mutation research 488 (2001) 195–209 forging the links...

101REFLECTIONS IN MUTATION RESEARCH: 1999 – 2019100 ELSEVIER

Mutation Research 488 (2001) 195–209

Forging the links between metabolism and carcinogenesis ☆F. Peter Guengerich *

Department of Biochemistry, Center of Molecular Toxicology, Vanderbilt University School of Medicine,638 Medical Research Building I, 23rd Avenue South at Pierce, Nashville, TN 37232-0146, USAAccepted 26 January 2001


Reflections in Mutation Research

Forging the links between metabolism and carcinogenesis�

F. Peter Guengerich∗

Department of Biochemistry, Center of Molecular Toxicology, Vanderbilt University School of Medicine,638 Medical Research Building I, 23rd Avenue South at Pierce, Nashville, TN 37232-0146, USA

Accepted 26 January 2001

Abstract

Metabolism plays important roles in chemical carcinogenesis, both good and bad. The process of carcinogen metabolismwas first recognized in the first half of the twentieth century and developed extensively in the latter half. The activationof chemicals to reactive electrophiles that become covalently bound to DNA and protein was demonstrated by Miller andMiller [Cancer 47 (1981) 2327]. Today many of the DNA adducts formed by chemical carcinogens are known, and extensiveinformation is available about pathways leading to the electrophilic intermediates. Some concepts about the stability andreactivity of electrophiles derived from carcinogens have changed over the years. Early work in the field demonstrated theability of chemicals to modulate the metabolism of carcinogens, a phenomenon now described as enzyme induction. Thecytochrome P450 enzymes play a prominent role in the metabolism of carcinogens, both in bioactivation and detoxication. Theconjugating enzymes can also play both beneficial and detrimental roles. As an example of a case in which several enzymesaffect the metabolism and carcinogenicity of a chemical, aflatoxin B1 (AFB1) research has revealed insight into the myriadof reaction chemistry that can occur even with a 1 s half-life for a reactive electrophile. Further areas of investigation involvethe consequences of enzyme variability in humans and include areas such as genomics, epidemiology, and chemoprevention.© 2001 Elsevier Science B.V. All rights reserved.

Keywords: Carcinogens; Metabolism; Cytochrome P450; Mutagenesis

1. Introduction

I start by admitting that I was somewhat surprisedwhen I was invited to prepare a Reflections article onthe title topic. The field was in full swing when I beganmy faculty position in 1975. 1 At the time I was 26

� This article is part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, readersshould contact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).

∗ Tel.: +1-615-322-2261; fax: +1-615-322-3141.E-mail address: [email protected](F.P. Guengerich).

1 My interest is not to dwell on my own career, but for therecord I should probably indicate how I came to this field. My

undergraduate career at the University of Illinois (Urbana) gave mea B.S. (1970) in Agricultural Science, a derivative of my agrarianyouth. During those undergraduate years, I became very interestedin biochemistry and subsequently received a Ph.D. (1973) fromVanderbilt, working under the direction of Prof. Harry P. Broquist,a biochemist interested in nutritional problems. My own thesisproject was in the area of alkaloid biosynthesis. Following a desireto learn more enzymology, I did postdoctoral work (1973–1975)under the direction of Prof. Minor J. Coon at the University ofMichigan where I first began my studies on cytochrome P450(P450) enzymes. (I add parenthetically that Profs. Broquist andCoon are among the finest people I have met in science and, alongwith my own father, have been very important influences in mycareer).In 1975, Leon Cunningham, Chairman of the Department ofBiochemistry at Vanderbilt, inquired as to my interest in applying

1383-5742/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S1 3 8 3 -5 7 42 (01 )00059 -X

196 F.P. Guengerich / Mutation Research 488 (2001) 195–209

years old, so what had happened before then I havelearned either from the literature or in verbal accountsfrom others. The person who should have written thisarticle is Prof. James Miller. 2 Over the years, he toldme some things about the early history of the field, andhe remained its senior scholar. The invitation, however,gives me the opportunity to dedicate this article tothe memory of the late James and Elizabeth Millerfor their many contributions to this field of research,including the interests they have stimulated. Readersare referred to a collection of reviews I found in myfiles that cover the early events much better than I canin the space available [1–8].

2. Early history of the field

The field of chemical carcinogenesis probably firstbegan with the epidemiological associations of tumorswith tobacco snuff and soots by Hill [9] and Pott[10], respectively, followed later by observations onurinary bladder tumors in workers handling aryl-amines in European “aniline dye” factories in the1890s [11]. Early work with experimental animalsinvolved polycyclic aromatic hydrocarbons (PAHs),coal tar by Yamagiwa and Ichikawa [12] and puri-fied dibenz[a,h]anthracene by Kennaway and Hieger[13]. Subsequently, benzo[a]pyrene was also isolatedby the latter group [14] and has served as an im-portant prototype for PAHs since. In 1938, Hueperdemonstrated that 2-naphthylamine, an arylaminerelated to the bladder tumors in chemical workersmentioned earlier [11], could also produce bladdertumors in dogs (but not rats) [11,15]. The 1932 reportof Lacassagne on the induction of tumors by estrone[16] is apparently the first recognized example ofan “endogenous” chemical that could cause cancer;

for a faculty position in the area of biochemical toxicology, inconnection with the Center in Toxicology then headed by BobNeal. With what I thought was a reasonable biochemical back-ground, I began this job with a limited knowledge of toxicologyand carcinogenesis. My long-term goal when I began in 1975 wasto characterize the roles of individual enzymes in the activation ofcarcinogens and also the chemistry involved in the modificationof DNA and proteins.

2 Prof. James A. Miller passed away on 24 December 2000, afterthe original manuscript was submitted.

the issue of hormonal carcinogenesis is still withus today.

Although chemicals could be demonstrated tocause cancer in animals and humans, the relationshipwith DNA damage and metabolism was unclear. Someof the first work on the metabolism of carcinogenswas done in 1938 by Wiley and in 1941 by Dobriner,Hoffman, and Rhoads, who demonstrated ring hydro-xylation of 2-naphthylamine [7]. In 1947, theMillers worked with the brightly colored azo dyeN,N-dimethylaminoazobenzene and could demon-strate the binding of some colored product to protein[17]. The amount of protein-bound dye in rat liverwas correlated with the carcinogenicity in a set ofrelated dyes [18,19]. Subsequent studies also revealedthe covalent binding of benzo[a]pyrene to proteins[19], followed by other compounds.

Today one may properly ask the question of whythese and other investigators were analyzing proteinsfor adducts instead of DNA. As Miller pointed out[8], at the time most scientists were of the opinionthat the genetic information resided in proteins, notDNA, notwithstanding the 1944 experiments of Averyet al. with pneumococcal transforming factor [20]. 3

The first attempts to find carcinogens bound to DNAwere negative but subsequent work with radiolabelsincreased the sensitivity [8,21]. (Today we can detectmany at extraordinarily low levels [22]; the issue isnot finding adducts but understanding if the levels arebiologically meaningful.)

Mustard gas is regarded as the first establishedchemical mutagen [23]. Treatment of DNA with mus-tard gas altered the UV spectrum [24], although theexact mode of binding was not known for a numberof years. The mutagenesis landscape changed appre-ciably in 1953 with the classic report on DNA structureby Watson and Crick [25]. Interestingly, their reportsuggested that mispairing of bases (in mutagenesis)might be the result of the existence of rare base

3 I admit that I cannot provide a firsthand account of the fieldin the early 1940s because I was not born yet. One of my men-tors, Jud Coon, also expressed a similar view of the dogma inbiochemistry in the early 1940s. According to him, DNA wasgenerally considered to be something akin to collagen, i.e. a struc-tural component of cells. Most scientists seemed to believe thatsomehow the proteins themselves were the carriers of the geneticinformation.


F.P. Guengerich / Mutation Research 488 (2001) 195–209 197

tautomers [25], a hypothesis that still exists andhas not been disproven [26,27]. Although some ev-idence had been obtained that “classic” chemicalcarcinogens (e.g. PAHs) could be mutagenic [28] thisview was not very generally accepted. Boyland andcoworkers [29,30] had suggested that naphthaleneoxidation might be explained by an initial epoxidation,providing a potential mechanism for generation ofelectrophiles.

In the 1950s, more evidence developed that therewas a positive correlation between carcinogenicityand the extent of DNA modification. In the 1960s, thisevidence became more developed with the char-acterization of the chemical structures of severalDNA adducts resulting from different carcinogens,including arylamines and N-nitrosamines [1].

A problem in much of the work was that therewas a rather weak correlation between mutagenicityand carcinogenicity. Many mutagens, when testedin animals, were carcinogens. However, only a frac-tion of carcinogens were mutagens. With a grow-ing appreciation of the nature of DNA–carcinogenadducts and of metabolism in general, the reasonfor poor correlation was recognized to be deficientmetabolism in test systems. A number of efforts weremade to improve the mutational systems. Maher et al.[31] demonstrated that bacterial DNA could be mod-ified with an electrophilic derivative of a carcinogen(esters of N-hydroxyarylamides) and used to detectmutations (in the tryptophan operon) in Bacillus sub-tilis. Some efforts were made to couple metabolicactivation systems with mutation screens, but BruceAmes’ Salmonella typhimurium his− strains, cou-pled with rat liver post-mitochondrial supernatant,became most popular [32]. The title of an Amesarticle [33] begins with “Carcinogens are muta-gens: . . . ”. The correlation between mutagenicityin these assays and carcinogenicity measured inrodent assays (and human epidemiological work) hasbeen compared [34–38]. The correlation (including“potency indices”) varies somewhat depending uponthe nature of the chemicals. Some chemicals operatepartially or largely through mechanisms involvingreceptor responses independent of metabolism (e.g.2,3,7,8-tetrachlorodibenzo-p-dioxin). Nevertheless,the Ames test is still a very useful initial screen forthe genotoxic potential of industrial chemicals andpharmaceutical agents.

3. Enzymes involved in carcinogen metabolism:early studies

Some of the first studies on the use of in vitro en-zyme systems were done by Mueller, working withthe Millers as a graduate student on the metabolismof N,N-dimethyl-4-aminoazobenzene with NADPH-fortified rat liver microsomes [39]. The significanceof this work was enhanced by the 1952 work ofRichardson et al. [40], in which administration of3-methylcholanthrene to rats decreased the tumorige-nicity of 3�-methyl-N,N-dimethyl-4-aminoazobezene.These studies, along with the demonstration that ran-cid food in the diet could alter the metabolism of car-cinogens [41], were explained by enzyme induction,which Remmer had also observed in patients usingbarbiturates [42]. Conney, then working as a studentwith the Millers, provided evidence for the role of enz-yme induction in altering carcinogen metabolism [43].

The field of enzyme induction developed slowly inthe 1960s, primarily due to the limitations of technol-ogy in the area of molecular biology. However, Nebertand Gelboin were able to demonstrate the inducibil-ity of aryl hydrocarbon metabolism by PAHs in cellculture in 1968 [44], and several years later Polandet al. demonstrated the existence of the “Ah” recep-tor in mediating this response [45]. In 1973, Shawand Kellerman [46,47] reported that humans couldbe grouped into phenotypes on the basis of the PAHinducibility of lymphocytes (in culture) and that theincidence of lung cancer in these smokers was corre-lated. Although the results were difficult to repeat fora number of years, subsequent mRNA analyses con-firmed the general hypothesis [48,49]. The Shaw andKellerman studies were probably the first major effortin molecular epidemiology in this area and stimulatedfurther interest, which has persisted to this day.

Although several researchers had been utiliz-ing NADPH-fortified liver microsomes to activateand detoxicate carcinogens by the 1960s, little wasknown about the enzymes involved. Work by Omuraand Sato on a pigment termed cytochrome P450(P450) [50] led to the demonstration that this wasthe terminal oxidase in an electron microsomaltransport chain [51]. Some of the mixed-functionoxidation reactions in microsomes could be att-ributed to the flavin-containing monooxygenase, e.g.N-oxygenations of some arylamines [52,53]. However,


the nature of the P450 component remained elu-sive until Lu and Coon separated the P450 andNADPH-P450 reductase components from rabbit livermicrosomes in 1968 [54]. Although carcinogens werenot used as substrates in this work, the solubilizationand reconstitution set the stage for a subsequent flurryof activity in the area.

4. P450 enzymology and carcinogen metabolism

By 1975 some work had been done on themetabolism of carcinogens by P450 enzymes [55,56].With my brief background in P450 enzymology andsome experience in the purification of these enzymes,I began my faculty career with the goal of defin-ing which of the rat liver enzymes were involved inspecific activation and detoxication reactions with car-cinogens and other toxicants. My group was able topurify some of these rat P450s and characterize theircatalytic activities [57–59]. Lu and Levin were alsosuccessful and contributed much in this area, particu-larly in the area of PAHs, working with Conney andJerina [60–62]. Ultimately our group purified nine dif-ferent rat liver P450s [59,63]. At the time, this workmight have seemed impressive although today werealize that there are 40–50 rat P450 genes (althoughnot all P450s contribute to carcinogen metabolism).

In the course of this early work in my own lab-oratory, we did some comparisons of the catalyticactivities of the rat liver P450s with rabbit liver P450sand found some seemingly large differences amongwhat we thought (at least at the time) were orthologs.These differences between the rat and rabbit enzymesraised particular concerns about their relevance to thehuman P450s. In the 1960s and 1970s, there werefew reports on the activities of human enzymes inthe metabolism of xenobiotic chemicals, and littlesolid evidence was available. In a 1980 review Wrightconcluded “human-liver microsomes generally do notpossess the high capacity for the oxidative metabolismof foreign compounds characteristic of rabbit- orrodent-liver microsomes” [2]. In retrospect, manyof these human preparations were probably of poorquality, including the ones we were able to procure.

Despite the difficulties in acquiring human liversamples for analysis of activities and purification ofenzymes, we felt strongly that this work would be

necessary if real insight would ever be made into hu-man chemical carcinogenesis. By a chance happening,we were able to establish an excellent relationship withthe local organ procurement agency, which providedexcellent tissues that they could not arrange to trans-plant. 4 We were able to isolate some human P450proteins to electrophoretic homogeneity, at first with-out characterizing their catalytic activities until afterpurification [64,65]. This was not a totally satisfyingapproach in that we did not have good way of relatingthese preparations to in vivo function.

In 1977, Smith reported that humans showedwhat seemed to be a genetic polymorphism in the4′-hydroxylation of the drug debrisoquine [66]. Irealized that we might use the approach of purifyingP450s from human liver on the basis of certain cata-lytic activities and then be able to better relate (invitro) results with the disposition of drugs in vivo,which might ultimately lead to in vivo predictionsabout carcinogen metabolism from in vitro work.This approach led to the purification of what are nowtermed P450s 1A1, 1A2, 2A6, 2B6, 2D6, 2E1, and3A4 [63,67–71].

With the purified P450s, we were able not only todo some direct characterization of catalytic activitiesand carcinogen activation but also identify chemicalinhibitors, e.g. quinidine [72], and raise antibodiesthat could be used for several purposes, including (i)estimations of levels of individual P450s in individualliver samples [73], (ii) immunoinhibition of catalyticactivities in microsomes as a means of estimating thecontribution of a particular P450 [74], and (iii) cloningof P450 cDNAs, which we contributed to, althoughthe bulk of the work in this area was done byGonzalez [75].

One approach that helped us a great deal wasthe introduction of the umu genotoxicity assay in ourlab in 1988 by Shimada [76–79]. This assay providesa screen for genotoxicity independent of mutagenesis.

4 Phil Wang, a postdoctoral in my group, was involved in all ofthe early human work and deserves a great deal of credit. Hiswife, a nurse, worked at another hospital in Nashville. One daythe organ transplant staff were in her ward (I forget why) and sheoverheard a conversation. She asked if they had any human liverher husband could have and they (Luke Skelley and the team) said“no problem”. This arrangement served us well for more than adecade, although it meant trips at odd hours to get material; I amvery glad to be making human P450s in bacteria now.



DNA damage by carcinogen metabolites induces theSOS response in bacteria [76]. One of the genes inthe SOS cascade, umuC, is linked with the reportergene lacZ in a chimeric plasmid, which produces�-galactosidase and can be monitored colorimetri-cally. Thus, one could obtain information about theactivation of procarcinogens with three major advan-tages to the Ames test: (i) the time involved in theassay was reduced to a few hours; (ii) the quantitationin the system was changed from counting colonies toa single spectrophotometric reading and (iii) the com-plication of microbial contamination in any humanliver samples was not an issue in the rapid assay.Within a few months a large number of chemicalcarcinogens could be processed in assays involvingpurified enzymes (the concentrations needed wereonly nanomolar) and various experimental designswith human liver microsomes [77,79–89]. We con-tinued to use the umu screening approach in variousways. With most classes of chemical carcinogens, ithas been possible to extend our studies to assays ofspecific carcinogen metabolites, e.g. [80,90–95].

Numerous systems have been developed for the het-erologous expression of P450s in various cell systems,both microbial and mammalian [96–98]. These sys-tems can be used to help establish roles of individual(human) P450s in various reactions. “Bicistronic” andother systems have been developed for the purposeof co-expressing NADPH-P450 reductase along withP450s [99,100]. Tetsuya Kamataki’s laboratory hasdeveloped an Ames test (S. typhimurium) with indivi-dual human P450s incorporated [101,102]. In col-laboration with Josephy et al. [103,104] we haveconstructed such “self-contained” human P450 sys-tems with a lac reporter [105]; recent work with Odahas yielded such systems in the umu assay platform[106,107].

5. Non-P450 enzymes involved in carcinogenmetabolism

The study of enzymatic conjugation of potentiallytoxic chemicals goes back to the 1800s [108–111].Subsequently, Boyland identified a glutathione (GSH)conjugate of naphthalene [112]. Most of the so-called“Phase II” enzymes (a term proposed by Williams[110] to distinguish the conjugating enzymes from

those involved in oxidation/reduction) participate insome detoxication reactions with carcinogens. The listincludes GSH transferase, UDP glucuronosyl trans-ferase, N-acetyltransferase, sulfotransferase, epoxidehydrolase, and methyl transferase [113].

What has been appreciated more in recent yearsis the variety of ways in which most of these con-jugating enzymes also activate pro-carcinogens.Sulfate and acetyl transferases were already knownto add good leaving groups by the 1960s (e.g. withN-hydroxy-N-acetylaminofluorene [114,115]). Whathas probably been less generally appreciated is thatsome of these reactions have been shown to enhanceboth tumor promotion and initiation [116].

My own interest in this latter area has been focusedon ethylene dihalides, compounds used widelyin industry that induce tumors in rodents [117].1,2-Dibromoethane is no longer used commercially;1,2-dichloroethane is the precursor of vinyl chloride.Our interest stemmed from our work on oxidation ofthese chemicals and was stimulated by the proposalof a GSH half-mustard product by Rannug [118].Our initial studies with 1,2-dichloroethane wereunsuccessful and later we took advantage of thehalide order, readily demonstrating the incorpora-tion of equimolar labels from GSH and ethylenefrom 1,2-dibromoethane into DNA, thus indicatinga (guanyl) ethylene–GSH adduct(s) [119]. Work onthe project has continued to the present and has beenreviewed recently [120]. As with many carcinogens,several different DNA adducts are formed and havebiological effects [121,122]. The question of howmuch mispairing can be attributed to the major N7-guanyl adduct is still open [121–123], in light of con-siderations of the mutagenicity of base tautomers [27].

Our experience with this project led us to study theactivation of dihalomethanes, a study motivated by in-creased interest in the rodent tumorigenicity of dichlo-romethane. The use of S. typhimurium expressing GSHtransferases has provided evidence of the roles of thetheta class enzymes in bioactivation [124–126].

Some other oxidation/reduction (Phase I enzymes)other than P450s can also generate reactive products.The list includes prostaglandin synthase, lipoxyge-nase, flavin-containing monooxygenase, and alcoholdehydrogenase. The reactive products are often thesame as or similar to those seen in the P450 reactions[127].


6. Aflatoxin B1 (AFB1) as a paradigm formetabolism

AFB1 provides an interesting case history in theissues involved in the metabolism and reactivity ofa procarcinogen. AFB1 was originally discoveredthough an incident with livestock, the poisoning ofturkeys in Britain by mold-contaminated peanut meal[128]. Subsequent work led to the characterization ofAFB1 and to its identification as a major contribu-tor to liver cancer in parts of the developing world[128,129].

AFB1 shows considerable variability in its toxi-city and carcinogenicity in different animal species[128]. A key development in AFB1 research was thecharacterization of the guanyl AFB1 DNA conjugate[130,131] (Fig. 1), which led to postulation of theepoxide as the electrophilic product involved in re-actions with macromolecules. More evidence for thishypothesis came from the identification of the GSHconjugate [139].

My colleague Harris and his group were able to syn-thesize the long-sought epoxide in 1988 [140]. Thissynthesis, in retrospect, allowed us to do a number ofkey experiments over the course of the next ten years.An important observation was that the exo isomer ofthe epoxide was at least 103 times more genotoxic

Fig. 1. Metabolism of AFB1. See [77,80,93,94,132–138] and included references.

than the endo form [134], a result akin to some ofthe important stereochemical differences seen withPAH diol epoxides [141]. This result, along with thedemonstrated difference in DNA reactivity, is best ra-tionalized in terms of the need for DNA intercalationand SN2 reaction of the epoxide [134,135]. Kineticstudies indicate that the half-life of the exo epoxide inneutral buffer is 1 second (k = 0.7 s−1) [135]. Despitethis short lifetime, this reactive epoxide is still stableenough to migrate into the cell nucleus and modifyDNA. The high reactivity of AFB1 exo-8,9-epoxidewith DNA is documented in a kinetic study; the reac-tivity can be understood in the context of both DNAaffinity and enhanced reactivity (k = 42 s−1) [138].

The enzymatic reactions related to AFB1 metabo-lism have been characterized. Shimada and I firstreported the significance of (human) P450 3A4 in theepoxidation reaction [77]. Subsequent work showedthat the reaction product was exclusively exo [93].Other P450s and some non-P450 oxygenases canalso oxidize AFB1 at lower rates [77,93]. P450 1A2may make some contribution to the oxidation butat least one-half of the product is the inactive endoisomer [77,93]. P450s also catalyze the oxidation ofAFB1 to other, inactive products, e.g. P450 3A4 tothe 3�-alcohol (aflatoxin Q1) and P450 1A2 to the9a-alcohol (aflatoxin M1) [93].



The availability of the synthetic epoxides hasallowed analysis of detoxication reactions. Rates ofGSH conjugation with AFB1 epoxides appear to bethe major reason for species differences in sensitivityto AFB1, and rates with human and rat GSH trans-ferases have been estimated [133,136]. The rapidnon-enzymatic hydrolysis of the epoxide renderscontribution of the enzyme epoxide hydrolase rathernegligible [137].

The dihydrodiol undergoes relatively slow, base-catalyzed ring opening to a dialdehyde, a reversible re-action [135]. The dialdehyde appears to bind to proteinlysine residues [142], although the direct involvementof the epoxide has not been examined. Both the exoand endo epoxides generate the dialdehyde (presum-ably enantiomeric at the hydroxyl �- to the aldehyde,although this might scramble due to enolization). Thedialdehyde is not genotoxic; if it were the dramatic dif-ferences in the binding of the epoxide steroisomers toDNA could not be rationalized [134]. However, the di-aldehyde may contribute to the acute toxicity of AFB1(or toxicity may be an issue in “promoting” the ini-tiating effects of DNA–AFB1 adducts). An aldehydereductase has been characterized that reduces AFB1dialdehyde to a dialcohol [143]. Although we felt wehad concluded our own studies on AFB1 with the workon the DNA kinetics [132], a conversation with JohnHayes led to a decision to re-evaluate the reductionof the dialdehyde reductase (AFAR). A concern aboutprevious work was that the dihydrodiol predominatesat neutral pH and cannot be a substrate for reduction.Reaction of base-stabilized dialdehyde with rat or hu-man AFAR (at neutral pH) indicated rapid reduction,first at the C8 position and then at C7.

Putting basic information about the enzymology ofAFB1 metabolism in the context of issues of humanrisk will require more time. Santella has found that thehigher levels of P450 3A4 in liver tissue near tumors(in individuals exposed to AFB1) are consistent witha role of this enzyme in hepatocellular cancer [144].Issues with P450 3A4 include the bifurcation between8,9-exo epoxidation and 3�-hydroxylation [93] in theliver and, perhaps even more importantly, in the smallintestine, a P450 3A4-rich area where AFB1 is firstencountered following oral administration and whereAFB1 activation to DNA adducts should not be tu-morigenic (due to rapid sloughing of cells). HumanGSH transferase M1 appears to be the most active

form in conjugating AFB1 exo-8,9-epoxide [133,136];in hepatocytes prepared from humans devoid of GSHtransferase M1 no GSH conjugates were detected[145]. Epidemiology studies have given conflictingresults to date on the role of the GSH transferaseM1 polymorphism in AFB1-related liver cancer[146,147].

7. What have we learned conceptually?

If we go back to the mid-1900s, the hypothesishas been proven that carcinogens are enzymaticallyactivated to reactive electrophiles that can bind toDNA and cause mutations and cancer. The same(pro)carcinogens are also detoxicated by the sameenzymes, sometimes the very same enzyme. Somecomments on the general scheme are in order. First,this paradigm does not apply to all chemical carcino-gens. Some chemicals (e.g. peroxisome proliferators)probably act by binding directly to receptors and in-creasing the transcription of specific (and as of yetpoorly understood) genes. In some cases the productsof metabolism are apparently not covalently boundbut bind tightly to certain proteins to evoke toxicresponses (e.g. trimethyl pentanol and α2u-globulin)[148]. High doses of chemicals can cause cell prolif-eration and contribute to tumorigenesis; the processmay often not be applicable at low doses. Having saidall this, a better appreciation of the role of mutationshas developed in recent years with (i) recognition ofthe significance of DNA repair [149], (ii) developmentof paradigms in which accumulation of mutations is apart of tumor progression [150], and (iii) demonstra-tion of chemical or physical agent-related patterns ofmutations in certain genes [151,152].

Often the pathways involved in carcinogen activa-tion are difficult to delineate due to the instabilityof products (AFB1, vide supra) or kinetic considera-tions. In the latter regard, work with urethane (ethylcarbamate) is an example. The Millers showed thatvinyl carbamate, the desaturation product, was moremutagenic and carcinogenic than ethyl carbamate(and required microsomal oxidation for activationto DNA-bound products) [153,154]. The resultssuggested the sequential oxidation scheme: ethylcarbamate → vinyl carbamate → vinyl carba-mate epoxide → DNA adducts (etheno derivatives).


However, vinyl carbamate was not detected in theusual incubations [153–155]. We utilized a sensitivegas chromatography and mass spectrometry methodand could demonstrate that the rate of epoxidationwas ∼103 times faster than that of desaturation,allowing observation of vinyl carbamate as an in-termediate [156]. This scenario probably has manycounterparts.

Not only P450s but also most Phase II enzymescan activate carcinogens. No single enzyme is alwaysgood or bad; the situation depends upon the carcinogenunder investigation. Thus, the use of chemopreventiveinterventions based on P450 inhibition (or induction)will be difficult to develop and will probably be mostuseful in settings in which a single carcinogen is ofconcern (e.g. oltipraz and AFB1).

One conclusion about the P450s is that a relativelysmall set of the 53 human P450s (http://www.drnelson.utmem.edu/CytochromeP450.html) seem to do mostof the carcinogen metabolism. P450s 1A1, 1A2, 1B1,2A6, 2E1, and 3A4 seem to be the major players inthe activation (and detoxication) of ∼90% of knowncarcinogens [157]. However, this view may changeas more is learned about the expression of P450s inextrahepatic target sites and what the most importantchemicals involved in the etiology of some cancersreally are.

Another point of interest is the catalytic differencesamong some seemingly orthologous P450s. In thisregard, Turesky and I considered the activation of thefood pyrolysate heterocyclic amines 2-amino-3,8-di-methylimidazo[4,5-f]quinoxaline (MeIQx) and 2-am-ino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)by rat and human P450 1A2 enzymes. Both enzymeshad essentially identical catalytic efficiencies in theoxidation of the model substrate 7-methoxyresorufinbut the human enzyme was >10 times more efficientin the oxidation of both MeIQx and PhIP to thehydroxylamines [158].

One of the concerns about rodent cancer (andother) bioassays is the gender and sometimes straindifferences that are seen with some chemicals. Someof these can now be rationalized in the contextof differences of the enzymes involved in carcino-gen metabolism. Rodents show considerable genderdifferences in the expression of some P450s, andthe patterns are controlled in development and canbe modulated by gonadectomy/hormone treatment

[159,160] and even growth hormone pulsatile patternsand certain signaling factors [161,162]. This researchprovides interesting biology but does not appear to behappening in humans, as judged by the lack of gendereffects on P450 expression and limited gender differ-ences in drug pharmacokinetics. Thus, what happenswith a carcinogen in one gender in a rodent modelmay not be indicative of what should be expected inhumans.

Another issue is the stability of reactive interme-diates, a subject treated in an earlier review [163]and discussed above under the heading of AFB1. Inthe mid-1970s, the concept came into vogue that thereactive intermediates generated from carcinogensand other toxicants were so reactive that they couldnever survive passage from the endoplasmic reticu-lum/cytosol to the nucleus (site of the DNA target).This view led to the concept that the most importantfraction of P450 must be in the nucleus, and a flurryof literature on nuclear P450 ensued. Two resultscame out of this. The first is that several experimentswith hepatocytes and carcinogens indicated that thebulk of the reactive metabolites were stable enough tomigrate out of the cells before reacting with (exoge-nous) DNA or GSH traps [164]. The second is thatthe P450s and other microsomal enzymes of interestwere determined to be located on the outer nuclearmembrane, which is contiguous with the endoplasmicreticulum [165] and not inside the nucleus. Over theyears our concepts have changed about the stabilityof reactive intermediates, as exemplified by the AFB1research (vide supra).

Another issue related to the stability of reactiveintermediates is the possibility of “coupling” of reac-tions, i.e. generation of reactive intermediates and im-mediate conjugation through “channeling” processesin which the electrophiles are “neutralized”. Althoughthis concept has a certain intellectual attraction andwe were able to demonstrate some enzyme–enzymeinteractions with purified proteins [166], 5 the ev-idence that such interactions occur in cells must

5 A problem with the interpretation of earlier experiments onthe coupling of P450 2B1 and epoxide hydrolase in the oxidationof naphthalene [166], later pointed out by van Bladeren et al.[167], is that of stereochemistry, in that a synthetic racemate ofthe epoxide was being compared with an enzymatically-generatedenantiomer.



still be regarded as hypothetical and, in the face ofdiscussion about rates of diffusion [138], may not beimportant.

Finally, the issue of the importance of the P450sand other enzymes of xenobiotic metabolism hasbeen addressed. Following the discovery of theseenzymes, there have been two general schools ofthought regarding their function [168]. One is thatthese enzymes have “physiological substrates” thathave been difficult to identify and that the reactionsseen with carcinogens and other xenobiotics are for-tuitous. Support for this view comes from (i) knowl-edge that some of these enzymes (e.g. P450s, GSHtransferases) do have critical roles in the metabolismof steroids, eicosanoids, etc. and (ii) the observationthat some of the “xenobiotic-metabolizing” membersof these enzyme families do use “physiological” sub-strates, e.g. P450 3A4 has high activity in testosterone6�-hydroxylation. The other view is that these en-zymes are not particularly critical to life and are partof a general, non-specific repertoire of systems fordealing with environmental stress (also included herewould be the efflux pump proteins such as MDR1).In this view, these enzyme systems are present forremoving natural products (e.g. alkaloids, terpenes)normally present in the diet. Drugs, industrial chem-icals, etc., are recognized because of general simi-larity, and reactive intermediates are unplanned partsof metabolic efforts to remove these. Inducibility ofthese enzymes is part of the general response. How-ever, there is probably little selection pressure onthese enzymes, so long as low doses of carcinogensare encountered, in that most chemically-inducedcancers would not be expected until after thereproductive age.

My personal view has always (or almost always)been the latter [168]. Support comes from the find-ing of humans with polymorphisms who are devoidof some of these enzymes, e.g. P450 2D6, GSHtransferase M1. These individuals may or may notbe at somewhat increased risks from some carcino-gens or toxicants, but the available evidence indicatesthat they generally fare well unless they encounterdrugs that have narrow therapeutic windows and themetabolism is critically dependent on a single en-zyme. Further evidence comes from the transgenic“knock-out” mouse work of Frank Gonzalez, in whichseveral P450s in families 1 and 2 and microsomal and

cytosolic epoxide hydrolase have been shown to benon-essential [169,170].

8. Concluding remarks

The study of the role of metabolism in chemicalcarcinogenesis has been an interesting one. The fieldhas played an important role in the development ofareas such as enzymology and gene transcription, aswell as matters directly related to cancer research.Further, this field has been an intellectually stimulat-ing one, in which I have been involved with syntheticand mechanistic chemistry, enzymology, mutagenesis,and cancer biology.

Much of the framework of what I have discussedwas already in place when I entered the field. Thefield has developed, and major questions still needto be addressed. We still need to learn more aboutsome of the basic properties of the enzymes underconsideration here. There are serious deficiencies inextrapolation of information from models and animalstudies to issues of human risk.

Finally, an open question is how much the wideinter-individual variations of the enzymes of interestcontribute to human cancer. Considerable precedentexists for cancer susceptibility in animal models[171,172], and we do know that the differences inthese enzymes can dramatically affect the in vivo dis-position of drugs. However, definitive epidemiologicalevidence for the importance of the enzyme varia-tions as factors in human cancer has been difficult toobtain. Problems include the difficulty in defining theetiology of human cancers (and which chemicals, ifany, contribute), the variation of enzyme levels overlong periods of time, and the general multifactorialnature of cancer. Ultimately the definition of theserelationships in humans is the rationale for the workdescribed here.

Acknowledgements

Research in this laboratory has been supported byUSPHS grants R35 CA44353 and P30 ES00267. Thisarticle is dedicated to the memory of the late Jamesand Elizabeth Miller for their many contributions tothis field of research, including the interest they have


stimulated. I thank Fred Kadlubar for his commentson the manuscript.

References

[1] J.A. Miller, Carcinogenesis by chemicals: an overview.G.H.A. Clowes Memorial Lecture, Cancer Res. 30 (1970)559–576.

[2] A.S. Wright, The role of metabolism in chemical muta-genesis and chemical carcinogenesis, Mutat. Res. 75 (1980)215–241.

[3] E.C. Miller, J.A. Miller, Searches for ultimate chemicalcarcinogens and their reactions with cellular macromole-cules, Cancer 47 (1981) 2327–2345.

[4] R.G. Harvey, Activated metabolites of carcinogenichydrocarbons, Acc. Chem. Res. 14 (1981) 218–226.

[5] A. Dipple, C.J. Michejda, E.K. Weisburger, Metabolism ofchemical carcinogens, Pharmacol. Ther. 27 (1985) 265–296.

[6] P.D. Lawley, Mutagens as carcinogens: development ofcurrent concepts, Mutat. Res. 213 (1989) 3–25.

[7] J.A. Miller, Brief history of chemical carcinogenesis, CancerLett. 83 (1994) 9–14.

[8] J.A. Miller, The metabolism of xenobiotics to reactiveelectrophiles in chemical carcinogenesis and mutagenesis:a collaboration with Elizabeth Cavert Miller and ourassociates, Drug Metab. Rev. 30 (1998) 645–674.

[9] J. Hill, in: R. Baldwin, J. Jackson, Cautions Against the Im-moderate Use of Snuff, Baldwin & Jackson, London, 1761.

[10] P. Pott, in: L. Hawes, Chirurgical Observations Relativeto the Cataract, the Polypus of the Nose, the Cancer ofthe Scrotum, the Different Kinds of Ruptures and theMortification of Toes and Feet, Hawes, London, 1775.

[11] L. Rehn, Ueber blasentumoren bei fuchsinarbeitern, Arch.Clin Chir. 50 (1895) 588–600.

[12] K. Yamagiwa, K. Ichikawa, Experimentelle studie überdie pathogenese der epithelialgeschwulste, Mitt. Med. Fak.Tokio 15 (1915) 295–344.

[13] E.L. Kennaway, I. Hieger, Carcinogenic substances and theirfluorescence spectra, Br. Med. J. 1 (1930) 1044–1046.

[14] J.W. Cook, C.L. Hewett, I. Hieger, The isolation of acancer-producing hydrocarbon from coal tar, Parts I, II, andIII, J. Chem. Soc. (1933) 394–405.

[15] W.C. Hueper, F.H. Wiley, H.D. Wolfe, Experimental pro-duction of bladder tumors in dogs by administration of beta-naphthylamine, J. Ind. Hygiene Toxicol. 20 (1938) 46–84.

[16] A. Lacassagne, Apparition de cancers de la mamelle chezla souris male, soumise a des injections de folliculine, C.R. Acad. Sci. 195 (1932) 630–632.

[17] E.C. Miller, J.A. Miller, The presence and significanceof bound amino azodyes in the livers of rats fedp-dimethylaminoazobenzene, Cancer Res. 7 (1947) 468–480.

[18] J.A. Miller, C.A. Baumann, The carcinogenicity of certainazo dyes related to p-dimethylaminoazobenzene, Cancer Res.5 (1945) 227–234.

[19] E.C. Miller, Studies on the formation of protein-boundderivatives of 3,4-benzpyrene in the epidermal fraction ofmouse skin, Cancer Res. 11 (1951) 100–108.

[20] O.T. Avery, C.M. Macleod, M. McCarty, Studies on thechemical nature of the substance inducing transformationof pnuemococcal types. I. Induction of transformationby a desoxyribonucleic acid fraction isolated frompneumonococcus type III, J. Exp. Med. 79 (1944) 137.

[21] F. Marroquin, E. Farber, The in vivo labeling of liverribonucleic acid by p-DAB-1′-C14, Proc. Am. Assoc. CancerRes. 4 (1963) 41.

[22] K.W. Turteltaub, J.S. Felton, B.L. Gledhill, J.S. Vogel,J.R. Southon, M.W. Caffee, R.C. Finkel, D.E. Nelson, I.D.Proctor, J.C. Davis, Accelerator mass spectrometry in bio-medical dosimetry: relationship between low-level exposureand covalent binding of heterocyclic amine carcinogens toDNA, Proc. Natl. Acad. Sci. U.S.A. 87 (1990) 5288–5292.

[23] C. Auerbach, J.M. Robson, Chemical production ofmutations, Nature 157 (1946) 302.

[24] A. Chanutin, E.C. Gjessing, Effect of nitrogen mustards onthe ultraviolet absorption spectrum of thymonucleate, uraciland purines, Cancer Res. 6 (1946) 599–601.

[25] J.D. Watson, F.H.C. Crick, Genetical implications of the stru-cture of deoxyribonucleic acid, Nature 171 (1953) 964–967.

[26] M.D. Topal, J.R. Fresco, Complementary base pairing andthe origin of substitution mutations, Nature 263 (1976)285–289.

[27] W. Suen, T.G. Spiro, L.C. Sowers, J.R. Fresco, Identificationby UV resonance Raman spectroscopy of an imino tautomerof 5-hydroxy-2′-deoxycytidine, a powerful base analogtransition mutagen with a much higher unfavored tautomerfrequency than that of the natural residue 2′-deoxycytidine,Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 4500–4505.

[28] J.G. Carr, Production of mutations in mice by 1:2:5:5-diben-zanthracene, Br. J. Cancer 1 (1947) 152–156.

[29] E. Boyland, Different types of chemical carcinogen and theirmode of action: a review, Cancer Res. 12 (1952) 77–84.

[30] J. Booth, E. Boyland, Metabolism of polycyclic compounds,5. Formation of 1,2-dihydroxy-1,2-dihyronaphthalenes,Biochem. J. 361 (1949) 365.

[31] V.M. Maher, E.C. Miller, J.A. Miller, W. Szybalski,Mutations and decreases in density of transforming DNAproduced by derivative of the carcinogens 2-acetylamino-flourene and N-methyl-4-aminoazobenzene, Mol. Pharmacol.4 (1968) 411–426.

[32] B.N. Ames, The detection of chemical mutagens with entericbacteria, in: A. Hollaender (Ed.), Chemical Mutagens,Principles and Methods for their Detection, Vol. 1, PlenumPress, New York, 1971, p. 267.

[33] B.N. Ames, W.E. Durston, E. Yamasaki, F.D. Lee,Carcinogens are mutagens: a simple test system combiningliver homogenates for activation and bacteria for detection,Proc. Natl. Acad. Sci. U.S.A. 70 (1973) 2281–2285.

[34] B.N. Ames, Identifying environmental chemicals causingmutations and cancer, Science 204 (1979) 587–593.

[35] B.N. Ames, R. Magaw, L.S. Gold, Ranking possiblecarcinogenic hazards, Science 236 (1987) 271–280.

[36] S.J. Rinkus, M.S. Legator, Chemical characterization of 465known or suspected carcinogens and their correlation withmutagenic activity in the Salmonella typhimurium system,Cancer Res. 39 (1979) 3289–3318.



[37] J. Ashby, R.W. Tennant, Chemical structure, Salmonellamutagenicity and extent of carcinogenicity as indicators ofgenotoxic carcinogenesis among 222 chemicals tested inrodents by the US NCI/NTP, Mutat. Res. 204 (1988) 17–115.

[38] J.K. Haseman, B.H. Margolin, M.D. Shelby, E. Zeiger, R.W.Tennant, Do short-term tests predict rodent carcinogenicity?Science 241 (1988) 1232–1233.

[39] G.C. Mueller, J.A. Miller, The metabolism of 4,4-dimethyl-aminoazobenzene by rat liver homogenates, J. Biol. Chem.176 (1948) 535–544.

[40] H.L. Richardson, A.R. Stier, E. Borsos-Nachtnebel, Livertumor inhibition and adrenal histologic responses in ratsto which 3′-methyl-4,4-dimethylaminoazobenzene and 20-methylcholanthrene were simultaneously administered,Cancer Res. 12 (1952) 356–361.

[41] R.R. Brown, J.A. Miller, E.C. Miller, The metabolism ofmethylated aminoazo dyes. IV. Dietary factors enhancingdemthylation in vitro, J. Biol. Chem. 209 (1954) 211–222.

[42] H. Remmer, The acceleration of evipan oxidation andthe demethylation of methylaminopyrine by barbiturates,Naunyn–Schmiedeberg’s Arch. Exp. Pathol. Pharmakol.237 (1957) 296–307.

[43] A.H. Conney, E.C. Miller, J.A. Miller, The metabolismof methylated aminoazo dyes. V. Evidence for inductionof enzyme synthesis in the rat by 3-methylcholanthrene,Cancer Res. 16 (1956) 450–459.

[44] D.W. Nebert, H.V. Gelboin, Substrate-inducible microsomalaryl hydroxylase in mammalian cell culture. II. Cellularresponses during enzyme induction, J. Biol. Chem. 243(1968) 6250–6261.

[45] A. Poland, E. Glover, A.S. Kende, Stereospecific, highaffinity binding of 2,3,7,8-tetrachlorodibenzo-p-dioxin byhepatic cytosol. Evidence that the binding species isreceptor for induction of aryl hydrocarbon hydroxylase, J.Biol. Chem. 251 (1976) 4936–4946.

[46] G. Kellerman, M. Luyten-Kellerman, C.R. Shaw, Geneticvariation of aryl hydrocarbon hydroxylase in humanlymphocytes, Am. J. Hum. Genet. 25 (1973) 327–331.

[47] G. Kellerman, C.R. Shaw, M. Luyten-Kellerman, Arylhydrocarbon hydroxylase inducibility and bronchogeniccarcinoma, N. Engl. J. Med. 298 (1973) 934–937.

[48] R.E. Kouri, C.E. McKinney, A.S. Levine, B.K. Edwards,E.S. Vesell, D.W. Nebert, T.L. McLemore, Variations inaryl hydrocarbon hydroxylase activities in mitogen-activatedhuman and nonhuman primate lymphocytes, Toxicol. Pathol.12 (1984) 44–48.

[49] T.L. McLemore, S. Adelberg, M.C. Liu, N.A. McMahon,S.J. Yu, W.C. Hubbard, M. Czerwinski, T.G. Wood, R.Storeng, R.A. Lubet, J.C. Eggleston, M.R. Boyd, R.N.Hines, Expression of CYP1A1 gene in patients withlung cancer: evidence for cigarette smoke-induced geneexpression in normal lung tissue and for altered generegulation in primary pulmonary carcinomas, J. Natl.Cancer Inst. 82 (1990) 1333–1339.

[50] T. Omura, R. Sato, A new cytochrome in liver microsomes,J. Biol. Chem. 237 (1962) 1375–1376.

[51] R.W. Estabrook, D.Y. Cooper, O. Rosenthal, The lightreversible carbon monoxide inhibition of the steroid

C21-hydroxylase system of the adrenal cortex, Biochem.Z. 338 (1963) 741–755.

[52] F.F. Kadlubar, J.A. Miller, E.C. Miller, MicrosomalN-oxidation of the hepatocarcinogen N-methyl-4-amino-azobenzene and the reactivity of N-hydroxy-N-methyl-4-aminoazobenzene, Cancer Res. 36 (1976) 1196–1206.

[53] C.B. Frederick, J.B. Mays, D.M. Ziegler, F.P. Guengerich,F.F. Kadlubar, Cytochrome P-450 and flavin-containingmonooxygenase-catalyzed formation of the carcinogenN-hydroxy-2-aminofluorene, and its covalent binding tonuclear DNA, Cancer Res. 42 (1982) 2671–2677.

[54] A.Y.H. Lu, M.J. Coon, Role of hemoprotein P-450 in fattyacid w-hydroxylation in a soluble enzyme system fromliver microsomes, J. Biol. Chem. 243 (1968) 1331–1332.

[55] G. Holder, H. Yagi, P. Dansette, D.M. Jerina, W. Levin,A.Y.H. Lu, A.H. Conney, Effects of inducers and epoxidehydrolase on the metabolism of benzo[a]pyrene by livermicrosomes and a reconstituted system: analysis by highpressure liquid chromatography, Proc. Natl. Acad. Sci.U.S.A. 71 (1974) 4356–4360.

[56] F.J. Wiebel, J.K. Selkirk, H.V. Gelboin, D.A. Haugen, T.A.van der Hoeven, M.J. Coon, Position-specific oxygenationof benzo[a]pyrene by different forms of purified cytochromeP-450 from rabbit liver, Proc. Natl. Acad. Sci. U.S.A. 72(1975) 3917–3920.

[57] F.P. Guengerich, Separation and purification of multipleforms of microsomal cytochrome P-450. Activities ofdifferent forms of cytochrome P-450 towards severalcompounds of environmental interest, J. Biol. Chem. 252(1977) 3970–3979.

[58] F.P. Guengerich, Separation and purification of multipleforms of microsomal cytochrome P-450. Partial characte-rization of three apparently homogeneous cytochromesP-450 prepared from livers of phenobarbital- and3-methylcholanthrene-treated rats, J. Biol. Chem. 253(1978) 7931–7939.

[59] F.P. Guengerich, G.A. Dannan, S.T. Wright, M.V. Martin,L.S. Kaminsky, Purification and characterization of livermicrosomal cytochromes P-450: electrophoretic, spectral,catalytic, and immunochemical properties and inducibility ofeight isozymes isolated from rats treated with phenobarbitalor �-naphthoflavone, Biochemistry 21 (1982) 6019–6030.

[60] D. Ryan, A.Y.H. Lu, J. Kawalek, S.B. West, W. Levin,Highly purified cytochrome P-448 and P-450 from rat livermicrosomes, Biochem. Biophys. Res. Commun. 64 (1975)1134–1141.

[61] D.E. Ryan, P.E. Thomas, D. Korzeniowski, W. Levin,Separation and characterization of highly purified forms ofliver microsomal cytochrome P-450 from rats treated withpolychlorinated biphenyls, phenobarbital, and 3-methyl-cholanthrene, J. Biol. Chem. 254 (1979) 1365–1374.

[62] A.W. Wood, W. Levin, A.Y.H. Lu, H. Yagi, O. Hernandez,D.M. Jerina, A.H. Conney, Metabolism of benzo[a]pyreneand benzo[a]pyrene derivatives to mutagenic products byhighly purified hepatic microsomal enzymes, J. Biol. Chem.251 (1976) 4882–4890.

[63] D. Larrey, L.M. Distlerath, G.A. Dannan, G.R. Wilkinson,F.P. Guengerich, Purification and characterization of the


rat liver microsomal cytochrome P-450 involved in the4-hydroxylation of debrisoquine, a prototype for geneticvariation in oxidative drug metabolism, Biochemistry 23(1984) 2787–2795.

[64] P. Wang, P.S. Mason, F.P. Guengerich, Purification ofhuman liver cytochrome P-450 and comparison to theenzyme isolated from rat liver, Arch. Biochem. Biophys.199 (1980) 206–219.

[65] P.P. Wang, P. Beaune, L.S. Kaminsky, G.A. Dannan, F.F.Kadlubar, D. Larrey, F.P. Guengerich, Purification and chara-cterization of six cytochrome P-450 isozymes from humanliver microsomes, Biochemistry 22 (1983) 5375–5383.

[66] A. Mahgoub, J.R. Idle, L.G. Dring, R. Lancaster, R.L.Smith, Polymorphic hydroxylation of debrisoquine in man,Lancet 2 (1977) 584–586.

[67] L.M. Distlerath, P.E.B. Reilly, M.V. Martin, G.G.Davis, G.R. Wilkinson, F.P. Guengerich, Purification andcharacterization of the human liver cytochromes P-450involved in debrisoquine 4-hydroxylation and phenacetinO-deethylation, two prototypes for genetic polymorphismin oxidative drug metabolism, J. Biol. Chem. 260 (1985)9057–9067.

[68] T. Shimada, K.S. Misono, F.P. Guengerich, Human livermicrosomal cytochrome P-450 mephenytoin 4-hydroxylase,a prototype of genetic polymorphism in oxidative drugmetabolism. Purification and characterization of two similarforms involved in the reaction, J. Biol. Chem. 261 (1986)909–921.

[69] F.P. Guengerich, M.V. Martin, P.H. Beaune, P. Kremers, T.Wolff, D.J. Waxman, Characterization of rat and human livermicrosomal cytochrome P-450 forms involved in nifedipineoxidation, a prototype for genetic polymorphism in oxidativedrug metabolism, J. Biol. Chem. 261 (1986) 5051–5060.

[70] C.-H. Yun, T. Shimada, F.P. Guengerich, Purification andcharacterization of human liver microsomal cytochromeP-450 2A6, Mol. Pharmacol. 40 (1991) 679–685.

[71] M. Mimura, T. Baba, Y. Yamazaki, S. Ohmori, Y. Inui, F.J.Gonzalez, F.P. Guengerich, T. Shimada, Characterization ofcytochrome P450 2B6 in human liver microsomes, DrugMetab. Dispos. 21 (1993) 1048–1056.

[72] F.P. Guengerich, D. Müller-Enoch, I.A. Blair, Oxidationof quinidine by human liver cytochrome P-450, Mol.Pharmacol. 30 (1986) 287–295.

[73] F.P. Guengerich, P. Wang, N.K. Davidson, Estimation ofisozymes of microsomal cytochrome P-450 in rats, rabbits,and humans using immunochemical staining coupled withsodium dodecyl sulfate–polyacrylamide gel electrophoresis,Biochemistry 21 (1982) 1698–1706.

[74] P.E. Thomas, A.Y.H. Lu, S.B. West, D. Ryan, G.T. Miwa,W. Levin, Accessibility of cytochrome P450 in microsomalmembranes: inhibition of metabolism by antibodies tocytochrome P450, Mol. Pharmacol. 13 (1977) 819–831.

[75] F.J. Gonzalez, Molecular genetics of the P-450 superfamily,Pharmacol. Ther. 45 (1990) 1–38.

[76] T. Shimada, Y. Oda, H. Yamazaki, M. Mimura, F.P.Guengerich, SOS function tests for studies of chemical car-cinogenesis in Salmonella typhimurium TA 1535/pSK1002,

NM2009, and NM3009, in: K.W. Adolph (Ed.), Methodsin Molecular Genetics, Gene and Chromosome Analysis,Vol. 5, Academic Press, Orlando, FL, 1994, pp. 342–355.

[77] T. Shimada, F.P. Guengerich, Evidence for cytochromeP-450NF, the nifedipine oxidase, being the principal enzymeinvolved in the bioactivation of aflatoxins in human liver,Proc. Natl. Acad. Sci. U.S.A. 86 (1989) 462–465.

[78] T. Shimada, M. Iwasaki, M.V. Martin, F.P. Guengerich,Human liver microsomal cytochrome P-450 enzymesinvolved in the bioactivation of procarcinogens detected byumu gene response in Salmonella typhimurium TA1535/pSK1002, Cancer Res. 49 (1989) 3218–3228.

[79] T. Shimada, M.V. Martin, D. Pruess-Schwartz, L.J. Marnett,F.P. Guengerich, Roles of individual human cytochromeP-450 enzymes in the bioactivation of benzo[a]pyrene,7,8-dihydroxy-7,8-dihydrobenzo[a]pyrene, and other dihy-drodiol derivatives of polycyclic aromatic hydrocarbons,Cancer Res. 49 (1989) 6304–6312.

[80] K.D. Raney, T. Shimada, D.-H. Kim, J.D. Groopman,T.M. Harris, F.P. Guengerich, Oxidation of aflatoxins andsterigmatocystin by human liver microsomes: significanceof aflatoxin Q1 as a detoxication product of aflatoxin B1,Chem. Res. Toxicol. 5 (1992) 202–210.

[81] T. Shimada, H. Yamazaki, H. Shimura, R. Tanaka, F.P.Guengerich, Metabolic deactivation of furylfuramide bycytochrome P450 in human and rat liver microsomes,Carcinogenesis 11 (1990) 103–110.

[82] T. Shimada, F.P. Guengerich, Inactivation of 1,3-, 1,6-, and1,8-dinitropyrene by human and rat microsomes, CancerRes. 50 (1990) 2036–2043.

[83] W.R. Brian, M.-A. Sari, M. Iwasaki, T. Shimada, L.S.Kaminsky, F.P. Guengerich, Catalytic activities of humanliver cytochrome P-450 IIIA4 expressed in Saccharomycescerevisiae, Biochemistry 29 (1990) 11280–11292.

[84] T. Shimada, F.P. Guengerich, Activation of amino-�-carbo-line, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine, anda copper phthalocyanine cellulose extract of cigarette smokecondensate by cytochrome P-450 enzymes in rat and humanliver microsomes, Cancer Res. 51 (1991) 5284–5291.

[85] H. Yamazaki, Y. Oda, Y. Funae, S. Imaoka, Y. Inui,F.P. Guengerich, T. Shimada, Participation of rat livercytochrome P450 2E1 in the activation of N-nitrosodimethyl-amine and N-diethylnitrosamine to products genotoxicin Salmonella typhimurium NM2009, Carcinogenesis 13(1992) 979–985.

[86] H. Yamazaki, Y. Inui, C.-H. Yun, M. Mimura, F.P.Guengerich, T. Shimada, Cytochrome P450 2E1 and 2A6enzymes as major catalysts for metabolic activation ofN-nitrosodialkylamines and tobacco-related nitrosaminesin human liver microsomes, Carcinogenesis 13 (1992)1789–1794.

[87] T. Shimada, E.M.J. Gillam, P. Sandhu, Z. Guo, R.H.Tukey, F.P. Guengerich, Activation of procarcinogens byhuman cytochrome P450 enzymes expressed in Escherichiacoli. Simplified bacterial systems for genotoxicity assays,Carcinogenesis 15 (1994) 2523–2529.

[88] H. Yamazaki, Y. Inui, S.A. Wrighton, F.P. Guengerich, T.Shimada, Procarcinogen activation by cytochrome P450



3A4 and 3A5 expressed in Escherichia coli and by humanliver microsomes, Carcinogenesis 16 (1995) 2167–2170.

[89] T. Shimada, C.L. Hayes, H. Yamazaki, S. Amin, S.S. Hecht,F.P. Guengerich, T.R. Sutter, Activation of chemicallydiverse procarcinogens by human cytochrome P450 1B1,Cancer Res. 56 (1996) 2979–2984.

[90] M.A. Butler, M. Iwasaki, F.P. Guengerich, F.F. Kadlubar,Human cytochrome P-450PA (P-450IA2), the phenacetinO-deethylase is primarily responsible for the hepatic 3-deme-thylation of caffeine and N-oxidation of carcinogenic aryl-amines, Proc. Natl. Acad. Sci. U.S.A. 86 (1989) 7696–7700.

[91] G.A. Csanády, F.P. Guengerich, J.A. Bond, Comparison ofthe biotransformation of 1,3-butadiene and its metabolite,butadiene monoepoxide, by hepatic and pulmonary tissuesfrom humans, rats, and mice, Carcinogenesis 13 (1992)1143–1153.

[92] E. Bauer, Z. Guo, Y.-F. Ueng, L.C. Bell, F.P. Guengerich,Oxidation of benzo[a]pyrene by recombinant humancytochrome P450 enzymes, Chem. Res. Toxicol. 8 (1995)136–142.

[93] Y.-F. Ueng, T. Shimada, H. Yamazaki, F.P. Guengerich,Oxidation of aflatoxin B1 by bacterial recombinant humancytochrome P450 enzymes, Chem. Res. Toxicol. 8 (1995)218–225.

[94] Y.-F. Ueng, T. Kuwabara, Y.-J. Chun, F.P. Guengerich,Cooperativity in oxidations catalyzed by cytochrome P4503A4, Biochemistry 36 (1997) 370–381.

[95] F.P. Guengerich, A.D.N. Vaz, G.M. Raner, S.J. Pernecky,M.J. Coon, Evidence for a role of a perferryl-oxygencomplex, FeO3+, in N-oxygenation of amines by cytochromeP450 enzymes, Mol. Pharmacol. 51 (1997) 147–151.

[96] B.W. Penman, L. Chen, H.V. Gelboin, F.J. Gonzalez,C.L. Crespi, Development of a human lymphoblastoidcell line constitutively expressing human CYP1A1 cDNA:substrate specificity with model substrates and promutagens,Carcinogenesis 15 (1994) 1931–1937.

[97] H.J. Barnes, C.M. Jenkins, M.R. Waterman, Baculovirusexpression of bovine cytochrome P450c17 in Sf9 cells andcomparison with expression in yesat, mammalian cells, andE. coli, Arch. Biochem. Biophys. 315 (1994) 489–494.

[98] F.P. Guengerich, E.M.J. Gillam, T. Shimada, New appli-cations of bacterial systems to problems in toxicology, Crit.Rev. Toxicol. 26 (1996) 551–583.

[99] J.A.R. Blake, M. Pritchard, S. Ding, G.C.M. Smith,B. Burchell, C.R. Wolf, T. Friedberg, Coexpression ofa human P450 (CYP3A4) and P450 reductase generatesa highly functional monooxygenase system in Escherichiacoli, FEBS Lett. 397 (1996) 210–214.

[100] A. Parikh, E.M.J. Gillam, F.P. Guengerich, Drug metabolismby Escherichia coli expressing human cytochromes P450,Nat. Biotechnol. 15 (1997) 784–788.

[101] A. Suzuki, H. Kushida, H. Iwata, M. Watanabe, T. Nohmi,K. Fujita, F.J. Gonzalez, T. Kamataki, Establishment ofa Salmonella tester strain highly sensitive to mutagenicheterocyclic amines, Cancer Res. 58 (1998) 1833–1838.

[102] P.D. Josephy, L.S. DeBruin, H.L. Lord, J. Oak, D.H. Evans,Z. Guo, M.-S. Dong, F.P. Guengerich, Bioactivation of

aromatic amines by recombinant human cytochrome P4501A2 expressed in bacteria: a substitute for mammaliantissue preparations in mutagenicity testing, Cancer Res. 55(1995) 799–802.

[103] P.D. Josephy, D.H. Evans, A. Parikh, F.P. Guengerich,Metabolic activation of aromatic amine mutagens by simul-taneous expression of human cytochrome P450 1A2,NADPH-cytochrome P450 reductase, and N-acetyltrans-ferase in Escherichia coli, Chem. Res. Toxicol. 11 (1998)70–74.

[104] P.D. Josephy, D.H. Evans, V. Williamson, T. Henry, F.P.Guengerich, Plasmid-mediated expression of the UmuCDmutagenesis proteins in Escherichia coli engineered forcytochrome P450-catalyzed activation of aromatic amines,Mutat. Res. 429 (1999) 199–208.

[105] C.G. Cupples, M. Cabrera, C. Cruz, J.H. Miller, A set of lacZmutations in Escherichia coli that allow rapid detection ofspecific frameshift mutations, Genetics 125 (1990) 275–280.

[106] P. Aryal, K. Yoshikawa, T. Terashita, F.P. Guengerich,T. Shimada, Y. Oda, Development of a new genotoxicitytest system with Salmonella typhimurium OY1001/1A2expressing human cytochrome P450 1A2 and NADPH-cyto-chrome P450 reductase, Mutat. Res. 442 (1999) 113–120.

[107] P. Aryal, T. Terashita, F.P. Guengerich, T. Shimada, Y. Oda,Use of genetically engineered Salmonella typhimuriumOY1002/1A2 strain co-expressing human cytochrome P4501A2 and NADPH-cytochrome P450 1A2 reductase andbacterial O-acetyltransferase in SOS/umu assay, Environ.Mol. Mutagen. 36 (2000) 121–126.

[108] W. Keller, Über verwandlung der benzoesäure inhippursäure, J. Liebigs Ann. Chem. 43 (1842) 108–111.

[109] M. Jaffe, Über die nach einführing von brombenzol undchlorbenzol im organismus entstehenden schwefelthaltigensäuren, Ber. Dtsch. Chem. Ges. 12 (1879) 1092–1098.

[110] R.T. Williams, Detoxication Mechanisms, Wiley, London,1947.

[111] F.P. Guengerich, Introduction and historical perspective, in:F.P. Guengerich (Ed.), Biotransformation, ComprehensiveToxicology, Vol. 3, Elsevier, Oxford, 1997, pp. 1–6.

[112] J. Booth, E. Boyland, P. Sims, An enzyme from rat livercatalysing conjugations with glutathione, Biochem. J. 79(1961) 516–524.

[113] F.P. Guengerich, Biotransformation, Vol. 3. ComprehensiveToxicology, Pergamon Press, Oxford, 1997.

[114] E.C. Miller, J.A. Miller, Mechanisms of chemical carci-nogenesis: nature of proximate carcinogens and interactionswith macromolecules, Pharmacol. Rev. 18 (1966) 805–838.

[115] J.H. Weisburger, P.H. Grantham, E. Vanhorn, N.H.Steigbigel, D.P. Rall, E.K. Weisburger, Activation and deto-xification of N-2-fluorenylacetamide in man, Cancer Res.24 (1964) 475–479.

[116] E.W. Boberg, A. Liem, E.C. Miller, J.A. Miller, Inhibition bypentachlorophenol of the initiating and promoting activites of1�-hydroxysafrole for the formation of enzyme-altered fociand tumors in rat liver, Carcinogenesis 8 (1987) 531–539.

[117] W.A. Olson, R.T. Habermann, E.K. Weisburger, J.M. Ward,J.H. Weisburger, Induction of stomach cancer in rats and


mice by halogenated aliphatic fumigants, J. Natl. CancerInst. 51 (1973) 1993–1995.

[118] U. Rannug, A. Sundvall, C. Ramel, The mutagenic effect of1,2-dichloroethane on Salmonella typhimurium. I. Activationthrough conjugation with glutathione in vitro, Chem. Biol.Interact. 20 (1978) 1–16.

[119] N. Ozawa, F.P. Guengerich, Evidence for formation ofan S-[2-(N7-guanyl)ethyl]glutathione adduct in glutathione-mediated binding of 1,2-dibromoethane to DNA, Proc.Natl. Acad. Sci. U.S.A. 80 (1983) 5266–5270.

[120] F.P. Guengerich, Mechanisms of mutagenicity of DNAadducts derived from alkyl and vinyl halides, Jpn. J.Toxicol. Environ. Health 43 (1997) 69–82.

[121] J.L. Cmarik, W.G. Humphreys, K.L. Bruner, R.S. Lloyd, C.Tibbetts, F.P. Guengerich, Mutation spectrum and sequencealkylation selectivity resulting from modification ofbacteriophage M13mp18 with S-(2-chloroethyl)glutathione.Evidence for a role of S-[2-(N7-guanyl)ethyl]glutathioneas a mutagenic lesion formed from ethylene dibromide, J.Biol. Chem. 267 (1992) 6672–6679.

[122] M.-S. Kim, F.P. Guengerich, Polymerase blockage andmisincorporation of dNTPs opposite the ethylene dibromide-derived DNA adducts S-[2-(N7-guanyl)ethyl]glutathione,S-[2-(N2-guanyl)ethyl]glutathione, and S-[2-(O6-guanyl)-ethyl]glutathione, Chem. Res. Toxicol. 11 (1998) 311–316.

[123] M. Persmark, F.P. Guengerich, Spectroscopic and thermo-dynamic characterization of the interaction of N7-guanylthioether derivatives of d(TGCTG*CAAG) with potentialcomplements, Biochemistry 33 (1994) 8662–8672.

[124] R. Thier, S.E. Pemble, J.B. Taylor, W.G. Humphreys,M. Persmark, B. Ketterer, F.P. Guengerich, Expression ofmammalian glutathione S-transferase 5–5 in Salmonellatyphimurium TA1535 leads to base-pair mutations uponexposure to dihalomethanes, Proc. Natl. Acad. Sci. U.S.A.90 (1993) 8576–8580.

[125] R. Thier, M. Müller, J.B. Taylor, S.E. Pemble, B. Ketterer,F.P. Guengerich, Enhancement of bacterial mutagenicity ofbifunctional alkylating agents by expression of mammalianglutathione S-transferase, Chem. Res. Toxicol. 8 (1995)465–472.

[126] Y. Oda, H. Yamazaki, R. Thier, B. Ketterer, F.P. Guengerich,T. Shimada, A new Salmonella typhimurium NM5004 strainexpressing rat glutathione S-transferase 5–5: use in detectionof genotoxicity of dihaloalkanes using an SOS/umu testsystem, Carcinogenesis 17 (1996) 297–302.

[127] T.A. Dix, L.J. Marnett, Metabolism of polycyclic aromatichydrocarbon derivatives to ultimate carcinogens during lipidperoxidation, Science 221 (1983) 77–79.

[128] W.F. Busby, G.N. Wogan, Aflatoxins, in: C.E. Searle (Ed.),Chemical Carcinogens, 2nd Edition, American ChemicalSociety, Washington, DC, 1984, pp. 945–1136.

[129] G.N. Wogan, Aflatoxins as risk factors for hepatocellularcarcinoma in humans, Cancer Res. Suppl. 52 (1992)2114s–2118s.

[130] J.M. Essigmann, R.G. Croy, A.M. Nadzan, W.F. BusbyJr., V.N. Reinhold, G. Büchi, G.N. Wogan, Structural iden-tification of the major DNA adduct formed by aflatoxin B1

in vitro, Proc. Natl. Acad. Sci. U.S.A. 74 (1977) 1870–1874.

[131] J.K. Lin, J.A. Miller, E.C. Miller, 2,3-Dihydro-2-(guan-7-yl)-3-hydroxy-aflatoxin B1 a major acid hydrolysisproduct of aflatoxin B1-DNA or -ribosomal RNA adductsformed in hepatic microsome-mediated reactions and in ratliver in vivo, Cancer Res. 37 (1977) 4430–4438.

[132] F.P. Guengerich, W.W. Johnson, Kinetics of hydrolysis andreaction of aflatoxin B1 exo-8,9-epoxide and relevance totoxicity and detoxication, Drug Metab. Rev. 31 (1999)141–158.

[133] K.D. Raney, D.J. Meyer, B. Ketterer, T.M. Harris, F.P.Guengerich, Glutathione conjugation of aflatoxin B1 exo andendo epoxides by rat and human glutathione S-transferases,Chem. Res. Toxicol. 5 (1992) 470–478.

[134] R. Iyer, B. Coles, K.D. Raney, R. Thier, F.P. Guengerich,T.M. Harris, DNA adduction by the potent carcinogenaflatoxin B1: mechanistic studies, J. Am. Chem. Soc. 116(1994) 1603–1609.

[135] W.W. Johnson, T.M. Harris, F.P. Guengerich, Kinetics andmechanism of hydrolysis of aflatoxin B1 exo-8,9-oxide andrearrangement of the dihydrodiol, J. Am. Chem. Soc. 118(1996) 8213–8220.

[136] W.W. Johnson, Y.-F. Ueng, B. Mannervik, M. Widersten,J.D. Hayes, P.J. Sherratt, B. Ketterer, F.P. Guengerich,Conjugation of highly reactive aflatoxin B1 8,9-exo-epoxidecatalyzed by rat and human glutathione transferases:estimation of kinetic parameters, Biochemistry 36 (1997)3056–3060.

[137] W.W. Johnson, Y.-F. Ueng, H. Yamazaki, T. Shimada, F.P.Guengerich, Role of microsomal epoxide hydrolase in thehydrolysis of aflatoxin B1 8,9-epoxide, Chem. Res. Toxicol.10 (1997) 672–676.

[138] W.W. Johnson, F.P. Guengerich, Reaction of aflatoxin B1

exo-8,9-epoxide with DNA: kinetic analysis of covalentbinding and DNA-induced hydrolysis, Proc. Natl. Acad.Sci. U.S.A. 94 (1997) 6121–6125.

[139] B. Coles, D.J. Meyer, B. Ketterer, C.A. Stanton, R.C.Garner, Studies on the detoxication of microsomallyactivated aflatoxin B1 by glutathione and glutathionetransferases in vitro, Carcinogenesis 6 (1985) 693–697.

[140] S.W. Baertschi, K.D. Raney, M.P. Stone, T.M. Harris,Preparation of the 8,9-epoxide of the mycotoxin aflatoxinB1: the ultimate carcinogenic species, J. Am. Chem. Soc.110 (1988) 7929–7931.

[141] A.W. Wood, R.L. Chang, W. Levin, D.R. Thakker, H.Yagi, J.M. Sayer, D.M. Jerina, A.H. Conney, Mutagenicityof the enantiomers of the diastereomeric bay-regionbenzo[c]phenanthrene 3,4-diol-1,2-epoxides in bacterial andmammalian cells, Cancer Res. 44 (1984) 2320–2324.

[142] G. Sabbioni, P.L. Skipper, G. Büchi, S.R. Tannenbaum,Isolation and characterization of the major serum albuminadduct formed by aflatoxin B1 in vivo in rats, Carcinogenesis8 (1987) 819–824.

[143] D.J. Judah, J.D. Hayes, J.-C. Yang, L.-Y. Lian, G.C.K.Roberts, P.B. Farmer, J.H. Lamb, G.E. Neal, A novelaldehyde reductase with activity towards a metabolite ofaflatoxin B1 is expressed in rat liver during carcinogenesisand following the administration of an anti-oxidant,Biochem. J. 292 (1993) 13–18.



[144] Y.J. Zhang, S.Y. Chen, W.Y. Tsai, H. Ahsan, R.M. Lunn,L.Y. Wang, C.J. Chen, R.M. Santella, Expression of cyto-chrome P450 1A1/2 and 3A4 in liver tissues of hepatocellu-lar carcinoma cases and controls from Taiwan and their rela-tionship to hepatitis B virus and aflatoxin B-1- and 4-amino-biphenyl–DNA adducts, BioMarkers 5 (2000) 295–306.

[145] S. Langouët, B. Coles, F. Morel, L. Becquemont, P.H.Beaune, F.P. Guengerich, B. Ketterer, A. Guillouzo,Inhibition of CYP1A2 and CYP3A4 by oltipraz results inreduction of aflatoxin B1 metabolism in human hepatocytesin primary culture, Cancer Res. 55 (1995) 5574–5579.

[146] K.A. McGlynn, E.A. Rosvold, E.D. Lustbader, Y. Hu, M.L.Clapper, T. Zhou, C.P. Wild, X.-L. Xia, A. Baffoe-Bonnie,D. Ofori-Adjei, G.-C. Chen, W.T. London, F.-M. Shen,K.H. Buetow, Susceptibility to hepatocellular carcinomais associated with genetic variation in the enzymaticdetoxification of aflatoxin B1, Proc. Natl. Acad. Sci. U.S.A.92 (1995) 2384–2387.

[147] C.J. Chen, M.W. Yu, Y.F. Liaw, L.W. Wang, S. Chaimprasert,F. Martin, A. Hirvonen, D.A. Bell, R.M. Santella, Chronichepatitis B carriers with null genotypes of glutathioneS-transferase M1 and T1 polymorphisms who are exposed toaflatoxin are at increased risk of hepatocellular carcinoma,Am. J. Hum. Genet. 59 (1996) 128–134.

[148] S.J. Borghoff, B.G. Short, J.A. Swenberg, Biochemicalmechanisms and pathobiology of α2u-globulin nephropathy,Annu. Rev. Pharmacol. Toxicol. 30 (1990) 349–367.

[149] E. Friedberg, G.C. Walker, W. Siede, DNA Repair andMutagenesis, Americn Society of Microbiology, Washington,DC, 1995.

[150] M. Hollstein, D. Sidransky, B. Vogelstein, C.C. Harris, p53mutations in human cancers, Science 253 (1991) 49–53.

[151] M. Barbacid, ras genes, Annu. Rev. Biochem. 56 (1987)779–827.

[152] M.S. Greenblatt, W.P. Bennett, M. Hollstein, C.C. Harris,Mutations in the p53 tumor suppressor gene: clues tocancer etiology and molecular pathogenesis, Cancer Res.54 (1994) 4855–4878.

[153] G.A. Dahl, J.A. Miller, E.C. Miller, Vinyl carbamate asa promutagen and a more carcinogenic analog of ethylcarbamate, Cancer Res. 38 (1978) 3793–3804.

[154] M.T. Leithauser, A. Liem, B.C. Stewart, E.C. Miller, J.A.Miller, 1,N6-Ethenoadenosine formation, mutagenicity andmurine tumor induction as indicators of the generation ofan electrophilic epoxide metabolite of the closely relatedcarcinogens ethyl carbamate (urethane) and vinyl carbamate,Carcinogenesis 11 (1990) 463–473.

[155] G.A. Dahl, E.C. Miller, J.A. Miller, Comparative carci-nogenicities and mutagenicities of vinyl carbamate, ethylcarbamate, and ethyl N-hydroxycarbamate, Cancer Res. 40(1980) 1194–1203.

[156] F.P. Guengerich, D.-H. Kim, Enzymatic oxidation of ethylcarbamate to vinyl carbamate and its role as an intermediatein the formation of 1,N6-ethenoadenosine, Chem. Res.Toxicol. 4 (1991) 413–421.

[157] F.P. Guengerich, T. Shimada, Oxidation of toxic and carci-nogenic chemicals by human cytochrome P-450 enzymes,Chem. Res. Toxicol. 4 (1991) 391–407.

[158] R.J. Turesky, A. Constable, J. Richoz, N. Vargu, J.Markovic, M.V. Martin, F.P. Guengerich, Differences inactivation of heterocyclic aromatic amines by rat and humanliver microsomes and by rat and human cytochromes P4501A2, Chem. Res. Toxicol. 11 (1998) 925–936.

[159] J.Å. Gustafsson, A. Mode, G. Norstedt, P. Skett, Sex steroidinduced changes in hepatic enzymes, Ann. Rev. Physiol.45 (1983) 51–60.

[160] D.J. Waxman, G.A. Dannan, F.P. Guengerich, Regulationof rat hepatic cytochrome P-450: age-dependent expression,hormonal imprinting, and xenobiotic inducibility of sex-specific isoenzymes, Biochemistry 24 (1985) 4409–4417.

[161] C. MacGeoch, E.T. Morgan, B. Cordell, J.Å. Gustafsson,Growth hormone regulates expression of rat liver cytochromeP-45015� at a pretranslational level, Biochem. Biophys.Res. Commun. 143 (1987) 782–788.

[162] H.W. Davey, S.-H. Park, D.R. Grattan, M.J. McLachlan,D.J. Waxman, STAT5b-deficient mice are growth hormonepulse-resistant, J. Biol. Chem. 274 (1999) 35331–35336.

[163] F.P. Guengerich, D.C. Liebler, Enzymatic activation ofchemicals to toxic metabolites, Crit. Rev. Toxicol. 14 (1985)259–307.

[164] A.L. Shen, W.E. Fahl, C.R. Jefcoate, Metabolism ofbenzo[a]pyrene by isolated hepatocytes and factors affectingcovalent binding of benzo[a]pyrene metabolites to DNAin hepatocyte and microsomal systems, Arch. Biochem.Biophys. 204 (1980) 511–523.

[165] Y. Sagara, T. Harano, T. Omura, Characterization of electrontransport enzymes in the envelope of rat liver nuclei, J.Biochem. 83 (1978) 807–812.

[166] L.S. Kaminsky, M.W. Kennedy, F.P. Guengerich, Differencesin the functional interaction of two purified cytochromeP-450 isozymes with epoxide hydrolase, J. Biol. Chem.256 (1981) 6359–6362.

[167] P.J. van Bladeren, J.M. Sayer, D.E. Ryan, P.E. Thomas, W.Levin, D.M. Jerina, Differential stereoselectivity of cyto-chromes P-450b and P-450c in the formation of naphthaleneand anthracene 1,2-oxides: the role of epoxide hydrolase indetermining the enantiomer composition of the 1,2-dihydro-diols formed, J. Biol. Chem. 260 (1985) 10226–10235.

[168] W.B. Jakoby, Detoxication enzymes, in: W.B. Jakoby (Ed.),Enzymatic Basis of Detoxication, Vol. 1, Academic Press,New York, 1980, pp. 1–6.

[169] S.S.T. Lee, J.T.M. Buters, T. Pineau, P. Fernandez-Salguero,F.J. Gonzalez, Role of CYP2E1 in the hepatotoxicity ofacetaminophen, J. Biol. Chem. 271 (1996) 12063–12067.

[170] J.M. Peters, H. Morishima, J.M. Ward, C.J. Coakley, S.Kimura, F.J. Gonzalez, Role of CYP1A2 in the toxicityof long-term phenacetin feeding in mice, Toxicol. Sci. 50(1999) 82–89.

[171] D.W. Nebert, The Ah locus: genetic differences in toxicity,cancer, mutation, and birth defects, Crit. Rev. Toxicol. 20(1989) 153–174.

[172] F.P. Guengerich, Roles of cytochrome P-450 enzymes inchemical carcinogenesis and cancer chemotherapy, CancerRes. 48 (1988) 2946–2954.


Reflections and meditations upon complexchromosomal exchanges ☆John R.K. Savage *

34 City Road, Tilehurst, Reading RG31 5HB, UKAccepted 22 August 2002



Reflections and meditations upon complexchromosomal exchanges�

John R.K. Savage∗

34 City Road, Tilehurst, Reading RG31 5HB, UK

Accepted 22 August 2002

Abstract

The application of FISH chromosome painting techniques, especially the recent mFISH (and its equivalents) where all 23human chromosome pairs can be distinguished, has demonstrated that many chromosome-type structural exchanges are muchmore complicated (involving more “break–rejoins” and arms) than has hitherto been assumed. It is clear that we have beengreatly under-estimating the damage produced in chromatin by such agents as ionising radiation. This article gives a briefhistorical summary of observations leading up to this conclusion, and after outlining some of the problems surrounding theformation of complex chromosomes exchanges, speculates about possible solutions currently being proposed.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Chromosome aberrations; Complex exchanges; FISH-painting; Ionising radiations; Interphase architecture

That structural changes can be produced in chromo-somes by ionising radiation has been known for a verylong time, and, down the years, there have been manybooks and reviews detailing ideas current at the timeof their composition, of which I list only a few [1–16].The recent introduction of fluorescence in situ hy-

bridisation (FISH) chromosome-painting as a stainingtechnique has undermined some of our establishedideas, and is having a profound impact on our under-standing of induced structural chromosomal aberra-tions. I would like, in this article, to reflect on someof the puzzles it raises.

� This article is part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, read-ers should contact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).

∗ Tel.: +44-118-9428159; fax: +44-118-9428159.E-mail address: [email protected](J.R.K. Savage).

I will begin by explaining the meaning of someterms which will be used.

1. Working definitions

For initial discussion purposes, we will accept thehypothesis that a chromosomal exchange arises frominteraction between pre-induced “lesions” within theDNA molecules of the participating chromosomes.Currently, these lesions are believed to be predomi-nantly double-strand breaks (dsb). However, irrespec-tive of their nature, or the actual mechanism of theexchange process, ultimately the damaged moleculesmust “touch”, and a break in continuity, followed byan illegitimate rejoining of the “break–ends”, must oc-cur. Conceptually, therefore, it is convenient to discussaberrations in terms of “breaks”, followed by “rejoin”interactions amongst the “break–ends”.

Although this may sound like the basics of thewidely accepted Breakage-and-Reunion Theory (B&R

1383-5742/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S1383 -5742 (02 )00066 -2


J.R.K. Savage / Mutation Research 512 (2002) 93–109 95

Pattern and Configuration is much more fre-quent, although resolution limitations and un-seen “cryptic” exchange events mean that muchof the time we still only have CAB(observed )

(Fig. 1D).4. A “COMPLEX” exchange is “one that involves

three, or more, breaks in two, or more, chromo-somes” [25].

5. Exchange “CYCLES”: The rejoining of DNAmolecular “ends” must always occur in a pair-wisemanner, maintaining strand polarity, but the endsinvolved may come from different chromosomes,or different parts of the same chromosome, andhave different orientations. Given the four endsof two breaks, three possible join situations exist:“Restitution” which restores the original continu-ity status, “Asymmetrical rejoining” (A) which, inthe simple two-break exchanges, always producesan acentric component, “Symmetrical rejoining”(S) which never produces an acentric component,unless the rejoin is incomplete [28].

Given more than four proximate break–endsand free interaction between them, the numberof possibilities increases, and the A and S termscan no longer be applied [29]. To cope with thisand allied problems, Sachs et al. proposed [30]that the various exchange possibilities be consid-ered as cycles of different orders. Thus, a simplefour-end pairwise interaction would be classed asan exchange cycle of order 2 (or c2 [27]), thatinvolving six break–ends as a cycle of order 3 (c3)and so on. Restitution then becomes a cycle of or-der 1 (c1). This provides an ingenious and logicalsolution for classifying the various rejoining pos-sibilities (Fig. 1D and E). Interestingly, a similarapproach was adopted by Fano [31], investigatingthe origins of complex exchanges in Drosophila.

Recently, to facilitate mathematical simulations,Sachs [32] has proposed an additional scheme,based on Graph Theory, for classifying the variousrejoining scenarios.

6. “CAB”: The initial arrangement of the interactivebreaks with respect to chromosomes and armscan be conveniently classified by the CAB system[25,26].

The acronym means the number of Chromo-somes, the number of Arms, the number of Breaks.Thus, CAB 3/4/4 means three chromosomes with

four breaks distributed between four arms. Incritical work, it is necessary to distinguish threeCABs:

(i) CAB(initial ), the disposition of the breaks im-mediately after induction and prior to anyrejoining or restitution interactions. For a par-ticular “real-life” situation, this must alwaysremain theoretical;

(ii) CAB(actual ), the disposition of the breakswhich have actually interacted to form theexchange Configuration;

(iii) CAB(observed ), the disposition of breaks de-duced from the observed Pattern (Fig. 1).Usually, CAB(observed ) < CAB(actual ) becauseof the limits of our techniques and resolution.

Given free interaction between available break–ends, each CAB(initial ) can be regarded as gen-erating a “family” of possible exchange Config-urations. If restitution (c1) is taken as a validinteraction, then C/A/3 families have 15 possibleoutcomes, C/A/4 have 105, C/A/5, 945, C/A/6,10395 and so on.

From CAB 2/2/3 to CAB 5/5/5 there are just26 possible CAB(initial ) families [33], and be-tween them, these can generate 15,060 exchangeConfigurations. Assuming homologues are neverinvolved in a Configuration, and painting singlechromosomes, 41,895 Patterns are possible only203 (0.5%) of which are distinctive, and theseform the basis of the S&S classification [34].

From these studies, we learn two importantlessons about single-paint Patterns. (a) DifferentConfigurations can generate the same Pattern. (b)The same Configuration can generate differentPatterns. Thus, it is almost always impossible toreconstruct the underlying exchange Configurationfrom a partial-paint Pattern [26,35], Fig. 1.

One, or more, of the variant single-paint Pat-terns may look like a c2, so reducing the estimateof complex frequency. These false signals aretermed “pseudosimple” [36–38], and in quantita-tive work where the whole genome is not painted,correction has to be made for them [34]. Thiscorrection becomes unnecessary when all chro-mosomes are distinctively painted. Although theterm “pseudosimple” is almost exclusively appliedto dicentrics in the literature, we must rememberthat it is also applicable to S events like reciprocal

94 J.R.K. Savage / Mutation Research 512 (2002) 93–109

[2–4,17]) the terms as defined here carry no implica-tion with regard to mechanism, and are therefore notsynonymous with those of that theory. In any case, adsb is not equivalent to the “primary break” of B&RTheory, and much confusion can arise by equating thetwo [18].

Furthermore, the molecular initiating events are tak-ing place in relatively relaxed interphase chromatin,and resultant aberrations are scored later in condensed,coiled chromosomes at metaphase, using light mi-croscopy with a resolution orders of magnitude re-moved from the initial processes. So, a one-to-onecorrespondence between the molecular and the visualmust not always be expected.

1. “SITE”: From the earliest days of radiation cyto-genetics, it has been recognised that the probabil-ity of rejoining interactions between break–endsis conditioned by the distance between them, and“proximity” has been the subject of much experi-ment and discussion (reviewed in [4,19–22]). Ini-tially, the maximum “rejoining distance” (h [4])was calculated to be around 1.0 �m, but currentideas have reduced this to 0.1–0.2 �m, or evenlower [23].

The observation that the between-cell distribu-tion of induced chromosome-type exchanges inTradescantia microspores was consistently under-dispersed, led Atwood [24] to conclude that therewas a limited number of discrete places within thenucleus where chromosome threads came closeenough for exchanges to form if lesions wereinduced there. They termed such places “Sites”.A very lively controversy followed (reviewed in[19,20]), and such severe limitation is now dis-counted.

This convenient term, though, is widely, butrather uncritically, used in the literature and forclarity, it is necessary to distinguish two kinds ofSites:(a) Pre-existing or potential Sites (p-Sites): Vol-

umes or regions where proximity conditionsfor exchange are satisfied for any lesions in-duced there. This corresponds to the originalWolff/Atwood definition.

Prior to lesion induction, it is impossible todefine either the shape or size of p-Sites. Thecrowded conditions within a nucleus make it

unlikely that they will be small discrete vol-umes as originally envisaged. Rather, if prox-imity is the primary determinant for exchange,one can expect suitable conditions to existover very large regions.

However, factors additional to a simple jux-taposition of threads are probably necessarybefore exchange can occur and this may re-duce the number of available regions.

(b) Lesion-defined Sites (l-Sites): Once one lesionis established, proximity conditions around itcan be investigated with respect to neighbour-ing lesions, and an l-Site can be defined asa spherical volume, centred on that lesion,within which one or more additional lesionsmust, or actually do, exist for exchange to oc-cur [20]. This is essentially how Lea [4] ex-pressed his concept of h.

Clearly, lesions may move to form suchl-Sites post-irradiation, if, for example, lo-calised enzymatic assemblies must be visitedto effect repair.

2. The “EXCHANGE CONFIGURATION” is the fi-nal structural entity with all its parts, resulting fromall the rejoining interactions that have taken placeamongst the participating chromosomes. One ormore of the rejoins may have failed (i.e. the ex-change is incomplete, in the structural sense, seelist-item 7); nevertheless, all the relevant fragmentsform part of the Configuration. The Configurationcorresponds to the CAB(actual) [25–27].

3. A “PATTERN” is the exchange as seen and scoredmicroscopically following a particular stainingtechnique. The technique used generates (or some-times fails to generate) a visible Pattern from theunderlying Configuration.

It is important to remember that, for chromo-some-type aberrations, it is very rare to see thecomplete configuration when conventional solidstaining is used, or when only a few chromosomesare painted. Consequently, most of our interpreta-tions and quantitative scoring are made from onlypart of the underlying Configuration (Fig. 1A–C).Under these conditions, only for the truly simpletwo-break exchanges will the Pattern and Config-uration correspond.

When all the chromosomes are painted dis-tinctively, as with mFISH, equivalence between



translocations (examples are given in Fig. 1Band C).

7. “INCOMPLETE”: An exchange Configuration orPattern is said to be incomplete if one, or more,of the potential rejoins has failed and both “open”break–ends are visible. Structural incompletenessis much more frequent in chromatid-type (30–50%)than in chromosome-type aberrations (<5%).

In cytogenetics it is a technical term, and shouldnot be confused with the more general usageof “failing to reach an expected, or recognised,standard”. Certain paint Patterns frequently haveone or more terminal segments missing [39–41]and these have sometimes been termed “incom-plete exchanges”, which of course they are in thegeneral sense, but the implication is that unre-joined segments have been lost. However, carefulwork using telomeric probes [42–45] has shownthat almost all termini with “missing” segmentsare telomerically capped, indicating that exchangehas involved segments too small to register posi-tive paint signals [46]. This confirms the long-heldview [12,28] that true incompleteness is rare (afew percent) for chromosome-type aberrations. Theterms “one-way exchange” or “terminal exchange”for these imbalanced forms are less ambiguous,and to be preferred.

The problem of missing terminal segments isstill present when all the chromosomes are painted,and sometimes, therefore, plausible assumptionshave to be made to “close” the exchange (accountfor, and assign all the participating segments)before an interpretation can be made [27].

8. “SEQUENTIAL EXCHANGE COMPLEXES”(SECs): As already known from chromatid-typeexchanges [28], complex exchanges involvingfour, or more, breaks are of two broad types; thosewhere all the break–ends participate in one grand“musical chairs” type rearrangement (i.e. exchange

�

PAINT—dic(BC), ace(cd), t–t(bDf), t(Ed), t(Fe)]. Panel E: If we assume that CAB(observed) = CAB(actual), i.e. that the Pattern we areseeing is actually the complete Configuration, then we can assign position and break–end numbering to the six breaks involved. Panel F:There is a limited number of break–end rejoins that will satisfy the colour-junctions observed in the Pattern, and these are listed. Becauseboth arms of one (blue) chromosome are involved, some alternative rejoins are permissible. Combining these rejoining possibilities, justfour possible solutions for the Configuration exist, and without additional information, it is impossible to decide between them. Tworequire a single cyclical exchange of order 6 (c6), a “musical chairs” interaction between all 12 break–ends. The other two are SequentialExchange Complexes (either “true” or “contingent”) comprising two exchange cycles of order 3 (2 c3). The complex Configuration istherefore “reducible”, and following Cornforth’s suggestion, we would record it as having an “obligate cycle structure” of 2 c3.

cycles of orders ≥4) and those compounded fromtwo, or more, apparently independent, lower-orderexchange cycles usually, but not necessarily, in-volving a common chromosome.

These latter are termed “Sequential ExchangeComplexes” (SECs), and are of two kinds. Thosewhere the component exchanges are truly inde-pendent events, (true SEC) and those arising bychance as one possible set of rejoinings from asingle cycle involving all the break–ends (contin-gent SEC [47]). It is impossible to decide betweenthese two origin possibilities in any particular case.Probabilities of contingent SEC frequencies canbe estimated from the CAB family expansions, butnumbers of true SEC will be super-imposed onthese frequencies according to unknown functions.

When just one, or a few chromosomes arepainted, only part of a Configuration is revealed,and under these circumstances, it is possible tointerpret many Patterns as SEC (discussion [48]).Unfortunately, bearing in mind the amount andcompaction of DNA in a metaphase chromosome,we can never rule out the possibility that all ex-changes ≥c3 are just collections of c2 too close forour light-microscope optics to resolve as separateevents.

Some of the complex Patterns now being re-vealed by mFISH, in particular those where bothhomologues, or both arms of a given chromosome,are participants, have several break–end rejoin so-lutions, and are often capable of being interpretedeither as single-cycle or as SEC. Such patternshave been termed by Cornforth “reducible” [27,49]and because a unique solution is not possible hehas proposed that for such cases one should choosethe solution which will minimise the cycle sizesrequired to achieve the observed pattern. He termsthis “the obligate cycle structure” [27] (examplegiven in Fig. 1D–F). Only when there is strictly one


Fig. 1. A sub-set of five chromosomes from a first post-irradiation metaphase following exposure in G0/G1, to illustrate some of the conceptsand terminology discussed in the text. Panel A: With solid staining, only Asymmetrical (A) chromosome-type aberrations are visible, andone dicentric and one acentric fragment can be seen. The logical inference is that these two elements are related, resulting from a simpletwo-break, four-end pairwise exchange (a cyclical exchange of order 2, or c2). There may also be (unseen) Symmetrical (S) exchangespresent. Panel B: Two chromosomes have been distinctively FISH-painted with a dual-paint technique. The remaining chromosomes havebeen counter-stained with a non-specific fluorochrome. The dicentric displays the expected Pattern for a c2 exchange. Another pair ofchromosomes have a Pattern indicating a reciprocal translocation (S exchange), logically interpretable as a c2. [Scoring: S&S—2A, 2B;PAINT—dic(BA), ace(ba), t(Ca), t(Ac)]. Panel C: A different pair of chromosomes has been painted with a dual-paint protocol. Thedicentric now has an anomalous Pattern, the terminal segment is not in the acentric fragment! This indicates that our previous inference ofa c2 exchange was wrong, a complex Configuration with a minimum of three breaks must be involved. The Pattern displayed in Panel Bis therefore a “pseudosimple”. There is another reciprocal translocation present. Careful inspection of the chromosomes involved indicatesthat the reciprocal translocation seen in Panel B also cannot be the result of a c2, and is therefore another example of a “pseudosimple”,this time involving an S event. [Scoring: S&S—2G, 2B; PAINT—dic(BA), t–t(bAc), t(Ca)]. Panel D: The chromosome sub-set has beenmFISH painted, giving each chromosome a distinctive colour. It is now seen that all five elements are related in a single complex Configu-ration, CAB(observed) 5/6/6. [Scoring: S&S—only applicable to single-paint patterns, the complex is compounded from 2A, 2G, 2B, 2B, 3O;



“limited number of Sites” controversy. It was pointedout that, with solid-staining, a given chromosomearm can never be observed taking part in more thanone dicentric, irrespective of the number of A or Sexchanges in which it has actually participated. Thus,there is a progressive loss of visible exchanges asradiation dose increases and this inevitably leads tounder-dispersed between-cell distribution of A ex-changes, accompanied by dose–response curve satu-ration (the “Distortion hypothesis”, reviewed in [19]).However, as these phenomena seemed to be confinedto cells with low numbers of chromosomes (<16), itwas assumed (mistakenly, as we now know) that theprobability of additional exchange participation wasnegligible for mammalian and human cells whichhave a lot more chromosomes.

4. Cracking bands

During the 1970s, methods were developed whichcaused chromosome arms of mammalian (but notplant) somatic cells seen at metaphase, to display lon-gitudinal arrays of alternating dark and pale bands.Thus, we had a situation similar to (but not identicalto) that in Drosophila salivary-gland cells.

Initially, this banding phenomenon was discoveredusing various fluorochromes and UVmicroscopy [53],but quickly a number of Giemsa-staining methods (G-,R-banding, etc. [54–56], reviewed in [57]) were foundthat could be used with light microscopy. The pat-terns proved to be consistent and specific for a givenarm region (except that the band frequencies werenot fixed, but chromosome contraction-dependent[58]). Thus, not only did it become possible to iden-tify individual chromosomes, but also to analyse (bypattern disruption [59]) many of the hitherto invis-ible chromosome-type S exchanges [60–63]. Thesebanding procedures revolutionised clinical and can-cer cytogenetics, and also opened up a new rangeof experimental possibilities in the field of inducedchromosomal aberrations [64].

From both clinical and cancer studies, evidence forthe formation and transmission of complex exchangesin human cells soon appeared [65,66], but surpris-ingly, given the numerous experimental studies ofradiation-induced changes using G- and R-banding,very few induced complex exchanges were reported,

certainly insufficient to give the impression that theywere at all common. Reasons for this discrepancy arenot clear. Possibly it arises from the fact that the qual-ity of banded metaphases varies considerably fromcell to cell, and that whilst in clinical studies, one hasthe same structural exchange in many cells, allowingmultiple assessments, in experimental studies of in-duced exchanges, every cell is unique and only onechance for analysis exists.

5. Widening the cracks

The advent of fluorescence in situ hybridisationchromosome painting (FISH-painting) changed thesituation completely. Initially, the number of fluo-rochromes available was limited, so only one, or afew pairs of homologues could be painted with thesame colour. Accurate recording of S exchanges nowbecame possible since they produced clear bi-colourjunctions [67–70].

However, numerous anomalous patterns began tobe recovered, indicating that many of the assumed c2exchanges were, in fact, more complicated [25,26,35,39,71].Addition of new fluorochromes allowing distinctive

painting of several homologues [36,38,72–75] con-firmed and extended the frequent occurrence of com-plex exchanges, and made it evident that we were stillnot getting the full story (Fig. 1B and C).

These strange, unexpected patterns required newdescriptive methods, and two complementary scor-ing schemes were developed, “S&S” and “PAINT”[25,26,35], reviewed in [34]. Examples of scoringcodes using the two systems are given in the captionsto Fig. 1B–D.Currently, using fluorochrome mixtures coupled

with sophisticated computer analysis and pseudo-colouring, it is possible to paint, distinctively, everypair of homologues in the human karyotype (mFISH[76]; “Sky” [77]; “COBRA” [78]). Now, for the firsttime, we can see the actual Configurations (or a majorpart of them) that generate the Patterns observed withearlier staining techniques (Fig. 1D). Results to dateconfirm the fact that complex formation is very muchmore common than we have realised [27,49,79–81].

So, it is now abundantly clear, that in our earliersolid-stain studies, we have been largely under-esti-


interacting break per chromosome is there a uniquerejoin sequence that satisfies the Pattern (discussion[47]).

2. Background

Until very recently, the vast bulk of our theoreti-cal, biophysical and mechanistic ideas have been de-rived from the study of aberrations (mostly those ofthe chromosome-type) seen in solid-stained mitoticmetaphase chromosomes.

For chromosome-type aberrations, only the four Aforms (those which give rise to acentric fragments[28]) are readily visible with this staining protocol,and the most plausible interpretation for these is thatthe dicentric, or centric-ring, is directly related toan accompanying acentric fragment (hence the scor-ing “rule” of linking one fragment to each exchange,Fig. 1A). Any excess fragments are either interstitial-(frequent) or terminal-deletions (rare).

According to the widely held B&R Theory[2–4,17,18], the terminal deletions are considered tobe the residue of originating primary breaks, i.e. un-used, or unusable precursors of restitution or rejoin-type interactions.

It is recognised, but seldom included in quantitativecalculations, that since A and S are probably alterna-tive modes of rejoining, there will be an approximatelyequal frequency of largely unseen S forms present inany metaphase sample (Fig. 1A–C). This can be de-duced from observation of chromatid-type changes,where, even with solid-staining, the majority of bothA and S forms can be seen and scored with equal fa-cility [28].As originally defined [4], the primary breaks were

regarded as complete severances of the chromosomebackbone, leaving “open” break–ends free to movearound within the nucleus, some of which would findunrelated ends and rejoin, illegitimately, with them toproduce exchanges. It is most important for us to re-member, when reading these early papers, that ideasof chromosome construction and interphase architec-ture were quite different from those held today (fordiscussions [13,20,50]).

Implicit in the B&R approach is the assumptionthat all the exchange aberrations arise from rejoin-ing between the four open-ends of pairs of breaks,

i.e. a c2 exchange. The visual impact of solid-stainedchromosome-type aberrations almost always leads tothis conclusion (Fig. 1A), so that classification, in-terpretation, and the very large body of mathematicalquantitative theory are all based upon this assumptionof an almost exclusive dominance of c2 exchanges.

3. Warning signs

Apart from the occasional tri- and tetra-centricexchanges seen at high levels of overall damage(always treated for scoring purposes as 2 c2, 3 c2,etc.) solid-stained chromosome-type aberrations aredevoid of the more complicated multi-chromosomalconcatenations frequently seen with chromatid-typeaberrations. This contrast between the two types withregard to the formation of complex interchanges (andintrachanges) is often noted in the literature, and inhind-sight, should have acted as a warning that wemight be missing something.

So also should the observations from the workwith Drosophila [1,31,51] which ante-dated, andin some respects set the stage for, the foundationalwork of the Sax and Lea groups. The solid-stainedpolytene salivary-gland chromosomes with which theDrosophila people worked display a specific array ofalternating dark and light bands, and the transmissibleS exchanges (unstable A forms having been elimi-nated during the embryonic cell divisions) can notonly be observed, but mapped with a very high degreeof accuracy. The resultant analyses quickly showedthat very complicated cyclical exchanges were fairlyfrequent, many involving interactions between severalbreak–ends from different chromosomes. Kaufmann[52] reported analysis of cells from a larva carryingcomplex rearrangements involving at least 32 breaks!These findings were considered to be such a contrastto the somatic cell results in the plant material, thatCatcheside, in his excellent 1948 review [6], seg-regates the Drosophila data from the plant studies,stating that “the mechanism of structural rearrange-ment following irradiation of sperm in Drosophilarequires separate consideration if only because cycli-cal exchanges occur between more than two breaks,leading to complex rearrangements. . . ”.

Hints that somatic chromosomes could be involvedin additional, unrecognised exchanges arose from the



related, compounded from the terminal segmentsderived from the observed dicentric or centric-ring(Fig. 1A). Possible relationships with other “normal”chromosomes were just never considered.

8. How and where?

The complex exchange frequencies and types thatare regularly recovered require either very large rejoin-ing distances for participating lesions, or much higherbreak densities within an l-Site than current biophys-ical data will allow [20]. We are faced then, with twobig problems:

“How can so many chromosomes/lesions be found(or become) so proximate that interactions of suchmagnitude can occur?” and “Where, in the nucleus,does this happen?”

Within the interphase nucleus, the bulk of the chro-mosome arm material is confined to discrete locationstermed domains, or territories. There does not seem tobe a massive intermingling of all chromatin as envis-aged by earlier workers (reviewed in [20,50,94,95]).Moreover, most of the DNA is condensed and splintedwith histone proteins, giving fibres of various dimen-sions, and with some degrees of super-coiling, andthis will restrict major movement of many lesions,and serve to keep them apart, so favouring restitution[96–98].

Viewing the nucleus as a whole, much of the chro-matin is therefore intra-domain, and thus many of therandom induced lesions will be located in positionswhich preclude inter-chromosomal exchange, i.e. onlya small volume of the nucleus is at risk for inter-change of any sort. Presumably, only intrachanges areformed from within-domain lesion interactions, and asmall proportion of these may, in turn, be incorporatedinto more complicated interchanges, most likely asnon-visible, “cryptic” events, given our current tech-niques.

We have, therefore, to look for meeting placesoutside the domain volumes, and to regions of less-condensed chromatin. Not all the chromatin is con-fined within the domain. Some is extruded and, oftenfor functional purposes, anchored to regions of theintra-nuclear matrix, or, running like “cables” in alldirections through the matrix continuum betweenthe domains, reaches the nuclear envelope becom-

ing associated with the lamina and the nuclear pores[50,96,99,100]. It has been suggested that the poresmay act like “press-studs” to which the chromatinrosettes needed for current cell requirements attach[101]. There may also be specific inter-chromosomelinks.

These observations provide us with three possiblelocations where chromatin can form p- or l-Sites forinter-chromosome exchanges involving several chro-mosomes: domain surface interfaces; the inter-domainspatial continuum; the vicinity of the lamina and poresof the nuclear envelope [20,50]. Before we considerthese locations in more detail, it will be helpful to sum-marise the sort of problems we face using a practicalexample.

9. Illustrating the problems

A good illustration can be obtained from 238Pu�-particle irradiation, which is very efficient for pro-ducing complexes [79,80,88]. The particles pass rightthrough the cells leaving linear tracks of ionisations(LET 121 keV/�m) consisting of a dense core sur-rounded by a “cloud” of � tracks. A reasonable esti-mate for the region of influence of a track would be astraight rod 10 nm in diameter.

Consider a spherical nucleus 6000 nm in diameter(d) and volume 1.13E+11 nm3. For random chordspassing through a sphere, the mean chord length is2d/3. A 10 nm rod of this length (4000 nm) has avolume ≈3.14E+5 nm3. Thus, for one average lengthtrack passing through this nucleus, only 2.78E−6(∼0.0003%) of the volume sees any radiation at all!Even five tracks of this length, which is the modalnumber per cell at about 37% survival [88], wouldonly increase the irradiated volume to 0.0014%, andthe probability of these tracks intersecting is vanish-ingly small.

For a human cell nucleus, it is estimated that the6000 nm diametrical track will transit 4–5 chromo-some domains, 8–10 arm domains [80] and depositabout 6–9 double-strand breaks scattered randomlyalong the length of the track, though it is suggestedthat there may be some localised clustering [102,103].The number of domains crossed will set a limit on thesize of the complex, and it is interesting to note thatapproximately five chromosomes is the average size


mating the level of induced chromosome damage.Consequently the shapes of our dose–response curves,upon which so much theory depends, are warped, andthis must inevitably lead us to question and re-assessthe validity of many previous qualitative and quanti-tative studies [30,82–84].

6. Complex universality

All radiations tested so far produce complex ex-changes [85,86], even carbon-K ultra-soft X-rays(USX) where the ionising tracks produced within thenucleus barely traverse a DNA molecule [87]. Someradiation qualities, like �-particles, are very efficient[79,80,88].

The recovered frequency is dose dependent[37,41,49,75,89,90] and in some cases, sample-timedependent. It will also be obvious from earlier para-graphs, that it is staining-protocol dependent.

For 4.0 Gy 137Cs acute �-rays given to unstimu-lated human lymphocytes, Loucas and Cornforth [49],using mFISH, found ∼26% exchanges were complexinvolving from 3 to 11 breaks for their formation. With4.0 Gy 250 kV X-rays given to contact-inhibited, un-transformed, human fibroblasts, Simpson and Savage[37] estimated (correcting for pseudosimples) fromsingle-paint results with five different chromosomes∼35% exchanges were complex, with five breaks asthe modal requirement. A 1.5 Gy average absorbeddose of 1.5 keV Al-K USX to contact-inhibited hu-man fibroblasts gave an estimated ∼26% complexexchanges involving chromosomes 1 and 2 [89]. A1.31 Gy mean absorbed dose of carbon-K USX tocontact-inhibited human fibroblasts produced ∼46%complexes involving chromosome 1 [87], thoughsomewhat lower frequencies have been recorded inmore recent experiments (Hill, personal communica-tion). A 0.41 Gy of 238Pu �-particles to fibroblastsproduced ∼39% complexes based on painting chro-mosomes 1 and 4 [88].

In so far as the shapes of dose–response curves canbe trusted [91], current studies suggest that the majorpart of the upward curvature of chromosome-type ex-changes characteristic of low-LET radiations resultsfrom the complex component. Simple c2 types appearto have a close to linear response [41,49], contra [90].However, there is a possibility that part of the linearity

may be an artefact of curve distortion, since scores of“simple” and “complex” exchanges are not indepen-dent [92].

As dose, and probably LET, rises not only the num-ber, but also the “complexity” (number of breaks, andnumber of chromosomes taking part) of the complexesincreases. It is not unusual to find five or six chro-mosomes involved. This means a greater number ofvariant Patterns from one Configuration when a singlechromosome is painted.

There is also an increased probability that both ho-mologues, or both arms, of a particular chromosomewill be involved, and this can lead to hidden exchangeevents (cryptic events). As mentioned earlier, one con-sequence when mFISH is employed, is that severalpossible rejoin sequences exist for the same Pattern,some of which will be interpretable as SEC, renderingthe Pattern “reducible” (Fig. 1D–F).

7. Why did we miss them?

Since, then, chromosome-type complex exchangesare so common within the range of radiation qualitiesand doses ordinarily used for aberration studies, whydid we miss them for so long?

Firstly, the only type of chromosome-type complexdetectable with solid staining is the multi-centric.Careful analysis of the configurations derived fromthe 26 CAB families 2/2/3 → 5/5/5 shows that nopattern greater than a tri-centric will appear, and thatthese form only 2.3% of the possible paint-detectedpatterns [20]. The vast majority of complexes aretherefore invisible when solid-staining is used.

Secondly, not only are S exchanges invisible withsolid-staining, but many A events too, because a givenchromosome arm can never be seen to take part inmore than one dicentric, even though it may have beeninvolved in several exchange events. Similarly, com-plex centric rings will rarely be formed because anyadditional exchange event (A or S) within the potentialring loop converts the ring to a dicentric (ring diminu-tion [93]) beginning the visual limitation sequence justdescribed.

Thirdly, solid-stain can only reveal that part of anyexchange which simulates an A event. In the absenceof evidence to the contrary, it was always assumedthat the accompanying acentric fragment was directly



In general, the participation of chromosome armsin exchanges appears to be approximately propor-tional to Relative Corrected Lengths (RCLs) of thearm [105,106]. Such proportionality to the product ofarm (domain) sizes is consistent with a surface-at-riskhypothesis. However, RCLs are computed on the as-sumption that all exchanges are c2, (as also is theLucas correction factor for obtaining full genome ex-change frequencies from a partial karyotype painting[70]) and several reports have suggested that somechromosomes (in some experiments) can depart fromexpectation [74,107–109] although considerable con-tradictions exist. Unfortunately, it is difficult to seehow one can calculate RCLs for multi-arm partici-pation to investigate the actual significance of suchproportionality departures.

11. Inter-domain spatial-continuump- and l-Sites

The foregoing discussion pre-supposes unrestrictedsurface–surface contact between domains, but thisis probably an unrealistic assumption given the ex-tent of chromatin extrusion that is occurring into theinter-domain continuum.

Selective localised chromatin decondensation takesplace regularly for many purposes such as transcrip-tion, replication, and perhaps repair [110], and it hasfrequently been suggested that such regions are vulner-able to structural damage. These metabolic processesoccur in association with highly organised assembliesof protein molecules, believed to be located predomi-nantly in the intra-nuclear matrix.

If we are going to make repair or “mis-repair” amechanism for structural exchange, then, in line withthe well investigated DNA repair systems, we are go-ing to require assemblies of proteins (repairosomes orgarages [20]) either pre-existent, or induced by emer-gency signals, to provide the necessary clamping andorientation and sequential enzyme processes. One thenhas to assume that damaged regions need to visit such“repairosomes” (unless the damage itself acts as a sig-nal commanding the construction of one on the spot).If these structures are a non-specific facility, damagedregions from several chromosomes may well meet upthere, forming an l-Site, and under pressure of anemergency situation, or of excessive damage, mistakes

leading to complex illegitimate rejoinings could oc-cur. Note, in passing, that “break–ends” have to movein pairs to preserve the universal occurrence of cycli-cal exchange and very low structural incompleteness,from which we infer that any distances travelled can-not be very large.

A few points need to be made about extrusion. Ifthe bulk of p-Site regions lie in matrix between thedomains, and the majority of interchanges occur here,then, in order to maintain the observed arm-size pro-portionality of participation in exchange, every do-main must extrude chromatin in an amount that is afunction of arm size.

Every extrusion must go out as a loop since thechromosome has a uninemic construction, and severalmicrons of loop may be required to reach the nuclearenvelope [50], unless extensive membrane invagina-tions are a regular feature of all nuclei. This meansthat many megabases of DNA lie external to the do-mains and the inter-domain spatial continuum mustbe crowded with chromatin “cables” criss-crossingin all directions. The existing p-Sites will not onlyinvolve contacts within groups of individual cables,but also take in passing contacts with domain-surfacechromatin. Thus, most extruded loops will have thepotential to interact with several chromosomes in theirtravels. Cable–cable and surface–cable interactionsmay prove to be much more important in complexformation than surface–surface ones, since it is obvi-ous that all the participating lesions do not have to belocalised in one l-Site.

Proximity conditions within individual extrudedloop cables must mean that some intrachange p-Siteslie outside the domain volume [21].

It is highly unlikely that extrusion will involve a“naked” DNA molecule, for that would be very dan-gerous. Lower order chromatin fibres or rosettes [111],or even micelles [112], providing matrix anchorageare more probable, which considerably increasesthe amount of externalised chromatin. Extensive an-chorage will, of course, limit chromatin movement,reducing the likelihood of forming additional l-Sites.However, the matrix is thought to have considerablefluidity, so this may not be a problem.

As discussed elsewhere [20,21], there is also apossibility that chromatin from several different chro-mosomes needs to meet in “functional associations”for normal metabolic or genetic purposes, and this


for �-irradiation, though a few much larger complexesare found [80]. In contrast, low-LET radiations appearto have a much wider spread of complex sizes [49].

If we assume a rejoining distance of 200 nm (basedon “nearest neighbour” considerations [20]), we cansegment a 6000 nm track into 30 discrete p-Sites. Ap-plying occupancy theory for 9 dsb randomly allocatedto these 30 p-Sites, one can show that only ∼8% ofsuch tracks will have any l-Sites with 3 or more dsb,and only ∼0.4% with 4 or more.

Bearing in mind that some of these l-Sites will beintra-domain, complexes derived from the interactionof ≥3 localised breaks ought to be rare. Even for nucleireceiving several tracks of varying lengths, individualbreak aggregates are unlikely to be enlarged becauseof the rarity of track intersections.

But, whilst complex derivation from multi-breakaggregates may be a valid model for low-LET ra-diations where dsb are scattered throughout the nu-cleus, this approach may be misleading for �-particleswhich have strictly linear break distributions alongtheir tracks ([80] and see further).

So, we have the situation where a very small volumeof chromatin is irradiated, a very limited supply of dsbdeposited, and very low probabilities of ≥3 dsb ag-gregates, and yet sufficient p-Sites are encountered, orl-Sites formed, during particle transit to make complexchromosome-type exchanges a regular and significantfeature of �-radiation exposure. And this paucity ofdsb numbers and aggregates, combined with an abun-dance of complex exchanges, seems to be the normfor all qualities of radiation.

10. Domain surface p- and l-sites

If all domains were of equal volume, the 3D spa-tial form of the inner elements would approximateto rhombic dodecahedra (12 faces), or truncated oc-tahedra (14 faces), the principal regular space-fillingsolids [104]. Peripheral elements would be less regularand have fewer faces. This means that most domainshave potential surface–surface contact with severalother chromosomes. Scattered dsb pairs, or clusters,embracing adjacent surface locations, such as wouldresult from low-LET radiations, could, provided therejoining distances are fairly large, produce concate-nations of chromosome arms to form a complex ex-

change. Almost all such complexes formed in this waywould be true SECs, and will have very limited re-joining combinations compared with those obtainablefrom free multiple break–end interactions.

This is readily seen for the �-particle situation[80]. A single track will traverse a linear array of nchromosome arm domains and could produce (n − 1)two-break l-Sites where one break is at the exit sur-face and the other at the entrance surface of adjacentdomains in the array. Of course, the number of suchl-Sites, and therefore the potential size of the result-ing complex, will be limited by the track length andnumber of dsb deposited.

Using the calculations of the previous section, ninedsb in a 6000 nm track would mean that a maximumof five arms could be involved in one complex, the re-sult of four linking c2 exchanges. However, the prob-ability of this complex size should be low, for only∼0.09% of the 6000 nm tracks have four l-Sites withtwo or more breaks, only ∼3% have three such l-Sitesand ∼23% have two. These percentages will be muchsmaller for shorter tracks. We could improve the situ-ation somewhat by increasing the rejoining distance,but it would have to become very large to accommo-date the frequencies of complexes recovered, and suchdistances are prohibited on biophysical grounds.

However, a really important factor which exacer-bates the problem further is the disposition of thosel-Sites in relation to the domain interfaces traversed.To get a chain of n arms forming one complex, onlythose tracks where the ≥2-break l-Sites are spaced atvery precise intervals, corresponding with the n − 1interfaces, will satisfy conditions for this exchange,which reduces, drastically, the number of tracks whichcould form large complexes by a surface–surfacemethod.

The segments of condensed chromatin lying be-tween the entrance and exit surfaces of those singlearms in the array centre will be present in the Configu-ration as insertions, the size of which will depend uponthe orientation of the domain with respect to the track.If large enough to be visible, the Pattern will always bea true SEC, but if all the insertions are too small to reg-ister as paint signals, the Pattern will appear as a singlecycle exchange. Intermediate cycle combinations willbe inferred if only some insertions are visible. It is in-teresting to note that Griffin et al. [88] found insertionsto be a frequent feature of �-induced exchanges.



If only short homologous stretches are required, theenormous amount of DNA in the eukaryotic nucleusmust mean that no dsb will be very far from a suitablesequence. The sequence does not have to be on thehomologous chromosome, as appears to be the casein yeast [120], for no consistent significant excess ofhomologous/homologous exchanges is observed in ei-ther plant or animal material.

In an emergency situation, as precipitated by a ra-diation insult, the mechanism may have to make dowith partial homology, and consequent enhanced prob-ability of aberrant rejoining. The yeast studies [120]showed that poor, or absent homology causes failureof dsb repair, and subsequent chromosome loss. Onecan speculate that, for a long homology requirement,partial segments might be gleaned from several lo-cal chromosomes, which would be another recipe forcomplex formation. The large amounts of interspersedrepetitive DNA sequences within the genome (LINESand SINES, etc.) will also facilitate highly localisedrecombinational events.

Whatever the mechanism, the one-lesion approachhelps to overcome the need for large numbers of in-teractive breaks in one l-Site.

The recent observations of an almost linear dose–response curve for simple (c2) exchanges, and the veryhigh efficiency of carbon-K and other USX radiationsfor producing complexes, offer some support for theidea that fewer lesions (perhaps of a more devastatingkind [102]) are required to effect exchanges.

Work using premature chromosome condensationhas shown that the dicentrics formed in human lym-phocytes during the first couple of hours after X-rayshave a linear dose–response curve, and it takes about8 h for significant curvature to develop [121]. This2-stage response might suggest that c2 exchanges formquickly, and that complexes form later, perhaps by adifferent mechanism [122].

Another possibility is that c2 form from relatively“clean” dsb, easily dealt with, but that the more shat-tered “dirty” dsb pose problems, their repair takeslonger, and is more error-prone so they are likelyto end up in complexes. A study of the types andfrequencies of chromosome-type complex exchangesproduced by restriction endonucleases (RE) might bequite informative here, for these enzymes producehigh densities of very clean dsb. They are very ef-ficient at producing complex chromatid-type inter-

and intra-changes, many of which appear to be SEC(non-obligate [28]). One might therefore expect thatthe majority of RE-induced chromosome-type com-plexes also to be SEC, or at least “reducible”.

Further support for a one-track, or one lesion,mechanism for c2 exchanges comes from an exper-iment where fractionated X-ray doses were given tocontact-inhibited untransformed human fibroblasts(Simpson and Savage, unpublished). Scoring ex-changes involving painted chromosomes 1 and 2, wefound that only the complex component declined.

However, mathematical simulation of the one-lesionprocess, using models that have proved very success-ful for the conventional B&R approach, leads to anunder-prediction of complex exchanges [123].

14. Exchange initiation by signal

There remains yet another possibility, namely thatthe dsb introduced are not themselves directly involvedin exchange formation, but act as a “signal” whichinitiates a chain of molecular events that can lead tothe formation of aberrations in (local?) undamagedchromatin. This is the basis of the “signal model”,introduced recently by Bryant [124] to account for theorigin of chromatid-type breaks. There is no evidenceat present to indicate that interchanges or complexmulti-chromosome-type exchanges can be triggered inthis way, but it could be a process by which the needfor a large collection of proximate breaks to producesuch aberrations is eliminated.

Studies of recurrent chromosome aberrations in can-cer cells indicate that there are numerous “hot-spots”in a karyotype prone to change, and there is plentyof evidence for controlled, directed, DNA breakageand reorganisation of several kinds [125], all of whichcould be initiated by appropriate signals. “Signalling”is becoming an increasingly important factor for con-trolling molecular events in cells, so there is likely tobe a fruitful field for studying its relevance to aberra-tion formation.

15. Epilogue

There is no doubt that FISH techniques have throwna lot of our ideas back into the melting pot. Reflecting


presumably requires the presence of additional ma-trix associated structures. Such structures, togetherwith matrix attachment sites, provide opportunityfor another type of induced lesion, the DNA–proteincross-link (DPC [113,114]). Very little is known aboutthe repair of such lesions, or whether they play a rolein exchange formation, but their removal is likelyto be complicated enough for them to contribute tostructural chromosome damage.

There will probably be many such associations ofvarious sizes within a nucleus, and it seems quitelikely that in actively cycling cells, extrusion andlinkage to matrix-bound structures is likely to repre-sent a dynamic and transitory situation, so the p-Site(and probably the l-Site) status is probably in a stateof continuous flux. The aberration results from anacute radiation dose then represent a “flash-photo”of a cell’s interphase situation at the time of expo-sure, and not some permanent feature of intra-nucleararchitecture.

However, even if there were only a few identicalfixed associations or links in every target cell, the num-ber of possible Configurations/Patterns that can arisefrom even a C/A/3(initial) means that one would needto score a very large number of exchanges to detectthem by significant departures from participation ran-domness.

Another possibility that might provide multi-chro-mosome p-Sites is rosette fusions. In their elegant EMstudies of chromosome construction, Mullinger andJohnson [115,116] depict chromatids being built upfrom rows of lateral fusions of the core regions ofrosettes. Various levels of fusion can be seen in manyplaces away from the chromatid cores, and whilstit is probable that the “constructional fusions” areextremely precise, one cannot help speculating fromthe photographs that rosette core fusions may be amore general phenomenon. If so, such fusions maybe a regular feature of extruded chromatin, enhancingpre-formed proximity conditions.

12. Nuclear-pore vicinity p- and l-Sites

Many of the suggestions discussed for inter-domainregions apply also to the considerable amount of chro-matin associated with the nuclear envelope laminaand the pore complexes. Part of the envelope itself is

formed at telophase from chromosome linked residuesof the parent nucleus. Many telomeric regions, chro-mocentres and nucleoli also have attachments to theenvelope membranes, and the rather frequent occur-rence of telomeric segments involved in exchanges(one-way exchanges) is suggestive of the importanceof this region in exchange formation.

The close relationship of the envelope to the endo-plasmic reticulum makes it highly likely that this par-ticular chromatin location is functional, and thereforewill be highly specific and organised, again providingnecessary conditions promoting multi-chromosomeexchange should large enough lesion clusters occur.However, if exchange formation was confined to thisregion, some departures from random arm participa-tion might be expected.

Bearing in mind the extreme attenuation ofcarbon-K USX as it transits a nucleus, part of its ef-ficiency for producing complex exchanges might liein the high entrance dose received by the membrane-associated chromatin [87], though such an explanationwould not suffice for �-particles.

13. Lesion–non-lesion interactions

So far we have assumed that all exchanges arisefrom the interaction between pre-formed radiation-induced lesions. The problem of achieving such largenumbers of lesions (and chromosomes) within a verysmall compass, required for the high exchange-cycleorders often recovered (discussion [20]), and whichwe have been outlining above, has led a number ofworkers to look for alternative possibilities. Severalnow favour the idea that only one (or a very few)lesions are actually required to initiate an exchangeprocess. These lesions then invoke interaction(s) withlocal undamaged chromatin, possibly by a recombi-national repair mechanism.The idea of single-lesionexchanges was formalised by Chadwick and Leen-houts [117–119] although the suggestion of such amechanism has been around for a very long time ([1],p. 623). The repair process envisaged requires veryclose association between DNA carrying the induceddsb and some unaffected homologous DNA. Therewill also be the need for an enzyme/protein assemblyto effect the recombinational processes that can leadto an exchange Configuration.



[34] J.R.K. Savage, J.D. Tucker, Nomenclature systems forFISH-painted chromosome aberrations, Mutat. Res. 366(1996) 153–161.

[35] J.D. Tucker, W.F. Morgan, A.A. Awa, M. Bauchinger, D.Blakey, M.N. Cornforth, G.L. Littlefield, A.T. Natarajan,C. Shassere, A proposed system for scoring structuralaberrations detected by chromosome painting, Cytogenet.Cell Genet. 68 (1995) 211–221.

[36] P.J. Simpson, J.R.K. Savage, Detecting ‘hidden’ exchangeevents within X-ray-induced aberrations using multicolourchromosome paints, Chromosome Res. 3 (1995) 69–72.

[37] P.J. Simpson, J.R.K. Savage, Estimating the true frequencyof X-ray-induced complex chromosome exchanges usingfluorescence in situ hybridization, Int. J. Radiat. Biol. 67(1995) 37–45.

[38] P.J. Simpson, D.G. Papworth, J.R.K. Savage, X-ray inducedsimple, pseudosimple, and complex exchanges involvingtwo distinctly painted chromosomes, Int. J. Radiat. Biol. 75(1999) 11–18.

[39] J.M. Brown, M.S. Kovacs, Visualization of non-reciprocalchromosome exchanges in irradiated human fibroblasts byfluorescence in situ hybridization, Radiat. Res. 136 (1993)71–76.

[40] J.N. Lucas, Cytogenetic signature for ionising radiation, Int.J. Radiat. Biol. 73 (1998) 15–20.

[41] P.J. Simpson, J.R.K. Savage, Dose–response curves forsimple and complex chromosome aberrations induced byX-rays and detected using fluorescence in situ hybridization,Int. J. Radiat. Biol. 69 (1996) 429–436.

[42] J.J.W.A. Boei, S. Vermeulen, J. Fomina, A.T. Natarajan,Detection of incomplete exchanges and interstitial fragmentsin X-irradiated human lymphocytes using a telomeric PNAprobe, Int. J. Radiat. Biol. 73 (1998) 599–603.

[43] J.J.W.A. Boei, S. Vermeulen, A.T. Natarajan, Analysis ofradiation-induced chromosomal aberrations using telomericand centromeric PNA probes, Int. J. Radiat. Biol. 76 (2000)163–167.

[44] J. Fomina, F. Darroudi, J.J.W.A. Boei, A.T. Natarajan,Discrimination between complete and incomplete chromo-some exchanges in X-irradiated human lymphocytes usingFISH with pan-centromeric and chromosome-specific DNAprobes in combination with telomeric PNA probes, Int. J.Radiat. Biol. 76 (2000) 807–813.

[45] J. Fomina, F. Darroudi, A.T. Natarajan, Accurate detection oftrue incomplete exchanges in human lymphocytes exposed toneutron radiation using chromosome painting in combinationwith telomeric PNA probes, Int. J. Radiat. Biol. 77 (2001)1175–1183.

[46] Y. Kodama, M. Nakano, K. Ohtaki, A.A. Awa, N. Nakamura,R. Delongchamp, Estimation of minimal size of translocatedchromosome segments detectable by fluorescence in situhybridization, Int. J. Radiat. Biol. 71 (1997) 35–39.

[47] J.R.K. Savage, Some thoughts on “Sequential ExchangeComplexes” (SEC) and the principle of exchangeconfiguration “reducibility”, unpublished monograph, 2001.

[48] A.A. Edwards, J.R.K. Savage, Is there a simple answer tothe origin of complex exchanges? Int. J. Radiat. Biol. 75(1999) 19–22.

[49] B.D. Loucas, M.N. Cornforth, Complex chromosomeexchanges induced by �-rays in human lymphocytes: anmFISH study, Radiat. Res. 155 (2001) 660–671.

[50] J.R.K. Savage, Interchange and intra-nuclear architecture,Environ. Mol. Mutagen. 22 (1993) 234–244.

[51] U. Fano, Mechanism of induction of gross chromosomalrearrangements in Drosophila sperms, Proc. Natl. Acad. Sci.U.S.A. 29 (1943) 12–18.

[52] B.P. Kaufmann, A complex induced rearrangement ofDrosophila chromosomes and its bearing on the problem ofchromosome recombination, Proc. Natl. Acad. Sci. U.S.A.29 (1943) 8–12.

[53] T. Casperson, L. Zech, C. Johansson, Analysis of humanmetaphase chromosome set by aid of DNA-bindingfluorescent agents, Exp. Cell Res. 62 (1970) 490–492.

[54] A.T. Sumner, H.J. Evans, R.A. Buckland, New techniquefor distinguishing human chromosomes, Nat. New Biol. 232(1971) 31–32.

[55] M. Seabright, A rapid banding technique for humanchromosomes, Lancet 2 (1971) 971–972.

[56] B. Dutrillaux, J. Lejeune, Sur une nouvelle techniqued’analyse du caryotype humain, C. R. Acad. Sci. Paris Ser.D 272 (1971) 2638–2640.

[57] A.T. Sumner, Chromosome Banding, Unwin Hyman,London, 1990.

[58] J.J. Yunis, Mid-prophase human chromosomes: the attain-ment of 2000 bands, Hum. Genet. 56 (1981) 293–298.

[59] J.R.K. Savage, Assignment of aberration break-points inbanded chromosomes, Nature 270 (1977) 513–514.

[60] M. Seabright, The use of proteolytic enzymes for themapping of structural rearrangements in the chromosomesof man, Chromosoma 36 (1972) 204–210.

[61] M. Seabright, High resolution studies on the patterns ofinduced exchanges in the human karyotype, Chromosoma40 (1973) 333–346.

[62] H.J. Evans, K.E. Buckton, A.T. Sumner, Cytologicalmapping of human chromosomes: results obtained withquinacrine fluorescence and the acetic–saline–Giemsatechniques, Chromosoma 35 (1971) 310–325.

[63] ISCN, An international system for human cytogeneticnomenclature (1995), Cytogenet. Cell Genet. 6 (1996)1–114.

[64] J.R.K. Savage, Application of chromosome bandingtechniques to the study of primary chromosome structuralchanges, J. Med. Genet. 14 (1977) 362–370 (Annotation).

[65] D.S. Borgaonkar, R. Shafer, N. Reisor (Eds.), Repositoryof human chromosomal variants and anomalies, Aninternational registry of abnormal karyotypes, 12th Listing,Medical Center of Delaware, Newark, USA, 1987, 778 pp.

[66] J.L. Gorski, M.L. Kistenmacher, H.H. Punnett, E.H. Zachai,B.S. Emanuel, Reproductive risks for carriers of complexchromosome rearrangements, Am. J. Med. Genet. 29 (1988)247–261.

[67] T. Cremer, S. Popp, P. Emmerich, P. Lichter, C. Cremer,Rapid metaphase and interphase detection of radiation-induced chromosome aberrations in human lymphocytes bychromosomal suppression in situ hybridization, Cytometry11 (1990) 110–118.


on my more than 40 years of chromosomal aberrationresearch, I find myself echoing the concluding remarksof Fano in his 1941 paper [31]:

All these considerations represent approximatelythe present day line of advance . . . an advancewhich has not yet yielded a clear picture of the phe-nomena under investigation. If the phenomena ap-pear now more complex, perhaps, than it had beenhitherto realised, this means that the advance hasprogressed just beyond its preliminary stage.

So, there are lots of exciting times ahead!

References

[1] H. Bauer, M. Demerec, B.P. Kaufmann, X-ray inducedchromosomal alterations in Drosophila melanogaster, Gene-tics 23 (1938) 610–630.

[2] K. Sax, An analysis of X-ray induced chromosomal aberra-tions in Tradescantia, Genetics 25 (1940) 41–68.

[3] K. Sax, Types and frequencies of chromosomal aberrationsinduced by X-rays, Cold Spring Harb. Symp. Quant. Biol.9 (1941) 93–103.

[4] D.E. Lea, Actions of Radiations on Living Cells, first ed.,Cambridge University Press, 1946 (second ed., 1955).

[5] D.G. Catcheside, Effects of ionizing radiations onchromosomes, Biol. Rev. 20 (1945) 14–28.

[6] D.G. Catcheside, Genetic effects of radiation, Adv. Genet.2 (1948) 271–358.

[7] N.H. Giles, Radiation induced chromosome aberrations inTradescantia, in: A. Hollaender (Ed.), Radiation Biology,vol. 1, Part 2, McGraw Hill, New York, 1954, pp. 713–761.

[8] S. Wolff, Radiation genetics, in: M. Errera, A. Forssberg(Eds.), Mechanisms in Radiation Biology, vol. I, AcademicPress, New York, 1961, pp. 419–475.

[9] H.J. Evans, Chromosome aberrations induced by ionizingradiations, Int. Rev. Cytol. 13 (1962) 221–321.

[10] H.J. Evans, Molecular mechanisms in the induction ofchromosome aberrations, in: D. Scott, B.A. Bridges, F.H.Sobels (Eds.), Progress in Genetic Toxicology, Elsevier,Amsterdam, 1977, pp. 57–74.

[11] B.A. Kihlman, Molecular mechanisms of chromosomebreakage and rejoining, Adv. Cell Mol. Biol. 1 (1971) 59–107.

[12] J.R.K. Savage, Radiation-induced chromosomal aberrationsin the plant Tradescantia: dose–response curves. I.Preliminary considerations, Radiat. Bot. 15 (1975) 87–140.

[13] J.R.K. Savage, The production of chromosome structuralchanges by radiation an update of Lea (1946) Chapter VI,Br. J. Radiol. 62 (1989) 507–520.

[14] D. Scott, Molecular mechanisms of chromosome structuralchanges, in: M. Alacevic (Ed.), Progress in EnvironmentalMutagenesis, Elsevier, Amsterdam, 1980, pp. 101–113.

[15] R.F. Kimball, The development of ideas about the effectof DNA repair on the induction of gene mutations andchromosomal aberrations by radiation and chemicals, Mutat.Res. 186 (1987) 1–34.

[16] M.N. Cornforth, J.S. Bedford, Ionizing radiation damageand its early development in chromosomes, Adv. Radiat.Biol. 17 (1993) 423–496.

[17] K. Sax, Chromosome aberrations induced by X-rays,Genetics 23 (1938) 494–516.

[18] J.R.K. Savage, A brief survey of aberration origin theories,Mutat. Res. 404 (1998) 139–147.

[19] J.R.K. Savage, Sites of radiation induced chromosomeexchanges, Curr. Topics Radiat. Res. 6 (1970) 129–194.

[20] J.R.K. Savage, Insight into sites, Mutat. Res. 366 (1996)81–95.

[21] J.R.K. Savage, Proximity matters, Science 290 (2000) 62–63 (Perspective).

[22] R.K. Sachs, A.M. Chen, D.J. Brenner, Proximity effectsin the production of chromosome aberrations by ionizingradiation, Int. J. Radiat. Biol. 71 (1997) 1–19 (Review).

[23] D.T. Goodhead, Radiation tracks in biological materials:initial damage in cells, DNA and associated structures,in: M.L. Mendelsohn (Ed.), Genes, Cancer and RadiationProtection, National Council on Radiation Protection andMeasurements, Bethesda, 1992, pp. 25–37.

[24] K.C. Atwood, Numbers of nuclear sites for aberrationformation and the distribution of aberrations, in: S.Wolff (Ed.), Radiation-induced Chromosome Aberrations,Columbia University Press, New York, 1963, pp. 73–86.

[25] J.R.K. Savage, P.J. Simpson, FISH “painting” patternsresulting from complex exchanges, Mutat. Res. 312 (1994)51–60.

[26] J.R.K. Savage, P.J. Simpson, On the scoring of FISH-“painted” chromosome-type exchange aberrations, Mutat.Res. 307 (1994) 345–353.

[27] M.N. Cornforth, Analyzing radiation-induced complexchromosome rearrangements by combinatorial painting,Radiat. Res. 155 (2001) 643–659.

[28] J.R.K. Savage, Classification and relationships of inducedchromosomal structural changes, J. Med. Genet. 13 (1976)103–122 (Annotation).

[29] J.R.K. Savage, A comment on the terms “Asymmetrical”,“Symmetrical”, applied to chromosomal exchanges, Clin.Cytogenet. Bull. 2 (1992) 97–99.

[30] R.K. Sachs, A.M. Chen, P.J. Simpson, L.R. Hlatky, P.Hahnfeldt, J.R.K. Savage, Clustering of radiation-producedbreaks along chromosomes: modeling the effects onchromosome aberrations, Int. J. Radiat. Biol. 75 (1999) 657–672.

[31] U. Fano, On the analysis and interpretation of chromosomalchanges in Drosophila, Cold Spring Harb. Symp. Quant.Biol. 9 (1941) 113–120.

[32] R.K. Sachs, J. Arsuaga, M. Vázquez, P. Hahnfeldt, L.R.Hlatky, Using Graph Theory to describe and modelchromosome aberrations, Radiat. Res., in press.

[33] J.R.K. Savage, The transmission of FISH-painted patternsderived from complex chromosome exchanges, Mutat. Res.347 (1995) 87–95.



[68] J.W. Gray, J.N. Lucas, D. Pinkel, A.A. Awa, Structuralchromosome analysis by whole chromosome painting forassessment of radiation-induced genetic damage, J. Radiat.Res. 33 (1992) 80–86.

[69] J.W. Gray, D. Pinkel, J.M. Brown, Fluorescence in situhybridization in cancer and radiation biology, Radiat. Res.137 (1994) 275–289 (Review).

[70] J.N. Lucas, A.A. Awa, T. Straume, M. Poggensee, Y.Kodama, M. Nakano, K. Ohtaki, H.U. Weier, D. Pinkel, J.Gray, G. Littlefield, Rapid translocation frequency analysisin humans decades after exposure to ionizing radiation, Int.J. Radiat. Biol. 62 (1992) 53–63.

[71] J.N. Lucas, R.K. Sachs, Using three-colour chromosomepainting to test chromosome aberration models, Proc. Natl.Acad. Sci. U.S.A. 90 (1993) 1484–1487.

[72] E. Schmid, H. Zitzelsberger, H. Braselmann, J.W. Gray,M. Bauchinger, Radiation-induced chromosome aberrationsanalysed by fluorescence in situ hybridization with a triplecombination of composite whole chromosome-specific DNAprobes, Int. J. Radiat. Biol. 62 (1992) 673–678.

[73] M. Bauchinger, E. Schmid, H. Zitzelsberger, H. Braselmann,U. Nahrstedt, Radiation-induced chromosome aberrationsanalysed by two-colour fluorescence in situ hybridizationwith composite whole chromosome-specific DNA probesand a pancentromeric DNA probe, Int. J. Radiat. Biol. 64(1993) 170–184.

[74] S. Knehr, H. Zitzelsberger, H. Braselmann, E. Schmid, M.Bauchinger, Analysis for DNA-proportional distribution ofradiation-induced chromosome aberrations in various triplecombinations of human chromosomes using fluorescence insitu hybridization, Int. J. Radiat. Biol. 65 (1994) 683–690.

[75] S. Knehr, H. Zitzelsberger, U. Nahrstedt, M. Bauchinger,FISH-based analysis of radiation-induced chromosomalaberrations using different nomenclature systems, Int. J.Radiat. Biol. 73 (1998) 135–141.

[76] M.R. Speicher, S.G. Ballard, D.C. Ward, Karyotypinghuman chromosomes by combinatorial multi-fluor FISH,Nat. Genet. 12 (1996) 368–375.

[77] E. Schrock, S. du Manior, T. Veldman, B. Schoell, J.Wienberg, M.A. Ferguson-Smith, Y. Ning, D.H. Ledbetter,I. Bar-Am, D. Soenksen, Y. Garini, T. Reid, Multicolorspectral karyotyping of human chromosomes, Science 273(1996) 494–497.

[78] H.J. Tanke, J. Wiegant, R.P.M. van Gijlswijk, V. Bezrookove,H. Pattenier, R.J. Heetebrij, E.G. Talman, A.K. Raap, J.Vrolijk, New strategy for multi-colour fluorescence in situhybridization: COBRA: Combined Binary Ratio Labelling,Eur. J. Hum. Genet. 7 (1999) 2–11.

[79] R.M. Anderson, S.J. Marsden, E.G. Wright, M.A. Kadhim,C.S. Griffin, Complex chromosome aberrations in peripheralblood lymphocytes as a potential biomarker of exposure tohigh-LET alpha particles, Int. J. Radiat. Biol. 76 (2000)31–42.

[80] R.M. Anderson, D.L. Stevens, D.T. Goodhead, mFISHanalysis shows that complex chromosome aberrationsinduced by �-particle tracks are cumulative products oflocalised rearrangements, Proc. Natl. Acad. Sci. U.S.A., inpress.

[81] M. Vazquez, K. Greulich-Bode, J. Arsuaga, M. Cornforth,M. Brückner, R.K. Sachs, P. Hahnfeldt, M. Molls, L. Hlatky,Computer analysis of mFISH chromosome aberration datademonstrates an excess of very complex metaphases, Int. J.Radiat. Biol., in press.

[82] A.M. Chen, J.N. Lucas, F.S. Hill, D.J. Brenner, R.K. Sachs,Chromosome aberrations produced by ionizing radiation:Monte Carlo simulations and chromosome painting data,Comput. Appl. Biosci. 11 (1995) 389–397.

[83] A.M. Chen, J.N. Lucas, P.J. Simpson, C.S. Griffin, J.R.K.Savage, D.J. Brenner, L.R. Hlatky, R.K. Sachs, Computersimulation of data on chromosome aberrations produced byX-rays or alpha-particles and detected by fluorescence insitu hybridization, Radiat. Res. 148 (1997) 93–101.

[84] R.K. Sachs, D. Levy, A.M. Chen, P.J. Simpson, M.N.Cornforth, E.A. Ingerman, P. Hahnfeldt, L.R. Hlatky,Random breakage-and-reunion chromosome formationmodel: an interaction-distance implementation based onchromatin geometry, Int. J. Radiat. Biol. 76 (2000) 129–148.

[85] R.R. Swiger, J.D. Tucker, Fluorescence in situ hybridization:a brief review, Environ. Mol. Mutagen. 27 (1996) 245–254.

[86] J.J.W.A. Boei, S. Vermeulen, L.H.F. Mullenders, A.T.Natarajan, The impact of radiation quality on the spectrumof induced chromosome exchange aberrations, Int. J. Radiat.Biol. 77 (2001) 847–857.

[87] C.S. Griffin, M.A. Hill, D.G. Papworth, S. Townsend,J.R.K. Savage, D.G. Goodhead, Effectiveness of 0.28 keVcarbon ultrasoft X-rays at producing simple and complexchromosome exchanges in human fibroblasts in vitro usingFISH, Int. J. Radiat. Biol. 73 (1998) 591–598.

[88] C.S. Griffin, S.J. Marsden, D.L. Stevens, P. Simpson,J.R.K. Savage, Frequencies of complex chromosomeexchange aberrations induced by 238Pu alpha-particles anddetected by fluorescence in situ hybridization using singlechromosome-specific probes, Int. J. Radiat. Biol. 67 (1995)431–439.

[89] C.S. Griffin, D.L. Stevens, J.R.K. Savage, Ultrasoft 1.5keV aluminium-K X-rays are efficient producers ofcomplex chromosome exchange aberrations as revealed byfluorescence in situ hybridization, Radiat. Res. 146 (1996)144–150.

[90] A.A. Edwards, P.A. Hone, J.E. Moquet, D.C. Lloyd, Simplechromosome exchanges are not linear with dose, Int. J.Radiat. Biol. 75 (1999) 1113–1117.

[91] J.R.K. Savage, D.G. Papworth, Excogitations about thequantification of structural chromosomal aberrations, in:G. Obe (Ed.), Advances in Mutagenesis Research, vol. 3,Springer, Berlin, 1991, pp. 162–189.

[92] J.R.K. Savage, On the dose–effect relationship of true simplechromosomal rearrangements, Int. J. Radiat. Biol. 76 (2000)1428–1429 (Letter).

[93] J.R.K. Savage, A note on inter-arm intrachange patternsresulting from dual-arm FISH painting, Mutat. Res. 373(1997) 265–269.

[94] C. Cremer, Ch. Münkel, M. Granzow, A. Jauch, S. Dietzel,R. Eils, X.Y. Guan, P.S. Meltzer, J.M. Trent, J. Langowski,T. Cremer, Nuclear architecture and the induction ofchromosomal aberrations, Mutat. Res. 366 (1996) 97–116.


[95] T. Cremer, C. Cremer, Chromosome territories, nucleararchitecture and gene regulation in mammalian cells, Nat.Rev. Genet. 2 (2001) 292–301.

[96] S.M. Gasser, U.K. Laemmli, A glimpse at chromosomalorder, Trends Genet. 3 (1987) 16–22.

[97] K.J. Pienta, D.S. Coffey, DNA organization in the interphasenucleus and metaphase chromosome, in: T. Sharma (Ed.),Trends in Chromosome Research, Narosa Press, New Delhi,1990, pp. 1–5.

[98] T. Haaf, M. Schmid, Chromosome topology in mammalianinterphase nuclei, Exp. Cell Res. 192 (1991) 325–332.

[99] R. Berezney, Mapping the organization and functionaltopography of the cell nucleus, in: W.C. Dewey, M.Eddington, R.J.M. Fry, E.J. Hall, G.F. Whitmore (Eds.),Radiation Research: A Twentieth Century Perspective,Academic Press, New York, 1982, pp. 217–223.

[100] S.M. Gasser, B.B. Amati, M.E. Cardenas, J.F.X. Hofmann,Studies on scaffold attachment sites and their relation togenome function, Int. Rev. Cytol. 119 (1989) 57–96.

[101] P. Engelhardt, K. Pusa, Nuclear pore complexes: “press-stud”elements of chromosomes in pairing and control, Nat. NewBiol. 101 (1972) 163–166.

[102] D.T. Goodhead, J. Thacker, R. Cox, Effects of radiationsof different qualities on cells: molecular mechanisms ofdamage and repair (Weiss lecture), Int. J. Radiat. Biol. 63(1993) 543–556.

[103] D.T. Goodhead, Initial events in the cellular effects ofionizing radiations: clustered damage in DNA, Int. J. Radiat.Biol. 65 (1994) 7–17.

[104] H.M. Cundy, A.P. Rollett, Mathematical Models, third ed.,Tarquin Publications, Norfolk, UK, 1981.

[105] J.R.K. Savage, D.G. Papworth, Frequency and distributionstudies of asymmetrical versus symmetrical chromosomeaberrations, Mutat. Res. 95 (1982) 7–18.

[106] J.R.K. Savage, Testing the participation of chromosomesin structural aberrations, in: R.C. Sobti, G. Obe (Eds.),Eukaryotic Chromosomes: Structural and Functional Aspe-cts, Narosa Press, New Delhi, 1991, pp. 111–125.

[107] S. Knehr, H. Zitzelsberger, H. Braselmann, U. Nahrstedt,M. Bauchinger, Chromosome analysis by fluorescence insitu hybridization: further indications for a non-DNA-proportional involvement of single chromosomes inradiation-induced structural aberrations, Int. J. Radiat. Biol.70 (1996) 385–392.

[108] J.F. Barquinero, S. Knehr, H. Braselmann, M. Figel, M.Bauchinger, DNA-proportional distribution of radiation-induced chromosome aberrations analysed by fluorescencein situ hybridization painting of all chromosomes of ahuman female karyotype, Int. J. Radiat. Biol. 74 (1998)315–323.

[109] Y. Xiao, A.T. Natarajan, Non-random involvement ofChinese hamster chromosomes 3, 4, 8, and 9 in X-rayinduced chromosomal aberrations, Int. J. Radiat. Biol. 75(1999) 943–951.

[110] E.A. Nigg, Nuclear function and organization: the potentialof immunochemical approaches, Int. Rev. Cytol. 110 (1988)27–92.

[111] T.A. Okada, D.E. Comings, Higher order structure ofchromosomes, Chromosoma 72 (1979) 1–14.

[112] J.Y. Ostashevsky, Higher-order structure of interphasechromosomes and radiation-induced chromosomal exchangeaberrations, Int. J. Radiat. Biol. 76 (2000) 1179–1187.

[113] N.L. Oleinick, S.M. Chiu, Chromatin and nuclear structuralfactors which modify the radiosensitivity of DNA, in:W.C. Dewey, M. Edington, R.J.M. Fry, E.J. Hall, G.F.Whitmore (Eds.), Radiation Research: A Twentieth CenturyPerspective, vol. II, Academic Press, San Diego, 1992,pp. 149–153.

[114] N.L. Oleinick, S.M. Chiu, Nuclear and chromatin structuresand their influence on the radiosensitivity of DNA, Rad.Protec. Dosim. 52 (1994) 353–358.

[115] A.M. Mullinger, R.T. Johnson, Packing DNA intochromosomes, J. Cell Sci. 46 (1980) 61–86.

[116] A.M. Mullinger, R.T. Johnson, Units of chromosomereplication and packing, J. Cell Sci. 64 (1983) 179–193.

[117] K.H. Chadwick, H.P. Leenhouts, The rejoining of DNAdouble-strand breaks and a model for the formation ofchromosomal rearrangements, Int. J. Radiat. Biol. 33 (1978)517–529.

[118] K.H. Chadwick, H.P. Leenhouts, The Molecular Theory ofRadiation Biology, Springer, Berlin, 1981.

[119] K.H. Chadwick, H.P. Leenhouts, On the dose–effectrelationship of true simple chromosomal rearrangements.Comments on the paper: simple chromosome exchanges arenot linear with dose, Int. J. Radiat. Biol. 76 (2000) 1425–1429 (Letter).

[120] M.A. Resnick, T. Nilsson-Tillgren, DNA homology andchromosome stability: a sensitive yeast genetic systemfor identifying double-stranded DNA damage, in: M.L.Mendelsohn, R.J. Albertini (Eds.), Mutation and theEnvironment. B: Metabolism, Testing Methods andChromosomes, Wiley, New York, 1990, pp. 363–369.

[121] R. Greinert, E. Detzler, B. Volkmer, D. Harder, Kineticsof the formation of chromosome aberrations in X-irradiatedhuman lymphocytes: analysis by premature chromosomecondensation with delayed fusion, Radiat. Res. 144 (1995)190–197.

[122] F. Darroudi, J. Fomina, M. Mijers, A.T. Natarajan, Kineticsof the formation of chromosome aberrations in X-irradiatedhuman lymphocytes using PCC and FISH, Mutat. Res. 404(1998) 55–65.

[123] R.K. Sachs, A. Rogoff, A.M. Chen, P.J. Simpson, J.R.K.Savage, P. Hahnfeldt, L.R. Hlatky, Underprediction ofvisibly complex chromosome aberrations by a recombina-tional-repair (one-hit) model, Int. J. Radiat. Biol. 76 (2000)129–148.

[124] P.E. Bryant, The signal model: a possible explanation forthe conversion of DNA double-strand breaks into chromatidbreaks, Int. J. Radiat. Biol. 73 (1998) 243–251.

[125] H.J. Evans, Cytogenetics: overview, in: M.L. Mendelsohn,R.J. Albertini (Eds.), Mutation and the Environment. B:Metabolism, Testing Methods and Chromosomes, Wiley,New York, 1990, pp. 301–323.



Shedding light on proteins, nucleic acids,cells, humans and fish ☆Richard B. Setlow *

Biology Department, Brookhaven National Laboratory, P.O. Box 5000, Upton, NY 11973, USAAccepted 8 January 2002



Shedding light on proteins, nucleic acids,cells, humans and fish�

Richard B. Setlow∗

Biology Department, Brookhaven National Laboratory, P.O. Box 5000, Upton, NY 11973, USA

Accepted 8 January 2002

Abstract

I was trained as a physicist in graduate school. Hence, when I decided to go into the field of biophysics, it was natural thatI concentrated on the effects of light on relatively simple biological systems, such as proteins. The wavelengths absorbed bythe amino acid subunits of proteins are in the ultraviolet (UV). The wavelengths that affect the biological activities, the actionspectra, also are in the UV, but are not necessarily parallel to the absorption spectra. Understanding these differences led me toinvestigate the action spectra for affecting nucleic acids, and the effects of UV on viruses and cells. The latter studies led meto the discovery of the important molecular nature of the damages affecting DNA (cyclobutane pyrimidine dimers) and to thediscovery of nucleotide excision repair. Individuals with the genetic disease xeroderma pigmentosum (XP) are extraordinarilysensitive to sunlight-induced skin cancer. The finding, by James Cleaver, that their skin cells were defective in DNA repairstrongly suggested that DNA damage was a key step in carcinogenesis. Such information was important for estimating thewavelengths in sunlight responsible for human skin cancer and for predicting the effects of ozone depletion on the incidenceof non-melanoma skin cancer. It took experiments with backcross hybrid fish to call attention to the probable role of the longerUV wavelengths not absorbed by DNA in the induction of melanoma. These reflections trace the biophysicist’s path frommolecules to melanoma. © 2002 Elsevier Science B.V. All rights reserved.

Keywords: Action spectra; DNA repair; Pyrimidine dimers; Skin cancers; Ultraviolet light; Excision repair; Photorepair; History of science

1. Introduction

I learned at Swarthmore College to love physics. Itwas, to me, a very logical, understandable and enjoy-able subject. So, I entered Yale in 1941 as a graduatestudent in the Physics Department supported by astipend, for the first year, that required me to help the

� This article is part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, readers shouldcontact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).

∗ Tel.: +1-631-344-3391; fax: +1-631-344-6391.E-mail address: [email protected] (R.B. Setlow).

faculty of the Medical School if they had problemsor concerns about physics. My co-worker was RolandMeyerott, a physics faculty member. We became goodfriends. Although I did not know it at the time, thiswas a valuable starting point for the interdisciplinarycareer that, unbeknownst to me, I was ultimately toembark upon.World War II saw an influx into Yale of groups of

students for undergraduate degrees who were candi-dates for Navy Officer Training. I became a TeachingAssistant, teaching physics laboratory and discussion/problem sessions in elementary physics. Because ofweak eyesight, I was not drafted and was appointed as

1383-5742/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S1 3 8 3 -5 7 42 (02 )00004 -2

2 R.B. Setlow / Mutation Research 511 (2002) 1–14

an Instructor with more teaching assignments. Rolandbecame my thesis advisor. My research, at that time,involved far-ultraviolet (UV) spectroscopy of N2, anessential part of which was to help design and build anappropriate spectrometer. I received a Ph.D. in 1947for “Spectroscopy of High Energy States of N2

+ andN2 ” [1] and became an Assistant Professor in physics,teaching more advanced undergraduate courses and arequired graduate laboratory course in spectroscopy.By then Ernest Pollard (Ernie), a Professor of nu-clear physics, had returned to Yale after a wartimeassignment and resumed research on the cyclotronthat he had put together before the war. He wanted touse physical techniques to investigate the structuresof viruses and large molecules by “target theory,”the destruction of functions resulting from bombard-ment by fast charged particles. In Ernie’s case, thismeant using 4 MeV deuterons and 8 MeV �-particlesfrom the Yale cyclotron. Ernie also was interested inusing UV to inactivate viruses. He organized a Bio-physics Division within the Physics Department. Oneof the graduate students, Donald Fluke, was assignedthe problem of building very large, water-prism UVmonochromators. Influenced by Ernie’s enthusiasm,I decided that biophysics was more interesting thanpure physics and so I joined the division and helpedDon construct the monochromator [2]. The first bio-physics experiments that I did were on the absorptionof peptide bonds at wavelengths <230 nm [3] and onthe inactivation of large proteins by deuterons [4,5].

It soon became apparent that the Physics Depart-ment was not comfortable in evaluating biophysicsstudents or faculty. Hence, a Biophysics Departmentwas created in the mid-1950s. I became an Asso-ciate Professor in both the physics and biophysicsdepartments, and taught courses in both, as well asbeing the Director of Undergraduate Studies for bothdepartments. Ernie and I developed a biophysicscourse for first year graduate students and seniors inthe department. We finally put it together as a book[6], “Molecular Biophysics” that was used interna-tionally. Ernie recognized that the biophysics facultyneeded assistance in carrying out biological and bio-chemical experiments, and was able to hire technicalassistants for us. I was most fortunate that my assis-tant, Barbara Doyle, was very talented, and invaluablein helping me with many experiments on the effectsof UV on proteins and nucleic acids.

2. The effects of UV on proteins

In the early 1950s, proteins were the best-studiedmacromolecules. Their functions and their aminoacid compositions were known. The effects of UVon proteins had not been systematically studied, buttheir quantum yields, the fractions of molecules in-activated per absorbed photon, were ∼0.01 variedamong proteins and seemed to depend on wavelength[7]. By then our high intensity monochromators werecompleted, and ready for use. Barbara Doyle and Iexposed dry chymotrypsin to monochromatic wave-lengths between 230 and 297 nm. The activity remain-ing decreased exponentially with dose in ergs/mm2

(Eq. (1)) at all wavelengths,A/A0 = e−sD, (1)

where A and A0 represent the activities after andbefore a dose D, and s is the inactivation cross sec-tion. The action spectrum, a graph of s versus λ wasnot parallel to the absorption spectrum, indicating thatthe quantum yield was not constant, but increased by∼two-fold at the shorter wavelengths [8]. The expla-nation for this finding did not become apparent to mefor several years. It depended on observations thatthe quantum yield at 254 nm for a number of proteinsincreased more or less linearly with their relative cys-tine content [9]. Photons absorbed by cystine had ahigher chance of inactivating a protein than photonsabsorbed in the aromatic amino acids. One could ex-plain the complicated action spectra of proteins bysumming the effects at 280 nm (absorption maximumof the aromatic amino acids) and the effects at 254 nm(photons absorbed by cystine at the absorption min-imum of aromatic amino acids) [10]. Proteins withnegligible cystine had action spectra very similar totheir absorption spectra (Fig. 1).

John Preiss, a graduate student, had built a vacuummonochromator that could be used to measure absorp-tion and action spectra down to 110 nm [11–13]. Thequantum yield for the inactivation of proteins rosegreatly at the lower wavelengths to ∼0.1 at 120 nm[14] and to 1.0, for ribonuclease, at 110 nm [15]. Thisrapid rise in quantum yield was associated with theemission of electrons from the proteins (a photoelec-tric effect). At 110 nm one absorbed photon resulted inan ionization that inactivated the molecule, just asobserved for the effects of incident ionizing radiation


R.B. Setlow / Mutation Research 511 (2002) 1–14 3

Fig. 1. Action spectra [10] for the inactivation of the enzymestrypsin, 4% cystine, (solid symbols, multiply ordinate by 10−19)and aldolase, 1% cystine (open symbols, multiply ordinate by10−18).

in giving a cross section equal to the size of themolecule.

3. The effects of UV on DNA and viruses

The fact that the action spectrum for proteins mightnot look like their absorption spectrum raised a ques-tion about action spectra for affecting nucleic acids.Their absorption maxima are ∼260 nm. Early dataindicated that these were the most effective wave-lengths for killing bacteria [16] and also for inducingmutations in fungal spores [17], thus implicatingnucleic acids as essential components for life. Theonly biochemical change in UV-irradiated DNA thatI knew of then was a decrease in viscosity. Barbaraand I set out to measure the action spectrum for thisviscosity decrease by exposing dried calf-thymus

DNA, dissolving the DNA and passing it through acapillary viscometer. The experiment failed. The ex-posed DNA clogged the viscometer! So, we turnedthe assay around and measured the cross-linking ofthe exposed DNA by filtering it and determining thesoluble fraction of the material passing through thefilter by its UV absorption [18]. Action and absorp-tion spectra had the same shape and the quantumyield was independent of wavelength. We used thedifference between the action spectra of proteins andnucleic acids to show that paramecin, the killer sub-stance of Paramecium, elaborated by resistant strainswhich killed sensitive strains, was a protein and not anucleic acid as was thought [19].

The four bases of DNA have very different absorp-tion spectra but strongly interact in double-strandedDNA, as indicated by the hyperchromic effect upondenaturation. It seemed possible that single-strandedDNA, such as in the virus �X174, might have anaction spectrum that was dependent on pH, becausethe pyrimidine absorption spectra were pH-dependent.Richard Boyce, a biophysics graduate student, and Idetermined the spectrum for inactivation at severalpHs [20]. High precision was easy to obtain becausethe survival curves were exponential over a range of103 to 104. The changes with pH indicated that proteinwas not involved and that the effects on the pyrim-idines were 2.5-fold greater than those on the purines.Moreover, it seemed that single- and double-strandedpolynucleotides had minima in their action spectra atdifferent wavelengths which might be useful in dis-tinguishing between their physical states in vivo [21].Somewhat later, the action spectrum for inactivatingtransforming DNA of Haemophilus influenzae aboveits melting point was shown to be of the single-strandform [22]. The action spectrum for inactivating T2phage showed a single-strand phase during its replica-tion in Escherichia coli [23]. The action spectrum forinactivating a t-RNA was obtained by a biophysics stu-dent, Faiza Fawaz; it also is single-stranded like [24].

It is of historical interest that Latarjet et al. in 1970determined an action spectrum for the inactivation ofthe scrapie agent [23]. Scrapie was originally calleda “slow virus” but is now understood to be a protein(a prion) containing the amino acid cystine [24]. Therelative sensitivities were determined at 237, 250, 254and 280 nm. The authors concluded that the sensitiv-ities did not correspond to an action spectrum for a


Fig. 2. Action spectra for the inactivation of scrapie (multiplyordinate by 10−19) from data in [23], trypsin as in Fig. 1, and thesmall single-stranded DNA virus �X174 [18] (multiply ordinateby 10−15).

nucleic acid. They failed to note, however, that therelative sensitivities were similar to those for affect-ing a cystine-containing protein such as trypsin [10]as shown in Fig. 2.

In 1960, I moved to the Biology Division of theOak Ridge National Laboratory. The Director wasAlexander Hollaender, a leader and innovator inradiation- and photobiology. The laboratory facilitieswere superb and the excellent staff covered all areasof biology. I was able to devote all my time to lab-oratory research and to collaborate with several staffso as to learn and to do experiments that could notbe done at Yale. An important collaborator was FredBollum, a matchless nucleic acid biochemist and anexpert on calf-thymus DNA polymerase. The knowl-edge of DNA biochemistry that I gathered from himwas invaluable, and he was a source of many ideas

as well as the many polynucleotides and radioactivepolymers essential for my future work. I also learnedfrom David Krieg, a phage geneticist, about the obser-vations of Beukers and Berends [25] who had exposeda frozen solution of thymine to 254 nm, observationsthat changed the main focus of my scientific life.They observed that irradiation generated dimers be-tween thymines and that if the dimers were furtherexposed in liquid solution to 254 nm, they split intomonomers. They suggested that the latter reactionmight have something to do with the phenomenon ofphotoreactivation (PR). I knew better. John Jagger,who had been a biophysics graduate student at Yale,had demonstrated that the most effective wavelengthfor PR of E. coli was in the neighborhood of 365 nm[26]. Thus, I decided to measure the action spectrumfor splitting thymine dimers. It was one of the easiestexperiments that I ever did. The formation of a dimersaturated the double bonds, and absorption at 260 nmdeclined. Splitting increased absorption. There wasa large quartz monochromator in John’s laboratory,where I was working. So, I would expose a solutionof dimers to various wavelengths and doses and thenrun down several hallways with the cuvette to anotherlaboratory where there was spectrophotometer, andmeasure the increase in UV absorption. Then I wouldrace back with the cuvette to the monochromatorand give a further dose, and so on. The action spec-trum increased monotonically from long to short UVwavelengths [27].

Reg Deering, a biophysics Ph.D. from Yale, wasbetween jobs and came to work for a while at OakRidge. He used the changes in UV absorption to in-vestigate dimer formation and splitting in thymidinedinucleotide and in polyT as a function of wavelength[28]. His results indicated that there was an equi-librium between the making and splitting of dimers,TT ↔ T = T, with the reaction shifted to the rightat long wavelengths, and to the left by short wave-lengths. I was fortunate that a highly skilled techni-cian, William (Bill) Carrier came to work with me.He was a great person and a constant scientific com-panion during my years at Oak Ridge; indeed he wasmuch more than a technician. He was an innovatorand a superior investigator. Bill and I used absorptionmeasurements to assess the formation of dimers indifferent DNA’s by exposure to large doses of a longwavelength, such as 280 nm, and their subsequent



splitting by a small dose of shorter wavelengths, suchas 239 nm [29]. The stage was now set to use thephotochemistry of DNA to show that dimers were re-sponsible for biological effects. Jane Setlow and I didthis using transforming DNA of H. influenzae as thetest system. The inactivation of its activity by 280 nmcould be largely reversed, by subsequent exposure to239 nm, with kinetics that were similar to those forsplitting of dimers [30]; accordingly the inactivationcould be ascribed to the formation of dimers. Dimerswere biological lesions. Further experiments showedthat the photochemical splitting of dimers overlappedenzymatic PR by an extract of yeast [31].

Subsequent work showed that dimers in templateDNA inhibited polymerization by calf-thymus poly-merase, creating a product with a deficit of adenine,suggesting that polymerization past a dimer could be acause of UV induced mutations [32]. Dimers also in-hibited the degradation, by nucleases, of UV-irradiatedDNAs and gave limit digests that contained mononu-cleotides, pN, and trinucleotides pNpT = pT [33]. Wealso observed some unidentified non-dimer productsthat had chromatographic mobilities similar to trinu-cleotides. Further work showed that dimers were notonly formed between adjacent Ts but also betweenCC and CT. [34–36]. These dimers had different pho-tochemical kinetics. Obviously, the phrase thyminedimers should be “pyrimidine dimers”.

A query, about the cause for the high sensitivity ofcells containing BrdUrd in place of dThd, posed byNeva Cummings, a high school student working in thelaboratory at Yale, led us to undertake action spectrastudies of substituted E. coli and of substituted T4phage. Richard Boyce, a Ph.D. student at Yale, and Ishowed that the action spectra were shifted to higherwavelengths because BrdUrd had a much higherabsorption coefficient than dThd above 300 nm. Thesensitivity of the substituted systems were 110-fold(cells) and 10-fold (phage) higher than the unsub-stituted ones at 313 nm [37,38]. Several years later,Menachem Lion, an Israeli visitor to my laboratoryat Oak Ridge in 1965, showed that while the BrdUrdreplacement reduced the numbers of thymine dimers,it led to another product that caused strand breaksas detected by sedimentation in alkaline solutions.Menachem was delayed in writing up his resultsbecause of the 1967 war in Israel, but the findingswere published a few years later [39].

4. The effects of UV on cells

At Yale, my focus on molecules in vitro had beenexpanded to include cells by graduate students RegDeering, Bob van Tubergen and Phil Hanawalt. Reginvestigated the effects of low doses of UV in inhi-biting cell division in E. coli B and the PR of the inhi-bition [40]. Bob used E. coli T−A−U− to follow, byradioautography, the distribution among progeny ofcells labeled with 3H thymine, arginine or uracil.The latter two were distributed randomly but thyminewas distributed asymmetrically, indicating that it waspresent in very large units [41]. To determine the auto-graphic exposure times, Bob used a simple procedureto estimate the radioactivity per cell. A few micro-liters of a labeled bacterial suspension was placedon a stainless steel planchette, dried, and the bacteria“fixed” to the planchette by a cytological technique.The planchette then was counted in a gas-flow win-dowless Geiger counter. The technique was laborious,but it was the only one possible since scintillationcounters were not available. It would be put to gooduse later in a critical experiment I carried out at OakRidge. Phil was interested in the effects of UV onmacromolecular synthesis in E. coli. He used the in-corporation of 32P, followed by separation into DNAand RNA components [42], to determine the effectsof different wavelengths on the synthesis of thesetwo polymers. At doses that stopped DNA synthesis,RNA synthesis and protein synthesis, measured bythe incorporation of 35S, continued at a linear rateuntil DNA synthesis resumed, at which point theirsyntheses increased exponentially [43].

I knew Ruth Hill when she was a graduate stu-dent in physics at Yale, before going to Columbia toobtain a Ph.D. degree in biophysics. She stayed on atColumbia and worked on ionizing radiation and UVeffects on viruses and bacteria and the PR of UV ef-fects. We kept in touch. She was mutagenizing cells,looking for radiation-resistant mutants of E. coli, whenshe came across a very sensitive mutant [44]. Whatwas more natural than to compare the effect of UVon DNA synthesis in this mutant, E. coli Bs−1, to thatin the wild type, B/r. Paul Swenson, a photobiologistfrom the University of Massachusetts at Amherst, hadjoined me for a sabbatical year at Oak Ridge. He andI began determining the effects of 265 nm on DNAsynthesis by measuring the incorporation of 3H-dThd,


using Bob’s planchette technique described above.The bacterial strains were not thymine auxotrophs, sowe used a trick, devised by Dick Boyce and me, toget 3H-thymidine incorporated [45]. In the resistantstrain, synthesis was inhibited for times that increasedwith dose, and then resumed exponentially. However,in the sensitive strain, synthesis stopped at much lowerdoses than for the wild type B/r and did not resume.The inhibitions were partially reversed in both strainsby exposure to PR light. Because we knew that dimersinhibited DNA synthesis we had to make sure thatequal numbers of dimers per unit dose were inducedin both strains. The doses were much too low for us toobserve any changes in DNA absorbance. Therefore,Bill Carrier adapted paper chromatography techniquesto separate labeled dimers from thymine in acid hy-drolysates of the irradiated cells. By this time FredBollum had a scintillation counter and water eluatesfrom the chromatograms could be counted in aqueousscintillation cocktails. The numbers of dimers werethe same in the two strains, and exposure to PR lightreduced the numbers by similar amounts [46]. Hence,we had no reasonable explanation for the differencein responses of the two strains. We speculated thatthe dimers in B/r might be in a form that did not per-manently inhibit synthesis. In attempting to detect achange in form, Bill and I took labeled cells exposedto ∼20 J/m2 and incubated them in growth mediumfor ∼60 min, the time it took for B/r to resume syn-thesis, and then exposed the cells to PR conditions.Dimers were reduced in the sensitive strain, but not inthe resistant one [47]. We had earlier shown that PRof dimers was much reduced in single stranded DNAcompared to double-stranded DNA [33]. Here wasevidence for a change in the form of dimer-containingDNA in B/r. The simplest separation we could thinkof was to compare the acid soluble to the acid insol-uble fraction of irradiated cells at ∼60 min after UV.Here, we found the answer to the conflicting findings.In sensitive cells the dimers were in the insolublefraction but in resistant cells the dimers were in thesoluble fraction. The dimers had been cut out of theresistant cells, the excised piece replaced by replica-tion using the good strand as a template, and DNAsynthesis resumed. We speculated that this repairmechanism was a general error-correcting one [47].

It is noteworthy that a year before our discovery,David Pettijohn, and his adviser Phil Hanawalt at

Stanford, had reported the aberrant incorporation ofBrdUrd into the DNA of a UV-exposed resistant strainof E. coli. The BrdUrd was located at the normal den-sity in a CsCl gradient rather than at a hybrid densitycharacteristic of semiconservative replication [48]. Ihad written to Phil in the summer of 1963 “Enclosedis a pre-print of some of the work we have doneon the effects of UV on DNA synthesis in bacteria.Some later results that I am just beginning to writeup for publication fit very well with those you haveobtained on bromouracil incorporation into bacterialDNA. These results indicate that thymine dimersin radiation-resistant cells are cut out of the DNAand appear in the acid soluble fraction of the cells,whereas in sensitive cells they are not cut out. Anobvious mechanism is that the lesion that is removedand perhaps the surrounding polynucleotide regionsare replaced by new bases from the medium. In thissense they act as if there had been turnover in DNA—the turnover being initiated by nucleases acting onthe UV lesion. If this is true, and our data seem to in-dicate this, then the bromouracil would be distributedrandomly along a single-strand and one wouldn’t ex-pect to find much melting of the heavy label.” (Philand I still keep in close touch with each other.) Thestimulated incorporation of label into parental DNAis “repair replication”. It also is called unscheduledDNA synthesis (UDS) when it is estimated radioau-tographically by the incorporation of 3H-dThd intocells during the non-S period of the cell cycle.

UV-resistant cells exhibit a phenomenon called“liquid holding recovery (LHR)” that is the recoveryof survival following UV exposure if the cells areheld in non-nutrient medium before plating on agar.Amleto Castellani, a visitor from Italy, John Jaggerand I showed that the recovery overlapped PR [49], in-dicating that LHR was associated with dimer excision.

Bill Carrier’s methods for measuring dimers insmall amounts [50] led to many other determinationsof repair. T4 phage, compared to the very similar T2phage, was resistant to UV inactivation [51] titered oneither resistant or sensitive strains of E. coli, indica-ting that a phage gene, called v, was responsible forthe resistance. We showed that in UV-irradiated T4,infecting E. coli Bs−1, dimers were rapidly removedfrom phage DNA [52]. The v gene codes for anendonuclease, T4endo, that now is used as a probefor dimers in irradiated DNA, and therapeutically as



a component of a skin lotion applied that reducessunlight-induced pre-malignant and malignant lesionsfrom the skin of UV sensitive individuals [53]. JohnBoyle, a post-doc from UK, used UV-exposed �phage infecting E. coli, to show that host cell reacti-vation depended on dimer excision from phage DNA,but if dimers also were induced in the host cells, theyinhibited the rate of removal of dimers from the phageDNA [54]. Betsy Sutherland, a graduate student atthe University of Tennessee in Knoxville, in her the-sis work in the Biology Division, demonstrated thatparamecia could do both excision and photorepair ofdimers [55,56]. James Regan (Jim) who had come toOak Ridge as a post-doc and stayed as a staff mem-ber, was an expert in the properties of mammaliancells in culture. He, together with another post-doc,Jim Trosko and Bill demonstrated that UV-irradiatednormal human fibroblasts could excise dimers fromtheir DNA [57].

Dimer excision was not the only explanation foran organism’s resistance to UV radiation. J. EdwardDonnellan was a radiation physicist in the BiologyDivision. He had been an undergraduate physicsmajor and a biophysics Ph.D. at Yale. He was veryinterested in bacterial spores because of their relativeresistance to all types of radiations. We irradiated3H-dThd-labeled spores of B. megaterium and an-alyzed them for dimers, following acid hydrolysis.There were none! Instead, there was a new product,spore photoproduct [58]. The properties of this prod-uct and the conditions that result in its formation weresummarized recently by Peter Setlow [59].

Surprisingly, in 1968 the mechanism of dimer for-mation and the structure of thymine dimers werequestioned [60]. DNA labeled in the methyl group ofthymidine with 3H and in the ring with 14C was irra-diated, hydroyzed, and chromatographed. The dimerswere eluted and counted in a scintillation counter.There were significantly more dimers labeled with14C than with 3H. Bill and I could not repeat theseresults [61]. We told the authors that they had madean error in the counting procedure. They withdrewtheir finding [62].

In 1968, James Cleaver showed that skin cellsfrom individuals with the genetic, skin cancer-pronedisease xeroderma pigmentosum (XP) were defectivein repair replication and UDS following exposure toUV [63]. The fact that repair of UV damage was

defective in these people suggested that unrepaireddamage to DNA could result in cancer. This resultlinked the initiation of carcinogenesis to mutagenesis.Fred de Serres, a biology staff member whose grouphad established an excellent in vivo assay for muta-gens that used Neurospora instilled into the peritonealcavity of mice, and I visited the NCI in an attempt tosolicit funding for assaying for potential carcinogens.We were told definitively, “there is no connectionbetween mutagenesis and carcinogenesis”. Within ayear, that point of view ceased to exist.

Jim Regan was very interested in XP and managedto obtain normal and XP cells from other laboratories.With them, we verified that an XP strain was defec-tive in dimer excision and because single-strand nicks(detectable by sedimentation in alkaline solution) didnot appear, that the defective step was the first endonu-clease step [64]. The excision assay was laborious andgave no information on the sizes of repaired regions.A better one was needed. Brainstorming among Jim,me and Ron Ley, a post-doc in the laboratory, led us todevelop a BrdUrd assay, that we called the BU-trick,that could estimate the numbers and sizes of repairedregions. Our idea was to let the repair of UV-exposed,3H-dThd-labeled cells take place in the presence of Br-dUrd and the repair of UV-exposed, 14C-dThd-labeledcells take place in the presence of dThd. The cells werethen mixed and exposed to large doses of 313 nm.Previous work had shown that such doses wouldresult in single-strand breaks in BrdUrd-substitutedDNA, but not in dThd-substituted DNA, detectableby alkaline sedimentation [39]. The rate of breakingof the BU regions with 313 nm dose gave the averagesize of the repaired regions, and the maximum num-ber of breaks gave the number of repaired regions per107 Da [65,66]. We used the BU-trick to determine therepair capabilities of a number of XP and normal cellstrains and found that all the XP strains had very lowrepair abilities, while the normal strains had a muchhigher, but broad, distribution of repair capabilities[67]. The method also was applied using a numberof carcinogenic chemicals. We found that they couldbe represented by either long patch (30 nucleotides)or short patch (five nucleotides) repair [68]. Ron Leyand I used the method on E. coli and showed that theaverage patch size was ∼12–20 nucleotides [69].

Rörsch et al. had shown that an extract from Micro-coccus lysodeicticus, now called M. luteus, reactivated


the replicative form of UV inactivated �X174 DNA[70]. Bill showed that the extract could excise dimersfrom UV-irradiated DNA [71], and he purified andcharacterized an endonuclease from it that nickedUV-irradiated DNA [72]. The extract could reacti-vate UV-exposed transforming DNA [73], and theendonuclease is now used to estimate the dimer fre-quency in UV-exposed cells [74]. An energetic visitorto the Biology Division was Ronald Hart (Ron) fromOhio State University. He collected fibroblast strainsfrom a large number of mammalian species with awide spread of lifespans. After exposing them to UV,we found that UDS levels following UV exposureincreased with lifespan from mice to humans, a rangeof ∼50-fold [75]. Some investigators could not re-peat this result, but a recent reanalysis shows thatthe average of several different experiments supportsthe original finding [76]. Ron and I also showed thatDNA repair (unscheduled synthesis) declined as hu-man cells age in culture. However, there also wasa decline in scheduled synthesis. We interpreted theresults as indicating that the failure of repair was nota causal event in the failure of cells to divide, butthat as cells age, the ability to carry out the manycoordinated steps in repair declines [77].

Through the efforts of Alexander Hollaender, theUniversity of Tennessee established a Graduate Schoolof Biomedical Sciences at Oak Ridge. The School hadseveral exceptional students, one of whom was Mal-colm (Mac) Paterson. He collaborated with John Boyleand me in investigating repair in cells deficient in DNApolymerase [78,79], and then went on to develop aquantitative system for measuring UV-induced DNAdamage, using the M. luteus endonuclease system [72].

After I moved from Oak Ridge to Brookhavenin 1974, I was fortunate in having many ingeniouspost-docs and visitors from other institutions workingwith me. They introduced me to a wealth of other cel-lular systems for exploring DNA repair following UVexposure. Steven D’Ambrosio, now at Ohio State Uni-versity, showed that post-replication repair, followingthe UV exposure of Chinese hamster cells and nor-mal human and XP cells, was enhanced by a smallerpre-exposure to UV [80,81]. Farid Ahmed demon-strated that although normal human cells repairedboth UV and bulky chemical damages to their DNA,the rate-limiting steps and the kinetics of repair forchemicals differed from those of UV repair [82]. Jim

Boyd, of the University of California at Davis, was anexpert on the genetics of Drosophila melanogaster. Heshowed that a mutant with reduced meiotic recombina-tion in females was hypersensitive to killing by a num-ber of chemical mutagens and that their cells in culturewere defective in excising dimers after exposure to UV[83]. Bob Rothman ascertained that the action spectrafor killing Chinese hamster cells, and for producingdimers in them were very similar [84]. He also used theBU-trick to show that repair patches in a UV-irradiatedrecL mutant of E. coli were very, very long, ∼350 nu-cleotides [85]. Studies by Lorne Taichman of the StateUniversity of New York at Stony Brook revealed thatthe rates of dimer repair were similar in human ker-atinocytes and fibroblasts [86]. Barry Rosenstein, nowat Mt. Sinai School of Medicine in New York, workedon the effects of UV on cultured frog cells findingthat the PR-sector was independent of the UV wave-length from 252–313 nm, and that unrepaired dimersacted as long term blocks to DNA synthesis [87,88].He and Dan Yarosh, now at AGI Dermatics, usedthe BU-trick to show that the repair patch size in T4phage was small, approximately four nucleotides [89].George Kantor, of Wayne State University, carriedout elegant, precise studies on non-dividing normalhuman and XP cells, demonstrating the similarities inthe shapes of the action spectra for inactivation [90].He also found that, as expected, the dose to inactivateboth types of cells by 254 nm was much less (approx-imately three-fold) than that by sunlamp exposure,and that the numbers of dimers per inactivating dosewere independent of the light source. The dimer lev-els for the two cell types also differed by the factorof 3 [91]. George also measured repair rates, by fol-lowing the disappearance of endonuclease-sensitivesites over a range of doses as low as 1 J/m2 of 254 nm(approximately three dimers per 108 Da). Repair innon-dividing cells was biphasic, with a rapid reactionremoving ∼70% of the sites in 1 day, followed by aslow reaction that could take as many as 20 days [92].

Jack Lipman, a graduate student at Stony Brook,was working on repair in articular chondrocytes fromhumans and rabbits. His experiments revealed thatthe DNA repair rate, measured by UDS, was about2.5-fold greater for human cells and obtained a sim-ilar finding for the removal of dimers, measured byendonuclease-sensitive sites using either centrifuga-tion or alkaline elution [93,94]. Helene Hill, of the



New Jersey Medical School, introduced me to thephotobiology of mouse melanoma cells. By irradiat-ing together melanotic and non-melanotic cells withdifferent radioactive isotopes, we showed that moredimers (endonuclease-sensitive sites) were formedin the non-melanotic cells at wavelengths <289 nmbecause of their higher UV-transmission at thesewavelengths [95].

Akihiro (Aki) Shima, of the University of Tokyo,had used fish cells in culture, because they have anactive PR system, to mimic what might be the situa-tion in the real world of sunlight because it includesdamaging UV wavelengths and reactivating PRlight. He exposed cells separately or concurrentlyto sunlamp UV (>280 nm) and to daylight radiation(>350 nm). The concurrent exposure increased sur-vival compared to the sunlamp alone [96]. A compar-ison of cell killing by sunlamp and 254 nm exposuregave results similar to those obtained for human cellsin culture [91]. The fractions of cells inactivated perendonuclease-sensitive site were about the same forthe two light sources. I am pleased that this interac-tion with Aki, begun to investigate UV effects, hascontinued and led us to collaborate on assessing theeffects of high energy cosmic ray nuclei, using fishas a model for the induction of human germ cellmutations [97]. The presence of PR activity in fishcells, but not in human cells in culture, results in verydifferent quantitative responses to sunlamp exposure[98]. The number of dimers per unit dose of sunlampexposure >304 nm to fish cells is only ∼one-sixthof those in human cells, although the numbers perunit dose are the same for 254 nm exposure, becauseof PR by the sunlamp’s longer wavelengths; more-over, fish cells inactivated by 254 nm are extensivelyreactivated by exposure to them.

5. The effects of UV on humans and fish

Stanfield Rogers at Oak Ridge had shown thatUV-exposed embryonic mouse lung cells implantedinto syngeneic mice could develop into adenomas[99]. I wanted to do a similar experiment with cellsthat contained PR activity (mice do not have any) soas to determine whether PR would reduce the tumori-genic potential of the UV-exposed cells. We discussedthe possibility of using small marsupials because

they do have PR [100], but no animal colonies wereavailable at that time. Jim Regan suggested that wemight try a fish, Poecilia formosa, the Amazon molly,because they grow in clones and were known to havePR activity [100]. There are no males in the species.The oocytes are diploid and are activated to divide bysperm from a male of a correlative species. Our ideawas to obtain tissue from several fish, homogenizeit to separate the cells, UV-irradiate them, and thenexpose one half of them to PR light. We would injectthe UV cells intraperitoneally into a group of isogenicrecipients and the UV + PR cells into another groupof recipients. Ron Hart and I began this experiment inOak Ridge and continued it when I moved, in 1974,to the Biology Department of the Brookhaven Na-tional Laboratory. We were fortunate in having AvrilWoodhead, an expert fish biologist and histologist,join us to continue the experiment. It worked. Inject-ing UV-exposed cells resulted in tumors in ∼100%of recipient fish, whereas injecting cells exposed toUV+PR yielded only ∼5% of recipients with tumors[101]. Missing from this fish story was the demonstra-tion that PR actually split the UV induced dimers inthe fish cells. Several years later, Philip Achey fromthe University of Florida, was on a sabbatical leave atBrookhaven and helped to do an experiment to verifythat the dimers were split [102]. He used the M. luteusUV-endonuclease to introduce nicks into the unla-beled DNA of UV-exposed fish cells and estimatedthe decrease in the molecular weight of single-strandsnot by sedimentation in alkaline gradients, but byelectrophoresis in alkaline agarose gels, a techniquedeveloped at Brookhaven [103]. This much simplertechnique for measuring DNA damages has replacedthe sedimentation technique and was upgraded to givequantitative results [104].

In 1972, I accepted an invitation to join the NationalResearch Council’s “Climatic Impact Committee”concerned with the environmental hazards of super-sonic aircraft. It was thought that water from theexhaust of such craft, flying in the stratosphere, wouldcatalyze the destruction of ozone. Because ozone ab-sorbs UV wavelengths <320 nm, it would increasethe transmission of UV in the range of 290–320 nmthat might affect the growth of plants and aquatic or-ganisms and also result in an increase in human skincancer. The epidemiological data were convincing thatthe incidence and prevalence of skin cancer increased


with proximity to the equator and, because the yearlyUV flux also increased with proximity to the equator,that skin cancer was engendered by exposure to UV.The latter conclusion is logically weak because theflux of all wavelengths increases with proximity to theequator, as does temperature. Data on non-melanomaskin cancer in mice implicated the UVB wavelengths(280–320 nm) as the most important [105], but unlessthe action spectrum was known, the effect of ozonedepletion could not be estimated quantitatively. At thetime, because XP individuals were defective in DNArepair and were prone to skin cancer, the only appro-priate spectrum seemed to be an action spectrum foraffecting DNA. Hence, I summarized all the availableaction spectra data on photoproducts in DNA, virusesand E. coli and constructed a spectrum, as theoreticallyappropriate to use for the induction of skin cancer, if itwere multiplied by the transmission of skin as a func-tion of wavelength [106]. Unfortunately, many peoplewho used this spectrum, calling it “Setlow DNA”,forgot about the correction for transmission. The com-mittee used that spectrum along with the sunlight fluxas a function of wavelength and latitude to estimatethe “skin cancer-inducing UV-dose” as a functionof latitude. Those data, along with the epidemiolog-ical data on skin cancer incidence/prevalence versuslatitude, yielded the dose-response relation for skincancer incidence/prevalence. A hypothetical decreasein ozone would result in an increase in UV-dose and,hence, a predicted increase in skin cancer [107,108].This was my introduction to skin cancer and the dif-ficulties in obtaining and using epidemiological data.Although it turned out that the water-vapor injectedinto the stratosphere by aircraft was unimportant com-pared to the chlorofluorocarbons arising from usingfreons, the calculations were still valid.

There was no animal model for malignant mela-noma, but the National Research Council Committeefelt it reasonable to use the same type of estimation aswas used above for non-melanoma skin cancer, eventhough mortality as a function of latitude was lesssteep than that for non-melanoma incidence. Subse-quent committees could not really solve the problemof estimating the possible changes in melanoma in-cidence. Melanomas arise from melanocytes, cellsnormally containing melanin. Since melanin absorbsat all wavelengths, and seems to be a photoactivepigment, I thought it possible that melanin could be

a photosensitizer that might lead to a reaction withDNA at wavelengths longer than UVB. Avril Wood-head, Eleanor Grist and I tried to obtain a fish modelfor UV-induced melanoma using the Amazon molly.We were not successful. However, Avril knew thatsome interspecies crosses in the genus Xiphophorussometimes developed melanoma spontaneously. Ge-netic evidence indicated that X. maculatus (platyfish)had both melanoma genes and melanoma suppressorgenes and rarely developed melanomas, whereas X.helleri (swordtail) had neither melanoma nor sup-pressor genes. We irradiated progeny of crosses andbackcrosses, of different ages, with a range of sun-lamp exposures, and finally found a backcross hybridof platyfish ( ) and swordtail ( ) that did developmelanomas after UV exposure [109]. The inductionof these melanomas could be reversed, in part, byPR. Backcross progeny using X. couchianus ( ) wereeasier to breed, and from 5–7-day-old exposed fishwe determined an action spectrum for melanoma in-duction [110]. The most effective wavelength wasat 302 nm, but there was appreciable sensitivity at365, 405 and 436 nm. We estimated that if the humanspectrum for melanoma induction were like the fishspectrum, 90% of melanoma would arise from wave-lengths >UVB, whereas 95% of non-melanoma skincancer would arise from UVB. Appreciable numbersof dimers were made in fish skin by UVB, and theywere photoreactivable, whereas no dimers were madeby 365 nm [111]. Other findings and quantitative epi-demiological arguments support the conclusion fromaction spectra that melanoma arises from wavelengths>UVB [112–114]. I hypothesize (unpublished) that anew, as yet unidentified photoproduct is responsiblefor melanoma and, because of the high incidence ofmelanoma among XP patients, it is repaired in normalcells, but not in XP cells, by nucleotide excision repair.

6. Further reflections

Doing Science has been an exhilarating experience.I have worked in many areas of science other thanthe effects of light on biological systems. Thus, Iregret that I have not explicitly acknowledged all ofthe many students, collaborators, colleagues, corre-spondents, committee members and friends from fivecontinents that have helped and influenced me, and



are part of my total experience. I have not forgotten“you-all”. Nor have I forgotten that those who arementioned in these reflections have done much morescience than I have indicated.

My 14 years in the Biology Division of the OakRidge National Laboratory were times of great sci-entific pleasure, excitement and accomplishments. Iowe a debt of gratitude to the late Alexander Hol-laender for creating such a splendid division, forinsulating me from upper management and also forsecuring funds to support all of us with a minimumof grant/report writing compared to today. He was aremarkable person [115–117].

Most of my research at Yale, Oak Ridge andBrookhaven was supported by the US Departmentof Energy (DOE) and its predecessors, the AtomicEnergy Commission (AEC) and the Energy Researchand Development Agency. I have had additional, andinvaluable funds from The American Cancer Soci-ety, The National Cancer Institute, the EnvironmentalProtection Agency and the National Aeronautics andSpace Administration. It was remarkable that theAEC and its successors had the foresight to fund thebasic research needed to obtain much of our presentinformation about DNA repair. For example, the workon UV repair stimulated the analysis of repair follow-ing X-ray damage to DNA in cells. Dick McGrath, inthe biophysics group in Oak Ridge, found that if helysed E. coli cells on top of an alkaline sucrose gra-dient, he could observe large, single-strands of DNA(∼2 × 108 Da, ∼1/15 of the double-strand chromoso-mal size). Hence, he could observe the breaks after anX-ray exposure of 2 Gy and show in 1966 that theywere readily repaired in strain B/r but not in Bs−1[118]. The quantitative aspects of the sedimentationanalysis depended on the calibrations made by BillStudier [119]. Bill had been an undergraduate bio-physicist (Yale, 1958) and had recently (1964) cometo the Biology Department of Brookhaven.It is a pity that the DOE has changed its scientific

priorities. Jim Cleaver, writing briefly about me, put itwell [120]. “We shared much, at a distance, and evenhit a minor roadblock together, in 1995–1996. Thatwas when the DOE reduced its budget, cut a swathacross the DNA repair field in numerous of its labo-ratories, citing what they considered to be a declinein productivity. At the same time they presented ourwork in glowing terms on facing pages of their record

of achievement to Congress.” I hope that DOE willreverse their thinking/priorities again.

Acknowledgements

I thank Kathryn Folkers, an Administrative Assis-tant in the Biology Department, for her invaluablehelp in attending, for many years, to most of my cor-respondence and manuscripts. Avril Woodhead helpededit this manuscript. Brookhaven National Laboratoryis operated by Brookhaven Science Associates, undercontract with the US DOE.

References

[1] R.B. Setlow, High energy states of N2+ and N2, Phys. Rev.

74 (1948) 153–162.[2] D. J Fluke, R.B. Setlow, Water-prism ultraviolet monochro-

mators, J. Opt. Soc. Am. 44 (1954) 327–330.[3] R.B. Setlow, W.R. Guild, The spectrum of the peptide

bond and other substances below 230 �m, Arch. Biochem.Biophys. 34 (1951) 223–225.

[4] R.B. Setlow, The inactivation of catalase by deuterons andheat, Arch. Biochem. Biophys. 34 (1951) 396–408.

[5] R.B. Setlow, The inactivation of urease by deuterons andheat, Arch. Biochem. Biophys. 36 (1952) 328–335.

[6] R.B Setlow, E.C. Pollard, Molecular Biophysics, AddisonWeseley, Reading, PA, 1962.

[7] A.D. McLaren, Photochemistry of enzymes, J. Polymer Sci.2 (1947) 107.

[8] R.B. Setlow, B. Doyle, The quantum yield for theinactivation of dry �-chymotrypsin hydrochloride, Arch.Biochem. Biophys. 46 (1953) 39–45.

[9] R.B. Setlow, A relation between cystine content and ultra-violet sensitivity of proteins, Biochim. Biophys. Acta 16(1955) 444–445.

[10] R.B. Setlow, B. Doyle, The action of monochromatic ultra-violet light on proteins, Biochim. Biophys. Acta 24 (1957)27–41.

[11] J.W. Preiss, R.B. Setlow, Spectra of some amino acids,peptides, nucleic acids and protein in the vacuum ultraviolet,J. Chem. Phys. 25 (1956) 138–141.

[12] R.B. Setlow, G. Watts, C. Douglas, Inactivation of proteinsby vacuum ultraviolet radiation, in: Proceedings of the firstNational Biophysics Conference, Yale University Press, NewHaven, pp. 174–183.

[13] R.B. Setlow, Ultraviolet wavelength-effects on proteins andnucleic acids, Rad. Res. Suppl. 2 (1960) 276–289.

[14] F.L. Gates, A study of the bactericidal action of ultravioletlight. Part I. The reaction to monochromatic radiations, J.Gen. Physiol. 13 (1929) 231–248.


[15] A. Hollaender, C.W. Emmons, Wavelength dependence ofmutation production in the ultraviolet with special emphasison fungi , Cold Spring Harbor Symp. Quant. Biol. 9 (1941)179–186.

[16] R.B. Setlow, B. Doyle, The action of radiation on drydesoxyribonucleic acid, Biochim. Biophys. Acta 15 (1954)117–125.

[17] R.B. Setlow, B. Doyle, The action of ultraviolet light onparamecin and the nature of paramecin, Biochim. Biophys.Acta 22 (1956) 15–20.

[18] R.B. Setlow, R. Boyce, The ultraviolet inactivation of �X174bacteriophage at different wavelengths and pHs, Biophys. J.1 (1960) 29–41.

[19] R.B. Setlow, The use of action spectra to determine thephysical state of DNA in vivo, Biochim. Biophys. Acta 39(1960) 180–181.

[20] J.K. Setlow, R.B. Setlow, Ultraviolet action spectra ofordered and disordered DNA, Proc. Natl. Acad. Sci. U.S.A.47 (1961) 1619–1627.

[21] J.K. Setlow, R.B. Setlow, Evidence for the existence of asingle-stranded stage of T2 bacteriophage during replication,Proc. Natl. Acad. Sci. U.S.A. 46 (1960) 791–798.

[22] F. Fawaz-Estrup, R.B. Setlow, Inactivation of solubleribonucleic acid by ultraviolet irradiation, Biochim. Biophys.Acta 87 (1964) 28–33.

[23] R. Latarjet, B. Muel, D.A. Haig, M.C. Clarke, T. Alper,Inactivation of the scrapie agent by near monochromaticultraviolet light, Nature 227 (1970) 1341–1343.

[24] S.B. Prusiner, Molecular biology of prion diseases, Science252 (1991) 1515–1522.

[25] R. Beukers, W. Berends, Isolation and identification of theirradiation product of thymine, Biochim. Biophys. Acta 41(1960) 550–551.

[26] J. Jagger, Photoreactivation, Bacteriol. Rev. 22 (1958)99–142.

[27] R.B. Setlow, The action spectrum for the reversal of the di-merization of thymine induced by ultraviolet light, Biochim.Biophys. Acta 49 (1961) 237–238.

[28] R.A. Deering, R.B. Setlow, Effects of ultraviolet lighton thymidine dinucleotide and polynucleotide, Biochim.Biophys. Acta 68 (1963) 526–534.

[29] R.B. Setlow, W.L. Carrier, The identification of ultraviolet-induced thymine dimers in DNA by absorbance measure-ments, Photochem. Photobiol. 2 (1963) 49–57.

[30] R.B. Setlow, J.K. Setlow, Evidence that ultraviolet-inducedthymine dimers in DNA cause biological damage, Proc.Natl. Acad. Sci. U.S.A. 48 (1962) 1250–1257.

[31] J.K. Setlow, R.B. Setlow, Nature of the photoreactivableultraviolet lesion in DNA, Nature 197 (1963) 560–562.

[32] F.J. Bollum, R.B. Setlow, Ultraviolet inactivation of deoxy-ribonucleic acid primer activity. Part I. Effects of differentwavelengths and doses, Biochim. Biophys. Acta 68 (1963)599–607.

[33] R.B. Setlow, F.J. Bollum, W.L. Carrier, Nuclease-resistantsequences in ultraviolet-irradiated deoxyribonucleic acid,Biochim. Biophys. Acta 91 (1964) 446–461.

[34] R.B. Setlow, W.L. Carrier, F.J. Bollum, Pyrimidine dimersin UV-irradiated poly dI:dC, Proc. Natl Acad. Sci. U.S.A.53 (1965) 1111–1118.

[35] R.B. Setlow, W.L. Carrier, Pyrimidine dimers in ultraviolet-irradiated DNA’s, J. Mol. Biol. 17 (1966) 237–254.

[36] R.B. Setlow, Cyclobutane-type pyrimindine dimers inpolynucleotides, Science 153 (1966) 379–386.

[37] R. Boyce, R.B. Setlow, The action spectra of ultraviolet-lightinactivation of systems containing 5-bromouracil substituteddeoxyribonucleic acid. Part I. Escherichia coli 15T−A−U−,Biochim. Biophys. Acta 68 (1963) 446–454.

[38] R.B. Setlow, R. Boyce, The action spectra of ultraviolet-lightinactivation of systems containing 5-bromouracil substituteddeoxyribonucleic acid. Part II. Bacteriophage T4, Biochim.Biophys. Acta 68 (1963) 455–461.

[39] M. Lion, Search for a mechanism for the increasedsensitivity of 5-bromouracil-substituted DNA to ultravioletradiation. Part II. Single-strand breaks in the DNA ofirradiated 5-bromouracil substituted T3 coliphage, Biochim.Biophys. Acta 209 (1970) 24–33.

[40] R.A. Deering, R.B. Setlow, Inhibition of cell division ofEscherichia coli by low doses of ultraviolet light, Science126 (1957) 397–398.

[41] R.P. van Tubergen, R.B. Setlow, Quantitative radioauto-graphic studies on growing cultures of Escherichia coli: thedistribution of parental DNA, RNA, protein, and cell wallamong progeny cells, Biophys. J. 1 (1961) 589–625.

[42] P. Hanawalt, Use of phosphorus-32 in microassay fornucleic acid synthesis in Escherichia coli, Science 130(1959) 386–387.

[43] P. Hanawalt, R. Setlow, Effect of monochromatic ultravioletlight on macromolecular synthesis in Escherichia coli,Biochim. Biophys. Acta 41 (1960) 283–294.

[44] R.F. Hill, A radiation-sensitive mutant of Escherichia coli,Biochim. Biophys. Acta 30 (1958) 636–637.

[45] R.P. Boyce, R.B. Setlow, A simple method of increasingthe incorporation of thymidine into the deoxyribonucleicacid of Escherichia coli, Biochim. Biophys. Acta 61 (1962)618–620.

[46] R.B. Setlow, P.A. Swenson, W.L. Carrier, Thymine dimersand inhibition of DNA synthesis by ultraviolet irradiationof cells, Science 142 (1963) 1464–1466.

[47] R.B. Setlow, W.L. Carrier, The disappearance of thyminedimers from DNA: an error-correcting mechanism, Proc.Natl. Acad. Sci. U.S.A. 51 (1964) 226–231.

[48] D.E. Pettijohn, P.C. Hanawalt, Deoxribonucleic acid repli-cation in bacteria following ultraviolet radiation, Biochim.Biophys. Acta 72 (1963) 127–129.

[49] A. Castellani, J. Jagger, R.B. Setlow, Overlap of photoreac-tivation and liquid holding recovery in Escherichia coli B,Science 143 (1964) 1170–1171.

[50] W.L. Carrier, R.B. Setlow, The excision of pyrimidine dimers(the detection of dimers in small amounts), in: L. Grossman,K. Moldave (Eds.), Methods in Enzymology, Vol. XXI,Part D, Academic Press, New York, 1971, pp. 203–207.

[51] W. Harm, Gene-controlled reactivation of ultraviolet inacti-vated bacteriophage, J. Cell Comp. Physiol. 58 (Suppl. 1)(1961) 69–77.



[52] R.B. Setlow, W.L.Carrier, The excision of dimers in vivo andin vitro, in: W.J. Peacock, R.D. Brock (Eds.), Replication andRecombination of Genetic Material, Australian Academyof Science, Canberra, 1968, pp. 134–141.

[53] D. Yarosh, J. Klein, A. O’Conner, J. Hawk, E. Rafel, P.Wolf, et al., Effect of topically applied T4 endonucleaseV in liposomes on skin cancer in xeroderma pigmentosum,Lancet 357 (2001) 926–929.

[54] J.M. Boyle, R.B. Setlow, Correlations between host cellreactivation, UV reactivation and pyrimidine dimer excisionin the DNA of bacteriophage, J. Mol. Biol. 51 (1970)131–144.

[55] B.M. Sutherland, W.L. Carrier, R.B. Setlow, Photoreacti-vation in vivo of pyrimidine dimers in Paramecium DNA,Science 158 (1967) 1699–1700.

[56] B.M. Sutherland, W.L. Carrier, R.B. Setlow, Pyrimidinedimers in the DNA of Paramecium aurelia, Biophys. J. 8(1968) 490–499.

[57] J.D. Regan, J.E. Trosko, W.L. Carrier, Evidence for excisionof ultraviolet induced pyrimidine dimers from the DNA ofhuman cells in vitro, Biophys. J. 8 (1968) 319–325.

[58] J.E. Donnellan Jr., R.B. Setlow, Thymine photoproductsbut not thymine dimers found in ultraviolet irradiatedbacterial spores, Science 149 (1965) 308–310.

[59] P. Setlow, Resistance of spores of Bacillus species toultraviolet light, Environ. Mol. Mutagen. 38 (2001) 92–104.

[60] S.Y. Wang, M.H. Patrick, A.J. Varghese, C.S. Rupert,Concerning the mechanism of formation of UV-inducedthymine photoproducts in DNA, Proc. Natl. Acad. Sci.U.S.A. 57 (1967) 465–472.

[61] R.B. Setlow, Photoproducts in DNA irradiated in vivo,Photochem. Photobiol. 7 (1968) 643–649.

[62] S.Y. Wang, M.H. Patrick, A.J. Varghese, C.S. Rupert,Addendum: concerning the mechanism of formation ofUV-induced photoproducts in DNA, Proc. Natl. Acad. Sci.U.S.A. 58 (1967) 2483.

[63] J.E. Cleaver, Defective repair replication in xerodermapigmentosum, Nature 218 (1968) 652–656.

[64] R.B. Setlow, J.D. Regan, J. German, W.L. Carrier, Evidencethat xeroderma pigmentosum cells do not perform the firststep in the repair of ultraviolet damage to their DNA, Proc.Natl. Acad. Sci. U.S.A. 64 (1969) 1035–1041.

[65] J.D. Regan, R.B. Setlow, R.D. Ley, Normal and defectiverepair of damaged DNA in human cells: a sensitive assayutilizing the photolysis of bromodeoxyuridine, Proc. Natl.Acad. Sci U.S.A. 68 (1971) 708–712.

[66] R.B. Setlow, J.D. Regan, Measurement of repair synthesis byphotolysis of bromouracil, in: E.C. Friedberg, P.C. Hanawalt(Eds.), DNA Repair: A Laboratory Manual of ResearchProcedures, Marcel Dekker, New York, 1981, pp. 307–318.

[67] R.B. Setlow, Variations in DNA repair among humans, in:C.C. Harris, H.N. Autrup (Eds.), Human Carcinogenesis,Academic Press, New York, 1983, pp. 231–254.

[68] J.D. Regan, R.B. Setlow, Two forms of repair in the DNAof human cells damaged by chemical carcinogens andmutagens, Cancer Res. 34 (1974) 3318–3325.

[69] R.D. Ley, R.B. Setlow, Repair replication in Escherichiacoli as measured by the photolysis of bromodeoxyuridine,Biophys. J. 12 (1972) 420–431.

[70] A. Rörsch, C.v.d. Kamp, J. Adema, Dark reactivationof ultraviolet irradiated deoxyribonucleic acid in vitro,Biochim. Biophys. Acta 80 (1964) 346–348.

[71] W.L. Carrier, R.B. Setlow, Excision of pyrimidine dimersfrom irradiated deoxyribonucleic acid in vitro, Biochim.Biophys. Acta 129 (1966) 318–325.

[72] W.L. Carrier, R.B. Setlow, Endonuclease from Micrococcusluteus which has activity toward ultraviolet-irradiateddeoxyribonucleic acid: purification and properties, J.Bacteriol. 102 (1970) 176–186.

[73] R.B. Setlow, J.K. Setlow, W.L. Carrier, Endonucleasefrom Micrococcus luteus which has activity towardultraviolet-irradiated deoxyribonucleic acid: its action ontransforming deoxyribonucleic acid, J. Bacteriol. 10 (1970)187–192.

[74] M.C. Paterson, P.H.M. Lohman, M.L. Sluyter, Use of aUV-endonuclease from Micrococcus luteus to monitor theprogress of DNA repair in UV-irradiated human cells,Mutat. Res. 19 (1973) 245–256.

[75] R.W. Hart, R.B. Setlow, Correlation between deoxyribo-nucleic acid excision-repair and life span in a numberof mammalian species, Proc. Natl. Acad. Sci. U.S.A. 71(1974) 2169–2173.

[76] G.A. Cortopassi, E. Wang, There is substantial agreementamong interspecies of DNA repair activities, Mech. AgeingDev. 91 (1996) 211–218.

[77] R.W. Hart, R.B. Setlow, DNA repair in late-passage humancells, Mech. Ageing Dev. 5 (1976) 67–77.

[78] J.M. Boyle, M.C. Paterson, R.B. Setlow, Excision repairproperties of an Escherichia coli mutant deficient in DNApolymerase, Nature 226 (1970) 708–710.

[79] M.C. Paterson, J.M. Boyle, R.B. Setlow, Ultraviolet andX-ray-induced responses of a deoxyribonucleic acid poly-merase-deficient mutant of Escherichia coli, J. Bacteriol.107 (1971) 61–67.

[80] S.M. D’Ambrosio, R.B. Setlow, Enhancement of post-repli-cation repair in Chinese hamster cells, Proc. Natl. Acad.Sci. U.S.A. 73 (1976) 2396–2400.

[81] S.M. D’Ambrosio, R.B. Setlow, Defective and enhancedpost-replication repair in classical and variant xerodermapigmentosum cells treated with N-acetoxy-2-acetylamino-fluorene, Cancer Res. 38 (1978) 1147–1153.

[82] F.E. Ahmed, R.B. Setlow, Kinetics of repair in ultravio-let-irradiated and N-acetoxy-2-acetylaminofluorene-treatedmammalian cells, Biophys. J. 24 (1978) 665–675.

[83] J.B. Boyd, M.D. Golino, R.B. Setlow, The mei-9a mutantof Drosophila melanogaster increases mutagen sensitivityand decreases excision repair, Genetics 84 (1976) 527–544.

[84] R.H. Rothman, R.B. Setlow, An action spectrum for cellkilling and pyrimidine dimer formation in Chinese hamsterV-79 cells, Photochem. Photobiol. 29 (1979) 57–61.

[85] R. Rothman, Dimer excision and repair replication patchsize in a recL152 mutant of Escherichia coli K-12, J.Bacteriol. 135 (1978) 444–448.


[86] L.B. Taichman, R.B. Setlow, Repair of ultraviolet lightdamage to the DNA of cultured human epidermal keratino-cytes and fibroblasts, J. Invest. Dermatol. 73 (1979) 217–219.

[87] B.S. Rosenstein, R.B. Setlow, Photoreactivation of ICR2A frog cells after exposure to monochromatic ultravioletradiation in the 252–313 nm range, Photochem. Photobiol.32 (1980) 361–366.

[88] B.S. Rosenstein, R.B. Setlow, DNA repair after ultravioletirradiation of frog ICR 2A frog cells: pyrimidine dimersare long acting blocks to nascent DNA synthesis, Biophys.J. 31 (1980) 195–205.

[89] D.B. Yarosh, B.S. Rosenstein, R.B. Setlow, Excision repairand patch size in UV-irradiated bacteriophage T4, J. Virol.49 (1981) 465–471.

[90] G.J. Kantor, J.C. Sutherland, R.B. Setlow, Action spectrumfor killing non-dividing normal human and xeroderma pig-mentosum cells, Photochem. Photobiol. 31 (1980) 459–464.

[91] G.J. Kantor, R.B. Setlow, Correlation between inactivationof human cells and numbers of pyrimidine dimers inducedby a sun lamp and 254 nm radiation, Photochem. Photobiol.35 (1982) 269–274.

[92] G.J. Kantor, R.B. Setlow, Rate and extent of DNA repairin non-dividing human diploid fibroblasts, Cancer Res. 41(1981) 819–825.

[93] R.B. Setlow, J.M. Lipman, L. Sokoloff, DNA repair byarticular chondrocytes. Part II. Direct measurements ofrepair of ultraviolet and X-ray damage in monolayercultures, Mech. Ageing Dev. 21 (1983) 97–103.

[94] J.M. Lipman, R.B. Setlow, DNA repair by articularchondrocytes. Part IV. Measurement of Micrococcus luteusendonuclease-sensitive sites by alkaline elution in rabbitarticular chondrocytes, Mech. Ageing Dev. 40 (1987)193–204.

[95] H.Z. Hill, R.B. Setlow, Comparative action spectra forpyrimidine dimer formation in Cloudman S91 mousemelanoma and EMT6 mouse mammary carcinoma cells,Photochem. Photobiol. 35 (1982) 681–684.

[96] A. Shima, R.B. Setlow, Survival and pyrimidine dimersin cultured fish cells exposed to concurrent sunlamp andphotoreactivating radiations, Photochem. Photobiol. 39(1984) 49–56.

[97] A. Shima, A. Shimada, Development of a possiblenon-mammalian test system for radiation-induced germ-cellmutagenesis using a fish, the Japanese medaka (Oryziaslatipes), Proc. Natl. Acad. Sci. U.S.A. 88 (1991) 2545–2549.

[98] F.E. Ahmed, R.B. Setlow, E. Grist, N. Setlow, DNA damage,photorepair, and survival in fish and human cells exposedto UV radiation, Environ. Mol. Mutagen. 22 (1993) 18–25.

[99] S. Rogers, The in vitro initiation of pulmonary adenomasin fetal mouse lung by a single exposure to ultravioletradiation of wavelength 2537 Å, J. Natl. Cancer Inst.15(1955) 1001–1004.

[100] C.S. Rupert, Enzymatic Photoreactivation: Overview, in: P.CHanawalt, R.B. Setlow (Eds.), Molecular Mechanisms forRepair of DNA, Plenum Press, New York, 1975, pp. 73–87.

[101] R.W. Hart, R.B. Setlow, A.D. Woodhead, Evidence thatpyrimidine dimers in DNA can give rise to tumors, Proc.Natl. Acad. Sci. U.S.A. 74 (1977) 5574–5578.

[102] P.M. Achey, A.D. Woodhead, R.B. Setlow, Photoreactivationof pyrimidine dimers in DNA from thyroid cells of theteleost, Poecilia formosa, Photochem. Photobiol. 29 (1979)305–310.

[103] M.W. McDonell, M.N. Simon, F.W. Studier, Analysis ofrestriction fragments of T7 DNA and determination ofmolecular weights by electrophoresis in neutral and alkalineagarose gels, J. Mol. Biol. 110 (1977) 119–146.

[104] S.E. Freeman, A.D. Blackett, D.C. Monteleone, R.B. Setlow,B.M. Sutherland, J.C. Sutherland, Quantitation of radiation-,chemical-, or enzyme-induced single-strand breaks in non-radioactive DNA by alkaline gel electrophoresis: applicationto pyrimidine dimers, Anal. Biochem. 158 (1986) 119–129.

[105] H.F. Blum, Carcinogenesis by Ultraviolet Light, PrincetonUniversity Press, Princeton, NJ, 1959.

[106] R.B. Setlow, The wavelengths in sunlight effective inproducing skin cancer: a theoretical analysis, Proc. Natl.Acad. Sci. U.S.A. 71 (1974) 3363–3366.

[107] Environmental Impact of Stratospheric Flight, NationalAcademy of Science, Washington, 1975.

[108] E.L. Scott, M.L. Straf, Ultraviolet radiation as a cause ofcancer, in: H.H. Hiatt, J.D. Watson, J.A. Winsten (Eds.),Origins of Human Cancer, Cold Spring Harbor Laboratory,Cold Spring Harbor, 1977, pp. 529–546.

[109] R.B. Setlow, A.D. Woodhead, E. Grist, Animal model forultraviolet radiation-induced melanoma: platyfish-swordtailhybrid, Proc. Natl. Acad. Sci. U.S.A. 86 (1989) 8922–8926.

[110] R.B. Setlow, E. Grist, K. Thompson, A.D. Woodhead,Wavelengths effective in the induction of malignant mela-noma, Proc. Natl. Acad. Sci. U.S.A. 90 (1993) 6666–6670.

[111] F.E. Ahmed, R.B. Setlow, Ultraviolet radiation-inducedDNA damage and its photorepair in the skin of the platyfishXiphophorus, Cancer Res. 53 (1993) 2249–2255.

[112] R.B. Setlow, A.D. Woodhead, Temporal changes in theincidence of malignant melanoma: explanation from actionspectra, Mutat. Res. 307 (1994) 365–374.

[113] J.A. Moan, A. Dahlback, R.B. Setlow, Epidemiologicalsupport for an hypothesis for melanoma induction indicatinga role of UVA radiation, Photochem. Photobiol. 71 (1999)243–247.

[114] R.B. Setlow, Spectral regions contributing to melanoma: apersonal view, J. Invest. Dermatol. Symp. Proc. 41 (1999)46–49.

[115] R.B. Setlow, Alexander Hollaender, Photochem. Photobiol.8 (1968) 511–513.

[116] R.B. Setlow, In memorial: Alexander Hollaender, 1898–1986, Environ. Mutagen. 9 (1987) 107–110.

[117] R.B. Setlow, Alexander Hollaender (1898–1986), Genetics116 (1987) 1–3.

[118] R.A. McGrath, R. Williams, Reconstruction in vivo ofirradiated Escherichia coli deoxyribonucleic acid; therejoining of broken pieces, Nature 212 (1966) 534–535.

[119] F.W. Studier, Sedimentation studies of the size and shapeof DNA, J. Mol. Biol. 11 (1965) 373–390.

[120] J.E. Cleaver, Richard B. Setlow, A commentary on seminalcontributions and scientific controversies, Environ. Mol.Mutagen. 38 (2001) 122–131.



Hormesis: changing view of the dose-response, a personal account of the history and current status ☆Edward J. Calabrese *

Department of Environmental Health Sciences, School of Public Health and Health Sciences,University of Massachusetts, Amherst, MA 01003, USAAccepted 23 April 2002



Hormesis: changing view of the dose-response, a personalaccount of the history and current status�

Edward J. Calabrese∗

Department of Environmental Health Sciences, School of Public Health and Health Sciences,University of Massachusetts, Amherst, MA 01003, USA

Accepted 23 April 2002

Abstract

This paper provides a personal account of the history of the hormesis concept, and of the role of the dose response intoxicology and pharmacology. A careful evaluation of the toxicology and pharmacology literatures suggests that the biphasicdose response that characterizes hormesis may be much more widespread than is commonly recognized, and may come torival our currently favored ideas about toxicological dose responses confined to the linear and threshold representations usedin risk assessment. Although hormesis-like biphasic dose responses were already well-established in chemical and radiationtoxicology by the early decades of the 20th century, they were all but expunged from mainstream toxicology in the 1930s.The reasons may be found in a complex set of unrelated problems of which difficulties in replication of low-dose stimulatoryresponses resulting from poor study designs, greater societal interest in high-dose effects, linking of the concept of hormesis tothe practice of homeopathy, and perhaps most crucially a complete lack of strong leadership to advocate its acceptance in theright circles. I believe that if hormesis achieves widespread recognition as a valid and valuable interpretation of dose-responseresults, we would expect an increase in the breadth of evaluations of the dose-response relationship which could be ofgreat value in hazard and risk assessment as well as in future approaches to drug development and/or chemotherapeutics.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Hormesis; Biphasic; U-shaped; J-shaped; Risk assessment; Biological switching mechanisms; Dose-response relationships;Reflections

1. Introduction

This article is concerned with a phenomenon thathas come to be known as hormesis. Hormesis is adose-response phenomenon which is characterized bya counterintuitive switchover from low-dose stimula-


∗ Tel.: +1-413-545-3164; fax: +1-413-545-4692.E-mail address: [email protected] (E.J. Calabrese).

tion to high-dose inhibition that is not infrequently en-countered in the course of a toxicity assay. The storybegins with a scene-setting recapitulation of the dis-covery and early development of the hormesis con-cept, followed by an attempt to place names, dates,places and concepts into an integrative and insight-ful whole so as to introduce and analyze an importantarea of research whose potential significance for toxi-cology (and especially for risk assessors) is seriouslyunderappreciated in the wider scientific community.

For all practical purposes, the story of hormesis isone of the efforts of numerous investigators who havebeen striving for decades to enhance our knowledge


182 E.J. Calabrese / Mutation Research 511 (2002) 181–189

of the various factors that influence the dose-responserelationship, the keystone in all of toxicology. In theunfolding of this reconstruction what will emergemost clearly, and perhaps surprisingly, is that eventhough the linear and threshold dose-response mod-els are the twin pillars of toxicology from whichresearchers and regulators have derived so much aca-demic/institutional guidance for so many years, theydo not reflect the most fundamental toxicologicalmodel, which is the hormesis model. This is a typeof whodunit review in which I provide my versionof how, when, and why early workers in the field oftoxicology made a mistake of historic proportions onwhat the most fundamental nature of the shape of thedose response should be, then discuss why multiplegenerations of toxicologists have continued to perpet-uate this error while imbuing it with an influence thathas transformed essentially all US and internationalestimates of the risks of chemicals and radiation. Thisstory is not pretty, but it should be told.

2. The hormesis concept: the early years

The story of hormesis began about 60 years be-fore it acquired its present name1 in the obscure andill-equipped laboratory of Prof. Hugo Schulz at theUniversity of Greifswald in Northern Germany [1–5].The scene was the early 1880s, when Joseph Lister,Louis Pasteur and Robert Koch were busy setting Eu-rope on fire with their many important discoveries onthe causes of infectious diseases and how they mightbe prevented. Schulz became interested in testing howa medicine long used to treat a form of gastroenteritismight work. He noted that research had recently deter-mined the causative bacterial agent and how to cultureit. So he grew the bacteria and exposed them to variousdoses of the medicine. To his surprise, the medicinehad no effect on the microbe, regardless of dose. Thisled Schulz to a hasty, and possibly erroneous, con-clusion that the chemotherapeutic agent acted on thehuman body by enhancing natural adaptive responsesrather than by attacking the microbe more directly.

1 Hormesis was given its name by Southam and Ehrlich [6],who had been studying the effects of red cedar extracts on fungiand were reporting the unusual biphasic dose-response curves thatthey kept seeing when they plotted their results.

This interpretation gave him a profound appreciationfor the medical practice of homeopathy.

Independently of this initial study, Schulz assessedthe effects of numerous bactericidal agents, includingmercury and phenol, on yeast metabolism. He foundthat most agents appeared to stimulate metabolism atlow concentrations only to inhibit them at higher con-centrations. At this point Schulz believed he had founda toxicological explanation for his developing home-opathic beliefs. As a result of the publicity followingthese initial studies he became the main academic herofor numerous advocates of homeopathy, and thus thetheory of hormesis was born in close association withhomeopathy as a preventive/therapeutic modality.

Homeopathy has long been an embattled approachto medical practice, and Schulz’s decision to embracehomeopathy may well have ensured that his careerremained in Greifswald rather than in some closer as-sociation with Koch, from whose lab three of the firstseven Nobel Prize winners in biology and medicinewould later be drawn. Nonetheless, Schulz’s work didattract a fair amount of attention, and was confirmedby a few and extended by many. The possibility thathis findings were of general significance occurredto a number of people, and the biphasic responsewas encountered in a variety of other organisms, in-cluding several species each of bacteria and plants.Interestingly, soon after Ferdinand Hueppe, a protegeof Koch and the author of a highly regarded texton bacteriology in 1896, claimed to have confirmedSchulz’s biphasic phenomenon in bacteria, he beganto argue with considerable passion that this conceptshould not be rejected simply because of Schulz’sclose association with homeopathy. Interestingly, theconcept of low-dose stimulation/high-dose inhibition,as seen initially in the work of Schulz and subse-quently in that of Hueppe, soon became known asHueppe’s Rule, primarily one supposes because thelatter had the prestige of Koch’s lab even thoughscientific primacy belonged to the more marginalizedSchulz. This situation changed shortly afterwards,however, when Grote and Schulz [7] published their1912 book on biphasic responses in which they gavecredit and esteem to the reputation of their deceasedcolleague, the homeopath physician, Rudolph Arndt,for his role in galvanizing Schulz’s attention to theissue as well as his role in the process of concept de-velopment. As a result, the phenomenon that was to


E.J. Calabrese / Mutation Research 511 (2002) 181–189 183

become known by a great many people as hormesissoon became known as the Arndt–Schulz law.2

Interest in the effects of low doses rapidly expandedespecially with many studies of interactions involv-ing (mainly) plants, bacteria and fungi, most notablyin Europe, the US and Japan. Convincing findingswere reported by outstanding turn-of-the-century re-searchers at leading universities in the US, includingCornell (Benjamin Duggar, later to be a mentor ofAlexander Hollaender at the University of Wisconsin),Wisconsin (Louis Kahlenberg, former student of Wil-hem Ostwald, Nobel Laureate), Illinois (F.L. Stevens),Columbia (C.O. Townsend, H.M. Richards), andStanford (G.H. Jensen) and many others [2,3]. Thesestudies were mostly well designed and executed, andcarefully interpreted. Their emphasis on plant, bac-terial and fungal systems enabled early investigatorsto make use of the larger numbers of doses that areso important if one wishes to describe the biologicaleffects at low concentrations that are now known ashormesis. Similar findings were made by outstandingEuropean scientists, one of whom (Charles Richet)was later to receive the Nobel Prize for his study ofanaphylaxis. It is important to note that none of thesereports of stimulation at low doses and inhibitionat higher doses were either controversial or linkedto any particular ideology. In the US, this researchwas not closely linked to homeopathy but rather wasseen as principally a scientific question. Of partic-ular note was the Yale Prof. Charles Winslow, theEditor-in-Chief of the Journal of Bacteriology, whodirected the students responsible for numerous disser-tations in the 1920s through the mid 1930s that ex-plored the occurrence of biphasic dose responses andtheir mechanistic framework. Thus, as the chemicallyoriented toxicology community was approaching themid 1930s, the concept that was to become hormesiswas part of the mainstream of scientific investiga-tion and had the support of an impressive line-up ofinvestigators.

2 Note: our analysis of the concept of hormesis and its supportivedata has resulted in our distinguishing between hormesis andthe Arndt–Schulz law. The Arndt–Schulz law assumed a directstimulation at low doses followed by an inhibition at higher doses.Our concept of hormesis incorporates two forms of stimulation;one is a direct one that could be similar to that of the Arndt–Schulzlaw, while a second stimulation may result from overcompensationto an initial disruption in homeostasis.

On a parallel track, but with a nearly two decadehandicap (since X-rays and radionuclides had not beendiscovered until the mid 1890s) an impressive array ofinvestigators was making similar claims of low-dosestimulation/high-dose inhibition by radiation mostly inplants, fungi, insects, but also in clinical practice [3–5].While these findings were less extensive than thosealready reported in the chemical literature, they weresufficient to have done justice to most other similarscientific hypotheses during that era. However, some-thing happened during the 1930s to both concepts (i.e.chemical and radiation hormesis) that was to resultin the hormetic concept becoming marginalized—andremaining there until the present time.

How could two viable hypotheses that had such animpressive start and that had clearly established a sci-entific beachhead simply melt away to near irrelevancyin an earlier version of “cold fusion”? The causes aremany and interacting, with hormesis being the inno-cent victim of inherent theoretical limitations and so-cietal priorities as well as external enemies and a lackof scientific leadership. A most difficult problem in-herent to hormesis is that the low-dose stimulation thatoccurs is modest; this makes it difficult to replicate.Societal interest during the early decades of the 20thcentury was also more focused on high-dose effects,such as ensuring the killing of bacteria and insects, oron reducing the exposure of industrial workers to rel-atively high doses of toxic agents, than on looking formodest low-dose stimulatory responses of uncertainsignificance.In the radiation field, hormesis came to be greatly

exaggerated by proselytizers, who oversold what ra-diation hormesis could never have delivered. For ex-ample, chemicals, such as radium were portrayed asthe next elixir, or as effective plant fertilizers [3–5].When the hormetic concept miserably failed to satisfyon these accounts, interest dried up quickly and moreserious observers concentrated on assessing adversehealth effects at higher doses. In addition, the inter-nationally famous A.J. Clark of the University of Ed-inburgh took aim at the Arndt–Schulz law, directinga fierce and unrelenting attack on it in his acclaimedHandbook of Experimental Pharmacology [8]. Thus,there was a combination of being difficult to prove,a subject of oversold promises, an obvious linkage tohomeopathy, governmental concern with high-dose ef-fects, ineffective scientific leadership and prestigious


opponents; all of this proved too much for hormesis toovercome. By the end of the 1930s, it was a marginal-ized concept.

3. Hormesis—a renewed interest

Over the next 60 years, the hormetic hypothesis hadseveral minor resurgences, only to go back to hav-ing essentially negligible name recognition and influ-ence. However, a strategic opening for hormesis to un-dergo a resurgence emerged slowly in the early 1980sas the EPA and other regulatory agencies acceptedlow-dose linearity modeling to estimate cancer risksand began to use this as a vehicle to answer the “howclean is clean?” question for contaminated sites. Sincehormesis implied thresholds for non-carcinogens andcarcinogens alike it was thought of as a means withwhich to confront the low-dose linearity paradigm. Infact, this was the issue motivating the First Confer-ence on Radiation Hormesis in 1985 (see Proceedingsin Health Physics, 1987) and some of the biologicaleffects of low level exposures (BELLE) (see belleon-line.com) activities over the past decade.

As an outgrowth of these activities we have beenattempting to obtain a better understanding of the na-ture of the dose-response relationship, especially inthe low-dose zone. Our principal motivation was a de-sire to resolve the long-standing scientific debate (andoften ideological feud) over whether the threshold orlinear response model should be used as the default as-sumption in the assessment of carcinogen risk. Theredid not seem to be an issue for non-carcinogens, giventhat the threshold response model already had broadacceptance here.

Assessing threshold versus linear models for car-cinogens has the potential to bring one in contactwith the biphasic-hormetic3 response model in whichresponses at low doses are opposite and not pro-portional to those observed at high doses (Fig. 1).

3 Definition of hormesis: hormesis is a dose-response phe-nomenon characterized by a low-dose modest stimulation (e.g.about 30–60% greater than the control at maximum), and ahigh-dose inhibition. This response appears to result in two ways,either as an overcompensation to a disruption in homeostasis orvia a direct stimulation/inhibition response. A detailed article anddebate on the definition of hormesis has been recently published[9].

Fig. 1. Stylized curves illustrating the linear, threshold and hormeticdose-response models for carcinogens.

In the case of carcinogens, this generally meansthat the agent-induced cancer risk disappears with aJ-shaped/hormetic curve, with low doses being as-sociated with cancer risks less than controls whileat higher doses one expects a dose-dependent car-cinogenic effect as normally predicted. The real sig-nificance of the hormetic model in the conflict overthreshold versus linear response models is of coursethat if hormesis could be unequivocally demonstratedas universal then it would establish a bona fide thresh-



old for carcinogenic effects. This would immediatelydiscredit the many uses of linearity models to esti-mate cancer risk at low doses. Thus, the unequivocaldemonstration and characterization of hormesis, andits acceptance as the default model in estimatinglow-dose effects, clearly has the potential to drasti-cally affect cancer risk assessment. In the absence ofhormesis it has been extremely difficult to differenti-ate a linear from a threshold model in the low-dosezone. In such situations, the linear model tends to beadopted, in large measure because the precautionaryprinciple framework is allowed to prevail.

4. Hormesis: establishing documentation

Despite the impetus to resolve the low-dose cancerrisk assessment question via the use of the hormeticmodel, we reasoned that if hormesis is a real phe-nomenon then it must be evolutionarily based andshould therefore be quite broadly distributed acrossbiological systems, working through multiple phys-iological processes with underlying molecular path-ways that should be amenable to analysis. Thus, itmade little sense to focus solely on the questionof chemical/radiation-induced cancer and its riskassessment, as important as this was. We felt thatthe question needed to be framed in a broader andmore fundamental biological/evolutionary way. Ourstrategy in assessing hormesis therefore was to estab-lish objective criteria to test whether it existed andwas generalizable, as evidenced (for example) by itsoccurrence independently of biological system, end-point, or chemical/physical stressor. If all of thesecriteria were satisfactorily met it would be importantto determine the quantitative features of the dose re-sponse, investigate its mechanistic foundations, andattempt to understand how these relate to homeostaticregulatory mechanisms. A starting point had to bethe development of a definition of hormesis that wastoxicologically and statistically based so that an ob-jective and consistent evaluative framework could begenerally agreed upon a priori.

Although some examples of hormesis were knownto exist in the toxicological literature, at the initialstage of our assessment we were not sure if thesewere relatively rare exceptions (assuming they did notlack adequate replication, statistical power or mech-

anistic plausibility, as not a few supposed examplesclearly did), or whether this phenomenon was morewidespread and predictable. If hormesis was to be ofwidespread significance in the biomedical and toxi-cological domains it needed the important features ofbiological centrality. That is, it should be commonlyobserved, and should be an evolutionary expectation,not an exception.

Our initial strategy was to identify and evaluate pos-sible examples of hormesis from the broad spectrumof biologically based research without restrictions asto system, endpoint or agent tested. We would thenapply our own quantitative evaluation methodologyto the data for each of the examples we uncovered.As a result of this approach we were able to iden-tify many hundreds of specific examples of hormeticdose responses in various computer databases using acombination of critical key word descriptors, consid-erable cross referencing, and systematic hand search-ing/inspection of all articles in more than two dozenjournals from their inception to the present. This sur-vey clearly established that hormetic effects are in factcommon when experiments have been properly de-signed to assess dose responses in the low-dose range.In fact, we now have an database of over three thou-sand examples of the hormesis phenomenon. A de-tailed evaluation of this database revealed certain im-portant common dose-response features that appearedto be independent of organism, biological system, end-point or agent tested [10–13].

The low-dose stimulatory responses within thehormetic dose-response curve appear to have a lim-ited amplitude, almost always going no more thana factor of two above the control value. The maxi-mum stimulatory response is more commonly in the30–60% above control range. Thus, we believe thatwhen a stimulatory response of greater than four-foldis observed it is likely to be a biological phenomenonthat is different from hormesis.

The stimulatory dose range for hormetic effects istypically less than a factor of 20-fold immediately be-low the NOAEL (no observed adverse effect level).This would account for about 70% of the examples sofar examined. However, in about 2% of the examplesof hormesis examined the stimulatory dose range isin excess of 1000-fold (Fig. 2). While the causes ofthis potential broad variability in stimulatory range re-main to be assessed, we have made some progress in


Fig. 2. Stylized dose-response curves reflecting the relative distribution of stimulatory dose ranges. Note: the maximum stimulatory responseis usually 130–160% of the control value.

understanding how the range can be modulated. Suchan understanding may be important from a risk as-sessment perspective, but may be even more importantfrom a clinical perspective where dose optimizationrather than exposure minimization is the driving force.By our definition, the hormetic response is always

linked to the traditional toxicological NOAEL [9].That is, the hormetic stimulation is contiguously fol-lowed by a transition into the traditional toxic re-sponse zone. The linkage of the hormetic response tothe toxicological NOAEL is critically important, sinceit provides a stable frame of reference in relation torisk assessment guideposts and the goals of toxicolog-ical testing and assessment. By relating the hormeticresponse to the dose-response continuum containingthe NOAEL, LOAEL (lowest observed adverse effectlevel), etc. in a predictable and stable manner, risk as-sessors are able to enhance the flexibility and accuracyof their assessments by permitting the incorporation ofthe concept of optimization of exposure/response intotheir methodology. This allows hormesis to achieve abroader recognition as a fundamental and central fea-ture of biological processes. On the theoretical sidethe linkage of the hormetic stimulation to the NOAELfunctionally enjoins it with the concept of homeosta-sis, a universal concept and phenomenon.

While the initial hormesis database was useful inestablishing the biological validity of this concept, itdid not provide sufficient insight into the frequency ofhormetic responses in the toxicological literature. Thiswas considered a vital step in any meaningful evalua-tion of the potential biological centrality of hormesis.

If hormetic effects were real but occurred in fewer than1% of properly designed studies, its utility would berather limited. However, if it were significantly morecommon (e.g. >30%) then more formal acceptance intesting and assessment procedures would be called for.

In order to estimate the frequency with whichhormetic effects could be identified in the toxicologi-cal literature we created a second hormesis database.In this evaluation, we screened over 20,000 articles inthree toxicological journals covering issues from themid 1960s to the present [10]. Only 1.5–2.0% of thestudies had study designs that were consistent withour rigorous a priori entry criteria. Of the dose re-sponses that did satisfy the entry criteria (in practicethis means having multiple doses below a toxicologi-cally derived NOAEL plus a clear toxicological doseresponse at the higher dose levels) approximately 40%were found to satisfy the evaluative criteria, i.e. to dis-play reasonably convincing evidence of hormesis. (Ifwe had relaxed the evaluative criteria to a limited butstill fairly demanding level, the frequency would havebeen well above 50%.) Even more impressive is thatthe frequency of statistically significant responses fordoses below the NOAEL was 32-fold more frequentin the hormetic direction than in the opposite direc-tion. This clearly supports the non-random nature ofthe below NOAEL responses.

While the frequency and quantitative features ofthe hormetic dose response are critical to the accep-tance of hormesis as a central theorem of toxicol-ogy, we also needed to develop a better mechanisticunderstanding before broader acceptance would be



forthcoming. Our strategy to address the issue ofmechanism was found in the pharmacological, ratherthan the more traditional toxicological, literature. Weneeded to know the mechanisms by which biolog-ical systems operate a switch from stimulation toinhibition. Although precisely focused dose-responseexplanatory data are quite rare in toxicology, they arerelatively commonplace in pharmacology, and indeedthe pharmacological literature provided evidence forhormesis-like dose responses for essentially all re-ceptor systems (e.g. dopamine, prolactin, adrenergic,opiate, adenosine, nitric oxide, various prostaglandinsand others) along with mechanistic explanations atleast down to the receptor level, and often at evengreater depths of understanding. Further investigationrevealed that opposing responses were often inducedby the same endogenous agonist depending upon dif-fering affinities to receptor subtypes that might lead toeither stimulatory or inhibitory pathways [11]. Whenassessed over a broad dose range the response wastypically biphasic with quantitative features like thoseof chemical and radiation-induced hormesis. Thus, wecould now demonstrate the existence of hormesis-likeeffects in both the toxicological and pharmacologicalliteratures [12–14], provide a reasonable estimate oftheir frequency over the past three to four decades[10], and confidently account for the biphasic featuresof the dose-response relationship in essentially allreceptor-based systems [11].4 This remarkable se-quence of events over a 5-year period brought us fromthe initial stage of not knowing whether hormesisexisted to a recognition that it is likely to be a centralfeature of a great many biological systems, poten-tially with enormous implications for pharmacology,toxicology and medicine. The original question ofthreshold versus linearity in cancer risk assessment,while obviously still very important, then becamesecondary to understanding a phenomenon that wasof even greater fundamental interest. Nonetheless,our emerging perspective on the generalizability ofhormesis ought to be able to contribute to an im-proved framework in which to envisage the nature of

4 Mechanisms underlying changes in the dose response (as in thehormetic biphasic response) are rarely discussed in the toxicologi-cal literature, including the chemical and radiation carcinogenesisliteratures. The different mechanistic focus between pharmacologyand toxicology is an important difference that has been generallyunder-appreciated.

carcinogen dose responses as well as their underlyingmechanisms.

5. Hormesis and cancer

While much of the motivation that re-invigoratedthe concept of hormesis is its theoretical potential tochange the current cancer risk assessment model it isimportant to be aware that although the hormesis con-cept has a long history, it is not that long since theidea that carcinogen dose responses may be U-shapedfirst emerged. Thus, for example Luckey [15], an ar-dent supporter of hormesis, did not even discuss thetopic of cancer in his first book on radiation horme-sis [15] whereas his second (1991) book contains alengthy discussion of matters relating to cancer [16].This probably reflects the fact that a number of animalmodel studies dealing with chemical and radiation car-cinogenesis and hormesis were first described in printin the 1980s, while a few similar studies (e.g. [17–24])were published prior to 1980. However, the conceptof hormesis in these studies was ignored by exter-nal reviewers and either ignored or de-emphasized bythe authors as well. Thus, what might normally havebeen reported as a striking new finding, was presentedas an “observation”, never highlighted or discussed.Nonetheless, many of these studies revealed hormeticresponses, and suggested that such findings might beobserved again if appropriate study designs were em-ployed. More recently, we published two papers inwhich we specifically discussed chemical [25] and ra-diation [26] induced cancer and hormesis. In additionto providing numerous examples of hormesis in ani-mal model studies, we also demonstrated that hormeticeffects could be observed at a number of stages priorto the development of a tumor, and discussed possi-ble mechanistic hypotheses that could account for thehormetic responses seen in the cancer bioassays.

The detection of hormetic responses presents uniquechallenges for cancer bioassays. First, it requires thatthere be a large number of doses, including severalbelow the apparent threshold. Second, it is necessaryfor the background incidence of tumors to be suffi-ciently high that hormetic decreases, if they shouldoccur, will definitely be detected. While there are sev-eral other complexities that might affect the outcome,selection of the animal/tumor system and the details


of study design are prominent, and these alone maygreatly restrict the numbers of experimental studiesfrom which to assess possible relationships betweenhormesis and cancer more rigorously. Nonetheless,we believe that a solid core of such studies alreadyexists and is very reliable, thus extending the hormetichypothesis into the general area of carcinogenesis(see [25,26] for reviews).

6. Final perspectives

While efforts to assess hormesis have been rewardedwith unexpected biological insights, it is important toemphasize that however common hormesis is in prop-erly designed experiments, it is not easily studied andis often therefore overlooked or ignored, thereby con-tributing to its general omission from toxicologicaltext books and its poor recognition and/or acceptance.It is difficult to study primarily because it requires theinvestigator to use a large number of properly spaceddoses, to identify a reliable NOAEL in preliminaryexperiments, and, frequently, to include a temporalcomponent in some experiments. It also requires typi-cally large sample sizes, mainly because the low-dosestimulation is modest and statistical power issuesquickly become critical. Likewise, with modest stim-ulatory responses it is always possible that chance is areasonable alternative explanation. Therefore, claimsof hormetic responses need to be carefully and mul-tiply replicated to provide evidence of proper causaldetermination. Thus, the study of hormesis can neverbe a casual exercise using the same study design rulesas will suffice in high-dose toxicity studies. Simplyadding an extra dose in the low-dose zone will not al-low one to properly test for hormesis. Yet, if a criticallong-term goal of toxicology is to better understandhow biological systems respond to low levels of envi-ronmental stressor agents then the careful assessmentof hormesis becomes central and critical. Hormesisis also important since it demonstrates that high- tolow-dose extrapolation assumptions used in cancerrisk assessment are no longer necessarily adequate incharacterizing low-dose risks [27].

Although we have approached the issue of horme-sis from a toxicological perspective, researchers fromother biological fields may have addressed the sameor closely related concepts under the guise of different

names, including the following: J-shaped, U-shaped,reverse dose response, opposite effect, overcompen-sation response, Arndt–Schulz law, Yerkes–Dodsonlaw, bi-directional responses, dual effects, subsidygradients, intermediate disruption hypothesis and al-most certainly others. Regardless of this academicdichotomy as seen through the use of different wordsfor similar concepts, the biological systems that pro-vide such information frequently show hormetic-likedose responses with remarkably similar quantitativefeatures.

Despite its widespread occurrence in diverse bio-logical systems, hormetic responses may not alwaysbe observed in properly designed studies focusing onlow-dose responses. We have sufficient evidence thatthis is true, along with adequate mechanistic expla-nations to account for such exceptions. However, ourview is that the biphasic features of the dose responseare essential attributes that cells and organisms requireto function and survive. Smooth muscles need to ei-ther relax or contract, cells must proliferate or not,neutrophils need to be recruited to sites of injury andinformed when to stop coming and when to returnhome. These and numerous other biological regula-tory processes are activated by complementary stim-ulatory and inhibitory pathways that are themselvesmodulated by agonist concentration gradients that turnon/off biological switches that toxicologists see exper-imentally as biphasic-hormetic curves.

The concept of the dose response is in an importanttransition. The issue will no longer be whether the re-sponse is a threshold or linear but the role of biologi-cal switching (i.e. hormetic mechanisms) in health anddisease. I believe that the recognition and knowledgeof hormetic responses will bring with it an opportunityto profoundly improve risk assessment, and especiallycancer risk assessment, to harmonize the risk assess-ment of both non-cancer and cancer effects, and to rev-olutionize approaches to drug development and clini-cal medicine in which biphasic responses are widelyacknowledged, yet where such knowledge is rarelyused for patient benefit or the avoidance of harm.

Finally, I return to the opening theme in which I as-serted that the concept of hormesis has been systemat-ically ignored by those in the field of toxicology sincethe late 1930s, as judged by its absence from majortexts, or professional society activities (including ses-sions at annual conferences), and especially perhaps



as judged by the actions of certain government regula-tory agencies which “institutionalized” toxicologicalconcepts, such as the NOAEL and low-dose linearity.Indeed, the whole field of modern toxicology hasbeen built on the twin assumed “facts” of NOAELand linearity at low doses. This reality directed thefield, the questions asked, the projects funded, and thebooks written, thereby making the marginalizationof the hormesis concept solid and even reinforced.It is my feeling that unless toxicologists develop arenewed interest in the concept of hormesis criticalprogress in many areas will be delayed and the fieldas a whole will suffer. More important than this, how-ever, is the troubling philosophical question of howa genuine toxicological hypothesis such as hormesis,could have been eliminated from debate in the mostopen of modern societies.

References

[1] E.J. Calabrese, L.A. Baldwin, Chemical hormesis: itshistorical foundations as a biological hypothesis, Hum. Exp.Toxicol. 19 (1) (2000) 2–31.

[2] E.J. Calabrese, L.A. Baldwin, The marginalization ofhormesis, Hum. Exp. Toxicol. 19 (1) (2000) 32–40.

[3] E.J. Calabrese, L.A. Baldwin, Radiation hormesis: itshistorical foundations as a biological hypothesis, Hum. Exp.Toxicol. 19 (1) (2000) 41–75.

[4] E.J. Calabrese, L.A. Baldwin, Radiation hormesis: the demiseof a legitimate hypothesis, Hum. Exp. Toxicol. 19 (1) (2000)76–84.

[5] E.J. Calabrese, L.A. Baldwin, Tales of two similar hypotheses:the rise and fall of chemical and radiation hormesis, Hum.Exp. Toxicol. 19 (1) (2000) 85–97.

[6] C.M. Southam, J. Ehrlich, Effects of extracts of westernred-cedar heartwood on certain wood-decaying fungi inculture, Phytopathology 33 (1943) 517–524.

[7] L.R. Grote, Hugo Schulz, Die Medizine Der GegenwartIn Selbstdarstellungen (Contemporary Medicine as Presentedby its Practitioners Themselves), NIH-98-134 (Ted Crump,Trans.), 1923, pp. 217–250.

[8] A.J. Clark, Handbook of Experimental Pharmacology,Springer, Berlin, 1937.

[9] E.J. Calabrese, L.A. Baldwin, Defining hormesis, BELLENewslett. 10 (2) (2002) 25–30.

[10] E.J. Calabrese, L.A. Baldwin, The frequency of U-shapeddose responses in the toxicological literature, Toxicol. Sci.62 (2001) 330–338.

[11] E.J. Calabrese, L.A. Baldwin (Eds.), The Scientific Founda-tions of Hormesis, Crit. Rev. Toxicol. 31 (2001) 349–691.

[12] E.J. Calabrese, L.A. Baldwin, The dose determines thestimulation (and poison), Int. J. Toxicol. 16 (1997) 545–559.

[13] E.J. Calabrese, L.A. Baldwin, A quantitatively basedmethodology for the evaluation of hormesis, Hum. Ecol. RiskAssess. 3 (1997) 545–554.

[14] E.J. Calabrese, L.A. Baldwin, C. Holland, Hormesis: a highlygeneralizable and reproducible phenomenon with importantimplications for risk assessment, Risk Anal. 19 (1999) 261–281.

[15] T.D. Luckey, Hormesis with Ionizing Radiation, CRC Press,Boca Raton, FL, 1980.

[16] T.D. Luckey, Radiation Hormesis, CRC Press, Boca Raton,FL, 1991.

[17] R.W. O’Gara, M.G. Kelly, J. Brown, N. Mantel, Inductionof tumors in mice given a minute single dose ofdibenz[a,h]anthracene or 3-methylcholanthrene as newborns.A dose–response study, J. Natl. Cancer Inst. 35 (1965) 1027–1042.

[18] R.J. Kociba, D.G. Keyes, J.E. Beyer, R.M. Carreon, C.E.Wade, D.A. Ditenber, R.P. Kalnins, L.E. Frauson, C.N.Park, S.D. Barnard, R.A. Hummel, C.G. Humiston, Resultsof a 2-year chronic toxicity and oncogenicity study of2,3,7,8-tetrachlorodibenzo-p-dioxin in rats, Toxicol. Appl.Pharmacol. 46 (1978) 279–303.

[19] Office of Technology Assessment (OTA), Cancer TestingTechnology and Saccharin, US Government Printing Office,Washington, DC, 1977.

[20] W.A. Foley, L.J. Cole, Urethan-induced lung tumors in mice.X-radiation dose-dependent inhibition, Radiat. Res. 27 (1966)87–91.

[21] R.L. Ullrich, M.C. Jernigan, G.E. Cosgrove, et al., Theinfluence of dose and dose rate on the incidence of neoplasticdisease in RFM mice after neutron irradiation, Radiat. Res.68 (1976) 115–131.

[22] R.L. Ullrich, M.C. Jernigan, J.B. Storer, Neutroncarcinogenesis: Dose and dose-rate effects in BALB/c mice,Radiat. Res. 72 (1977) 487–498.

[23] R.L. Ullrich, J.B. Storer, Influence of γ irradiation on thedevelopment of neoplastic disease in mice. II. Solid tumors,Radiat. Res. 80 (1979) 317–324.

[24] J.J. Broerse, S. Knaan, D.W. van Bekkum, C.F. Hollander,A.L. Noteboom, M.J. van Zwieten, Mammary carcinogenesisin rat after X- and neutron-irradiation and hormoneadministration, Late Biol. Eff. Ionizing Radiat. II (1978) 13–27.

[25] E.J. Calabrese, L.A. Baldwin, Can the concept of hormesisbe generalized to carcinogenesis? Reg. Toxicol. Pharmacol.28 (1998) 230–241.

[26] E.J. Calabrese, L.A. Baldwin, Radiation hormesis and cancer,Hum. Ecol. Risk Assess. 8 (2001) 327–353.

[27] R. Sielken, D. Stevenson, Some implications for quantitativerisk assessment if hormesis exists, Hum. Exp. Toxicol. 17(1998) 259–262.


Observations at the interface of mutation research and regulatory policy ☆W. Gary Flamm * 1

4401 Sunset Drive, Vero Beach, FL 32963, USAAccepted 13 March 2003



Observations at the interface of mutationresearch and regulatory policy�

W. Gary Flamm∗,1

4401 Sunset Drive, Vero Beach, FL 32963, USA

Accepted 13 March 2003

Abstract

Between 1970 and 1975 developments in environmental mutagenesis proceeded with amazing speed. These developmentswere both structural and conceptual in nature. A new infrastructure was built and new concepts about how best to protectconsumers from exposures to mutagens emerged. The internal dynamics within the Food and Drug Administration played animportant role and is discussed with regard to modifications in testing protocols as well as changes in the overall approachused to protect consumers. It is clear that this exciting period in the early days of environmental mutagenesis has provided abase for growth and development of the field and continues to affect and guide future developments.© 2003 Elsevier Science B.V. All rights reserved.

Keywords: Mutagenesis; Genetic toxicology; Science policy; History of science; Food and Drug Administration; Federal, Food, Drug, andCosmetic Act (FDCA)

To those watching at the time (ca. 1970), the wholefield of environmental mutagenesis seemed to explodeinto existence like some cosmic big bang. While tomany this big bang came out of nowhere, the truthis that it was a predictable consequence of two greatforces coming together. The environmental movementhad been gathering public and political strength forseveral years and was rapidly becoming a powerfulforce for funding research. In need of new sources offunding was a group of radiation geneticists and biolo-gists who were beginning to look beyond the radiation

� This article is a part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, readers shouldcontact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).

∗ Corresponding author.E-mail address: [email protected] (W.G. Flamm).

1 Dr. Flamm is the former Branch Chief of Genetic Toxicologyat the US Food and Drug Administration, Washington, DC.

field and their principal patron, the Atomic EnergyCommission (AEC), for that support. The intellectualprowess of the group constituted the other great force,which was unleashed by declines in the funding of ra-diation genetics and mutagenesis. Heading the effortto bring these forces together was the late and greatAlexander Hollaender who had the vision and deter-mination to create, with help from others, the com-ponents of an entire scientific field and to accomplishthis creation “overnight.”

With a wave of his magic wand, new books onthe subject appeared, review articles were written,federal contracts were granted for testing chemicalsfor mutagenicity, the Environmental Mutagen Society(EMS) was founded with Alex as its first president,an informative newsletter was created, a new journalsection was developed, and the Environmental Muta-gen Information Center was established by HeinrichMalling, one of the talented young scientists attracted

1383-5742/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S1383-5742(03)00031-0


2 W.G. Flamm / Mutation Research 544 (2003) 1–7

to the Biology Division that Alex had built at OakRidge National Laboratory. By the end of 1970, Alexhad, with the help of a small army of highly qual-ified scientists, put everything in place: workshops,annual meetings of the EMS, plans to establish EMSsocieties in other countries and to create an interna-tional society to bind them together, involvement ofthe National Academy of Sciences (NAS), interest inCongress sufficient to generate funding, the attentionof the National Institutes of Health (NIH) and last butnot least, strong interest by regulatory agencies. Thesewere indeed amazing accomplishments of an amaz-ing man. One major key to his success was an abilityto impart to others the importance of the tasks thatlie ahead and by the same token, to impart to thema genuine sense of their importance. He was a greatlistener and always considered new ideas carefully. Inshort, Alex just made you feel better about yourselfand the value of your work. Small wonder he had somany willing and ready recruits helping him realizehis vision of a world in which mutagenesis researchand testing would play a major role. The recruitsknew also that Alex would praise and promote theirefforts and would never take the credit they earned.

Marvin Legator, who often referred to Alex as the“Dean of Mutagenesis,” was one such talented andresourceful recruit, strategically located at the Foodand Drug Administration (FDA). He somehow per-suaded the FDA to establish the Genetic ToxicologyBranch, in which he served as Branch Chief, andto fund a large research and testing contract, testinghundreds of food ingredients for mutagenicity. Asregulatory agencies tend to be extremely conserva-tive in adopting new technologies and science, thiswas a stunning achievement. Though Marvin can tellhis story better than I, the reason the Agency acqui-esced to extensive mutagenicity testing related, inpart, to the enormous pressure FDA was under fromthe public and Congress to provide assurances thatthe substances added to food were adequately testedand shown to be safe. At the time, hundreds of sub-stances were permitted to be added to food on thepresumption of safety as a result of “grandfathering”or judgement calls that may or may not have had astrong scientific foundation. Mutagenicity testing, asthen set forth by the collaborative efforts of MarvinLegator, Samuel Epstein and Warren Nichols, ap-pealed to some at FDA because it was relatively quick

and inexpensive, making it possible to test hundredsof substances in a matter of a few years for a cost ofmillions of dollars as opposed to tens of millions forcarcinogenicity studies on an equivalent numbers ofsubstances. Indeed, unlike carcinogenicity studies, thetime and expense involved was sufficiently minimalto make repeat testing practical in cases where the re-sults were statistically ambiguous. Thus, questionableresults could be clarified by further testing providingconfidence that the results were in fact reproducible.

Upon succeeding Marvin as Branch Chief of Ge-netic Toxicology in 1972, I was pleased but a bit over-whelmed by the breadth and amount of activities oc-curring in the Branch at the time. My reflections on theinterface of science and regulatory policy will neces-sarily focus on the FDA, but I suspect that parallels willbe evident to those involved in science and regulationin other agencies and in other countries. The strengthof the Genetic Toxicology Branch at FDA derivedfrom the dedication of its competent researchers, in-cluding David Brusick, who left shortly after I arrived,Sidney Green, Vernon Mayer, Errol Zeiger, MichaelPrival, Ken Palmer, and later, Virginia Dunkel, all ofwhom have gone on to make major contributions tomutagenesis research. Fortunately, the group willinglytook time away from their own research to assure thatthe contract work established to conduct testing offood ingredient substances was done in accordancewith professional standards.

The food ingredients being tested were the so-called“generally recognized as safe” (GRAS) substances,which had been placed on a list compiled by the Na-tional Academy of Sciences. GRAS substances were anew category of food ingredients created by the 1958amendments to the Federal, Food, Drug, and CosmeticAct (FDCA) (Section 201(s)). This provision of theAct provided that a substance determined to be GRASfor its intended use in food by qualified food safetyexperts was exempt from the legal requirements im-posed on food additives as set forth in Section 409of the FDCA. For instances, a GRAS substance didnot have to be approved by FDA prior to marketingthat substance in food as did a regulated food additivesubject to Section 409. GRAS substances have to bedemonstrated as safe for their intended use in food butthe means of demonstrating safety is left to the dis-cretion of food safety experts, including those expertsoutside FDA.

W.G. Flamm / Mutation Research 544 (2003) 1–7 3

While there was great concern about the safetyof GRAS substances at the time by the public andCongress, most such substances were in reality com-mon food spices or flavors, essential vitamins andminerals, common inorganic salts, vinegar, naturalamino acids, dietary fats, fatty acids and other com-mon dietary lipids, salts of citrate, and other sub-stances commonly found naturally in food. Why allthe fuss and concern? Partly because most peoplenever knew or appreciated how innocuous the sub-stances were that comprised the GRAS list and partlybecause cyclamate and saccharin, two artificial sweet-eners, were on the GRAS list and also in the newsand reputed to be “bad actors” though they continueto be marketed in over 40 countries worldwide today.

Despite the innocuous nature of the GRAS sub-stances, we in the Branch were more cautiously cir-cumspect. None of these substances had been testedfor mutagenicity before, so who knows what the out-come would be. Others outside the Branch had a verydifferent view and would take us aside to say—“youguys in the Branch will be in a lot of trouble if youfind that common components of food are mutagenicin your test systems. It will reflect badly on your tests,not on the food components you claim are positive inmutagenicity tests. No one will pay any attention toyou or the tests you have been using if simple thingslike salt, sugar and vinegar are found or asserted to bemutagenic based on your test results.” At this pointI began to understand that the agency was not takingthe big risk that I had thought by subjecting GRASlisted food ingredients to mutagenicity testing. Thetables were actually reversed; it was we the suppor-ters of testing who were taking the risk by testing foodsubstances that might prove to be beyond reproach bymutagenicity test findings. Quite a revelation.

The redeeming aspect of this revelation was the un-deniable kernel of inherent reason and logic withinit. After some careful consideration, I basically con-cluded that testing the test systems first was the rightapproach as did the scientists in the Branch as I un-derstood their positions. This issue will be discussedfurther under the subject of government’s authority todemand testing from the regulated industry.

One thing we in the Branch knew for certain wasthat the studies would be carefully reviewed and au-dited by the Branch and the results would be reportedhonestly without undue or inappropriate influence

from industry or consumer groups. As it turned out,the results of testing the GRAS-listed substanceswere negative, establishing that the tests were notoverly prone to generating false positive results. Whatcould not be determined was whether the tests weresensitive enough to detect weak mutagens. Or, statedanother way, were the tests sensitive enough to pro-tect consumers adequately? The theory behind thetests was solid and, as previously stated, they werereasonably practical in both cost and time to con-duct, particularly when compared to the high costsof carcinogenicity testing in rodents. The three testswere cytogenetic analysis in mammalian cells [1], thehost-mediated assay in the mouse using Salmonellatyphimurium and yeast as indicator organisms [2],and the dominant lethal test in the rat [3].Confidence that the tests were relevant was based on

theory which was all that was then available to supportthe approach. The theory supporting the use of cytoge-netic testing is that it involves the use of mammaliancells, including human cells, and is capable of detect-ing chromosomal abnormalities such as those knownto occur in humans, especially translocations and ane-uploidy. Further, general damage to the genome maybe reflected as chromosomal breaks and gaps which,while not mutations per se, may correlate with actualdamage to cellular DNA. The host-mediated assaywas seen as bridging the gap between testing poten-tial mutagens in simple microorganisms as comparedto testing in mammals. The test begins by injectingindicator organisms (such as S. typhimurium or yeast)into a host animal (usually into the peritoneum) priorto administering the potential mutagen by a differentroute than used for the indicator organism. After asufficient period, the indicator organism is withdrawnand the mutation frequency is assessed. The method,in theory, provides for both metabolic activation anddetoxification of the test substance and/or its metabo-lites. One drawback is that the assay does not addressDNA repair mechanisms of mammals which can havemajor effects on the mutagenic outcome of exposuresto mutagens. In the dominant lethal test in rats, thetest substance is administered to male rats, which arethen mated with groups of untreated females. Domi-nant lethal mutations are then measured by countingthe number of early fetal deaths (post-implantationloss) and by reduction in the number of implantedconceptuses in the uterus compared to control



females. Given that the apparent frequency of domi-nant lethal mutations among human births is high, thetest in rodents would appear to be relevant to humansand, unlike the other two tests, utilizes germ cells.

Thus, the tests used to examine hundreds of GRASsubstances which are permitted for addition to foodwere well supported by general theory, but the realquestion was whether they actually worked. Were theysensitive enough in detecting mutations that occur inhumans? Do they work well enough to be useful inprotecting consumers from inadvertent exposure tomutagens? We did not have answers to these fun-damental questions and yet we were inundated withquestions from inside FDA. While these were some-times exasperating they helped keep us thinking andlooking for answers and better approaches. A sampleof these questions follows:

1. Since some mutations are good mutations shouldwe try to prevent them?

2. Do we not need new mutations for the human raceto adapt to changes in the environment which arecertain to come in the future?

3. What increases in mutation rate in humans are ac-ceptable and what is an unacceptable increase?

4. Can any kind of mutagenicity testing be justifiedwithout knowing what increases in mutation rateare unacceptable?

5. How can simple in vitro tests help if they cannotprovide evidence concerning the degree or magni-tude of the effect on the mutation rate?

6. If all tests in a battery of tests are negative withonly one exception, can it ever be concluded thatthe substance is not positive?

7. Is there a concern that test batteries may grow insize and thereby increase the false positive rate fortest chemicals?

8. Considering the dominant lethal test is an invivo test utilizing germ cells for which classicno-observed effect levels can be established as inother toxicity tests, should it not be regarded asthe definitive test overriding or even making theother tests meaningless from a safety perspective?

The answer given to questions 1 and 2 was thatwe currently have all the “good” mutations we arelikely to need for selection for the next thousand yearsor more—do not worry about having enough “good”mutations, and if you want to know more, read these

papers by population geneticists (this response seemsarrogant in retrospect). Furthermore, the answer wenton to say that we are concerned about increases in themutation rate because some of those mutations willbe harmful and some will accumulate in the humangene pool to the detriment of the species. Whetherthe answers were satisfactory or not is not known, butthankfully questioning on this subject did seem to end.

The answer to question 3 was simply that we didnot know, but certainly any increase approaching orexceeding a doubling of the background rate seemedexcessive and totally unacceptable. Later on, in 1975,this question was addressed by Committee 17 ofEMS chaired by John W. Drake [4] which came tothe following conclusion: “In general no mutageniccompound should be distributed unless it serves atruly useful purpose and unless no efficacious sub-stitute is available. We recommend specific limitsfor environmental distribution of mutagenic agents,including both ionizing radiations and chemical com-pounds, such that the resulting genetic damage doesnot exceed a 12.5% increase over the spontaneousmutational background.” Again, whether the an-swer was satisfying or not, questions on this topicceased.

The answer to questions 4 and 5, based on theconclusions and recommendations of Committee 17,is that only those mutagens offering unique benefitswould require specific knowledge about the degreeto which exposure from intended use would increasethe human mutation rate. For all other substances,establishing their mutagenic potential might be suf-ficient knowledge to limit consumer exposure. Thebest approach under this philosophy would be to usethe cheapest, fastest, most reproducible, most com-prehensive and reliable tests even if they were in vitrotests in simple organisms. Whether they agreed ornot, I do not recall anyone at FDA arguing the point.Doubtlessly, discussions were held to which I wasnot privy but there is no evidence to suggest that theyresulted in any new policy on mutagenicity testing offood ingredients within the Agency.With regard to question 7, we were definitely con-

cerned that test batteries could become so large asto suffer from increased false positive error rates. Itcaused us to rethink the battery approach and to re-focus our testing efforts using a tier system approachto mutagen testing [5] as will be further discussed.


Again we encountered no real disagreement from ourFDA colleagues.

Addressing the underlying concept of question 8was difficult and, to an extent, remains so to this day.The job of a safety assessor is to determine the level ofexposure at which a substance is safe. When it is onlypossible to determine whether mutagenic potential ex-ists or not, only half the job is being addressed. Aspotential mutagens are not uncommon in nature, canarise from normal human metabolism and can evenbe produced from cooking food, it is desirable to as-sess their quantitative effects so that control efforts areaimed at the most critical exposures. Ultimately, forreasonable decision making, we need to know the de-gree and extent to which exposure to these mutagensis elevating the human mutation rate. But the viewthen was that we needed to learn much more aboutthe nature and distribution of mutagens in food andin the environment, in general, before insisting thatmutagens be subject to testing capable of ascertainingthe degree to which their exposures were elevating themutation rate. As for the dominant lethal test, otherissues had arisen casting doubt on its value in futuretesting paradigms.

After becoming fully acquainted with the innocu-ous nature of the GRAS listed substances that weretested, we were not surprised that none were found tobe positive. Nevertheless, we found ourselves ques-tioning whether the battery approach with cytogenet-ics, host-mediated assay and the dominant lethal testwas appropriate. Bridges [6] seemed to have the an-swer. He wrote that testing each new chemical intro-duced into the environment with every available testsystem is an impossible task and that, to be effective,an understanding of priorities needs to be built intoany recommended approach. He proposed instead atier system approach.

We followed up on his suggestion, adapting hisideas to our problem with food ingredients in a pre-sentation to the First International Conference on En-vironmental Mutagens, Asilomar, California [5]. Weagreed with Bridges that the battery of three tests wehad used was not an efficient means of identifyingmutagens in food. We also noted that all three testssuffered from an inability to detect certain classes ofmutagens, e.g. mutagens that act by intercalation ofDNA or require metabolic activation. Even more trou-bling was the finding that the dominant lethal test

failed to detect several classes of chemical mutagens[7] while generating inconsistent results for severaleconomically important substances, leading to con-troversy over whether the substances were positiveor negative [5]. We used the opportunity of the In-ternational Conference to announce our intentions tochange both our testing approach from one of using abattery of tests to using a tier system of tests and ourintention to change individual tests as well.The first tier was referred to as a prescreen, and

it would need to be sufficiently sensitive and com-prehensive to identify qualitatively any class of mu-tagens, rapidly and cheaply. Furthermore, its rate offalse positives would need to be low. Ultimately, werecommended the Ames test [8] for gene mutationsand diploid yeast for chromosomal damage [9] withmammalian metabolic activation [10]. In retrospect,we might have considered the induction of mitoticrecombination in yeast as an interesting genetic end-point in its own right but focused on mammalian cellsfor the primary assessment of chromosomal damage.

The second tier of testing was designed to iden-tify and confirm that the presumptive mutagen fromtier one is truly mutagenic in animals. Testing wasto include both mitotic and meiotic cells. Gene mu-tations in mammalian cells grown in culture withmetabolic activation would constitute one opportunityfor tier two testing. Additional tests in Drosophila forboth gene and chromosomal mutations could providedefinitive information about the nature of inducedgenetic events in complex, multicellular organisms.The heritable translocation test in the mouse, whichhad been developed by Generoso [11], was identifiedas holding promise for use in the second tier.

The third tier would be reserved for those substancesthat are either impossible to remove from the environ-ment or considered impossible to do without, neces-sitating risk/safety evaluation in an effort to quantifymutagenic risk from human exposure using explicitgenetic tests in mammals. Invoking tier three testingwas not expected to occur often because of the costsand time involved. The tests that might be utilizedfor this tier were an expanded heritable translocationtest involving thousands of F1 progeny and the spe-cific locus test developed by Russell [12]. It was alsopointed out that we needed new mammalian tests capa-ble of measuring gene mutations reliably with greaterease, efficiency, and sensitivity. The development of



transgenic animals for mutagenicity research can beconsidered a step in this direction [13,14].

Over the next 3–4 years, the Branch undertook aprogram similar to the tier system as described to retestmany of the GRAS substances plus some other chemi-cals. Once again, the results of mutagenicity testing ofGRAS substances were negative, which again was notsurprising. A small number of non-GRAS substanceswere positive in tier one and subsequently tested intier two. None got to the final tier three stage. In timethe agency became even more convinced that qualita-tive identification of chemical mutagens would meet99% of their regulatory needs and slowly let go ofthe idea that establishing no observed effect levels formutagens would be required.

Over time, the agency began to consider mutagenic-ity testing more as a prescreen for carcinogens thanas a means of detecting substances which may pose agenetic risk to the human gene pool. However, giventhe risk-averse conservatisms applied to food-bornecarcinogens, particularly if they are added substances,it seems likely that mutagenic risk is adequately ad-dressed by efforts to control carcinogenic risks. Inthis regard, it is noted that FDA guidelines for testingfood ingredients recommends comprehensive testingencompassing all known types of mutations of con-cern in humans. Thus, it can be argued that as long astesting for mutagenicity is adequate, it does not mat-ter whether its stated purpose is to protect consumersfrom carcinogens or to protect against an increasein the rate of human mutations. The consumer willbe protected against both. This seems to be a goodargument, but is it true? Should we not consider pub-lic discussion of this matter, setting up workshops orpossibly a symposium to determine whether there isor is not good agreement among geneticists and otherscientists that the cancer screens are sufficient to pro-tect against unwitting increases in mutations due tochemical exposures?

To FDA’s credit, making good use of new de-velopments in mutagenicity research to improve theguidance given to the industry on food ingredienttesting has been ongoing since Marvin Legator waschief of the Genetic Toxicology Branch. Neverthe-less, it should be recognized that the dynamics ofmixing science (research) with regulatory policy isa uniquely bumpy road—as well it should be, givenits potential for both good and harm. Regulations are

derived from laws passed by the Congress and signedinto law by the president, e.g. the FDCA. Such lawsand their individual provisions or the regulations de-rived from them can be struck down by the courtsif found to be in violation of the Constitution of theUnited States which places severe limits and con-straints on government’s authority over its citizens.Thus, a government agency (like the FDA) must becertain that any burdens placed on the regulated (inthis case, the food and food-ingredient industries) canbe justified on constitutional grounds. Constitutionalgrounds for requiring mutagenicity testing by the foodand food-ingredient industries demand a showing thatsuch testing is necessary to protect the rights of theconsumer. Such a showing requires strong scientificconsensus that a clear need exists to protect consumersfrom potential mutagens in the food supply and thatthe proposed testing guidance is scientifically appro-priate. Additionally, the tests must be practical andreasonable in costs and availability so that they do notbecome, or are not held by the courts to be, a de factoprohibition on the use of food ingredients. There are,of course, other factors that place additional bumpsin the road, such as congressional oversight hearings,objections raised by so-called advocacy groups, me-dia attention, and objections raised by the regulatedin administrative procedures. It can be frustrating butthe system does work—though slowly.

In looking back on developments leading up tomutagenicity guidelines for food ingredients, I amimpressed with the role played by the EMS, other sci-entific societies and the many scientists who partici-pated in numerous meetings in which the tests and theapproaches were discussed. This needs to be a con-tinuing process to ensure that new developments andnew knowledge are carefully considered with respectto the current status of FDA’s recommendations to theregulated industry. Making certain that the guidanceand requirements of regulatory agencies are foundedon current science is not only desirable but necessaryunder the law in the US and many other countries.

References

[1] S. Green, A. Auletta, J. Fabricant, R. Kapp, M. Manandhar,C.J. Sheu, J. Springer, B. Whitfield, Current status ofbioassays in genetic toxicology—the dominant lethal assay: a


report of the US Environmental Protection Agency Gene-ToxProgram, Mutat. Res. 154 (1985) 49–67.

[2] M.S. Legator, E. Bueding, R. Batzinger, T.H. Connor, E.Eisenstadt, M.G. Farrow, G. Ficsor, A. Hsie, J. Seed, R.S.Stafford, An evaluation of the host-mediated assay and bodyfluid analysis: a report of the US Environmental ProtectionAgency Gene-Tox Program, Mutat. Res. 98 (1982) 319–374.

[3] R.J. Preston, W. Au, M.A. Bender, J.G. Brewen, A.V. Carrano,J.A. Heddle, A.F. McFee, S. Wolff, J.S. Wassom, Mammalianin vivo and in vitro cytogenetic assays: a report of the USEPA’s Gene-Tox Program, Mutat. Res. 87 (1981) 143–188.

[4] J.W. Drake, S. Abrahamson, J.F. Crow, A. Hollaender, S.Lederberg, M. Legator, J.V. Neel, M.W. Shaw, E.H. Sutton,R.C. von Borstel, S. Zimmering, F.J. deSerres, W.G. Flamm,Environmental mutagenic hazards, Science 187 (1975) 503–514.

[5] W.G. Flamm, A tier system approach to mutagen testing,Mutat. Res. 26 (1974) 329–333.

[6] B.A. Bridges, Some principles of mutagenicity screening anda possible framework for testing procedures, Environ. HealthPerspect. 6 (1973) 221–226.

[7] S.S. Epstein, E. Arnold, J. Andrea, W. Bass, Y. Bishop,Detection of chemical mutagens by the dominant lethal assay

in the mouse, Toxicol. Appl. Pharmacol. 23 (1972) 288–325.

[8] B.N. Ames, W.E. Durston, E. Yamasaki, F.D. Lee,Carcinogens are mutagens: a simple test system combiningliver homogenates for activation and bacteria for detection,Proc. Natl. Acad. Sci. U.S.A. 70 (1973) 2281–2285.

[9] D.J. Brusick, V.W. Mayer, New developments in mutagenicityscreening techniques using yeast, Environ. Health Prespect.6 (1973) 83–96.

[10] H.V. Malling, Dimethylnitrosamine: formation of mutageniccompounds by interaction with mouse liver microsomes,Mutat. Res. 13 (1971) 525–529.

[11] W.M. Generoso, Evaluation of chromosomal aberration effectsof chemicals on mouse germ cells, Environ. Health Perspect.6 (1973) 13–22.

[12] W.L. Russell, X-ray induced mutations in mice, Cold SpringHarbor Symp. Quant. Biol. 16 (1951) 327–336.

[13] J.C. Mirsalis, J.A. Monforte, R.A. Winegar, Transgenic animalmodels for measuring mutations in vivo, Crit. Rev. Toxicol.24 (1994) 255–280.

[14] J. Vijg, H. van Steeg, Transgenic assays for mutations andcancer: current status and future perspectives, Mutat. Res.400 (1998) 337–354.



Mutation research at ABCC/RERF: cytogeneticstudies of atomic bomb exposed populations ☆Akio Awa

Cytogenetics Laboratory, Department of Genetics, Radiation Effects Research Foundation, 5-2 Hijiyama Park, Minami-ku, Hiroshima 732-0815, JapanAccepted 20 September 2002



Mutation research at ABCC/RERF: cytogeneticstudies of atomic bomb exposed populations�

Akio AwaCytogenetics Laboratory, Department of Genetics, Radiation Effects Research Foundation,

5-2 Hijiyama Park, Minami-ku, Hiroshima 732-0815, Japan

Accepted 20 September 2002

Keywords: Atomic bomb; Cytogenetics; Radiation risk assessment; Human population monitoring; Hiroshima/Nagasaki; ABCC; RERF

1. Introduction

The decade immediately following 1953 must havebeen the most celebrated and exciting period in biol-ogy and medicine in the whole of the 20th century. Nobiologist is likely to disagree with my view that oneof the very finest of all biological achievements wasthe work of Watson and Crick [1] that led to the publi-cation in 1953 of the concise but epoch-making paperthat described the double-helical structure of the DNAmolecule. Many of us have been fortunate enough tohave lived through both that famous discovery and theless than half a century since then that it has takenresearchers to determine what Presidents and PrimeMinisters have described as “the complete DNA se-quence of the human genome.”

Perhaps surprisingly, it was not until some 3 yearsafter the nature of the DNA molecule had been de-termined that Tjio and Levan finally established to(almost) every one’s satisfaction that the correct chro-mosome number for man is 46 [2]. These two investi-


1 E-mail address: [email protected] (A. Awa).

gators were only able to reach this conclusion becauseof the special facility with which they were able toapply advanced tissue culture techniques to some ofthe most extraordinarily careful studies of mammalianchromosomes conducted in that era. Their work andthe techniques they developed led to a tremendousexpansion in the relatively new field of human cy-togenetics. I found this an extraordinarily excitingtime to be a postgraduate student in mammaliancytogenetics.

Even in these early years, it was recognized thatthe estimation of radiation hazards to human popu-lations was of enormous importance. As Bender [3]pointed out, although there was an enormous body ofdata on the chromosome damage inflicted by radia-tion on non-mammalian organisms such as fruit fliesand plants, there were several important technicaldifficulties that had to be overcome before we couldexpect to obtain very much data on human subjects.With this thought in mind, Bender [3] devised ra-diation experiments with epithelioid diploid humankidney cells in which he employed some relativelynew methods for spreading chromosomes in tissueculture that Hsu and Pomerat had just developed [4].Although he was able to examine the frequency ofX-ray-induced chromosome alterations in terms ofchromatid deletions and exchanges relative to radia-


2 A. Awa / Mutation Research 543 (2003) 1–15

tion dose administered, it was still a little too earlyfor his approach to be widely adopted as a method ofassessing radiation hazards in humans.

In 1960, Moorhead et al. [5] described a sim-ple culture method for use with human peripheralblood leukocytes, having achieved success in culti-vating human white blood cells—mostly mature Tlymphocytes—by adding small amounts of phyto-hemagglutinin (PHA, an extract of kidney bean) toa suitable culture medium. Surprisingly, large num-bers of lymphocytes appeared to enter the first invitro mitosis in synchrony some 48 h after the culturehad been initiated, with a second wave of mitosistaking place after the cells had been in culture for atotal of 72 h. Discovery of this technique has had anenormous impact on research progress in all fields ofhuman cytogenetics, especially those pursued by themany subsequent investigators who made strenuousefforts to develop chromosomal mutation assays foruse in estimating the risks of radiation (as well as in-numerable other environmental mutagens) to humans.

Also in 1960, Tough et al. reported the results of apreliminary study which they conducted in Edinburghon the persistence of gross chromosome damage inthe peripheral blood lymphocytes of patients whohad previously received therapeutic X-ray treatmentfor ankylosing spondilitis [6]. This study was laterextended by Buckton et al., who confirmed one ofthe major findings of the Tough et al. study, namelythat radiation-induced structural rearrangements ofchromosomes are able to persist in circulating bloodlymphocyte populations for several years after therelevant exposure [7]. Tough et al. classified theradiation-induced chromosome damage that they hadobserved into the following two classes: (1) unstablechromosome aberrations, which included dicentricchromosomes, ring chromosomes, and acentric frag-ments (traditionally referred to as asymmetrical ex-changes of chromosomes), and (2) stable chromosomeaberrations, which included reciprocal translocationsand inversions (symmetrical exchanges).

A little later in the same year as the Buckton et al.(1962) report appeared, Bender and Gooch [8] pub-lished an analysis of chromosome aberrations in thecirculating lymphocytes of several people who hadbeen exposed to a mixture of gamma and neutronradiations at the time of the Y12 criticality accidentsin Oak Ridge, Tennessee. Their conclusions were

very similar to those of Tough’s group in Edinburgh,in that they too found that lymphocytes carryingradiation-induced aberrations were perfectly capableof persisting in the peripheral blood for very longperiods, and that all increases in the frequency withwhich chromosome aberrations could be detectedappeared to be dose-dependent.

At about this time, the Atomic Bomb CasualtyCommission was making its very important deci-sion to establish a new laboratory in which a majornew cytogenetic project would be begun with theexpressed aim of developing sensitive methods forassessing the effects of atomic bomb radiation onthe survivors of Hiroshima and Nagasaki who wereunfortunate enough to have been exposed to the dev-astating atomic bomb explosions that occurred overHiroshima on 6 August 1945 and over Nagasaki afew days later (9 August 1945). The remainder ofthis memoir will be concerned with my recollec-tions of this famous, and already historic, cytogeneticproject.

2. ABCC and RERF

The Atomic Bomb Casualty Commission (ABCC)was established in 1947 in Hiroshima, and in 1948 inNagasaki, by the US National Academy of Sciences/National Research Council (NAS/NRC) in responseto President Truman’s directive to begin a long-termand comprehensive epidemiological and genetic studyof the atomic bomb survivors. The Japanese NationalInstitute of Health under the Ministry of Health andWelfare joined with the ABCC in the initiation of thesestudies from their beginning in 1948 and helped toensure that they would be continued until the presentday (and hopefully for some years to come).

In 1975, the Radiation Effects Research Founda-tion (RERF) was set up to replace and assume theresponsibilities of the ABCC. RERF is a non-profitJapanese research foundation which is binationallymanaged and funded equally by the governments ofJapan (through the Ministry of Health and Welfare)and the US (through the NAS/NRC, under contractto the US Department of Energy and its various pre-decessor departments). The research objectives ofRERF as described in its Act of Endowment are asfollows:


A. Awa / Mutation Research 543 (2003) 1–15 3

To conduct research and studies, for peaceful pur-poses, on the medical effects of radiation on manand on diseases which may be affected by radi-ation, with a view to contributing to the mainte-nance of the health and welfare of atomic bomb sur-vivors and to the enhancement of the health of allmankind.

An important characteristic of the people who wereexposed to the Hiroshima and Nagasaki A-bombs isthat their individual radiation doses necessarily var-ied over a very wide range, depending primarily onthe distance between the burst point of the bomb andthe location of each individual survivor, and addition-ally (and very importantly) on the form and extent ofany shielding that they were lucky enough to receive.This has meant that one of the key research objectivesat ABCC/RERF necessarily has involved investigatorsin carefully planned attempts to determine the extentof the damage, and the nature of any health effectsthat may have resulted from that damage, in relationto each individual A-bomb survivor’s estimated radia-tion dose. One consequence was that four key platformprotocols were developed and have underpinned re-search at ABCC/RERF since the very early days. Theyare: (1) the life span study (LSS), (2) the adult healthstudy (AHS), (3) the Pathology Study, and (4) the Ge-netic Study. Unfortunately changing circumstances inthe RERF workforce made a decision to discontinuethe Pathology Study inevitable by the middle 1980s,but it seems likely that the others will continue wellinto the foreseeable future.The LSS population consists of a fixed cohort of

some 120,000 survivors who were listed in the 1950national census. Information concerning the numbersof deaths that occur within this group, and about theircauses, is obtained on a periodic basis from the na-tional death certificate system in Japan. In more re-cent times, considerable emphasis has been placed onrecording cancer incidence among A-bomb survivors.The AHS cohort was enrolled from within the LSS andconsists of some 20,000 survivors who were willing toundergo biennial clinical health examinations. A hugebody of information about all of these volunteers, andin particular about the incidence and prevalence of awide variety of diseases within this exceptionally largecohort of regular clinic attendees, has been collected,stored and analyzed over the last several decades.

The Genetic Study population consists of some80,000 children, of whom approximately one-halfwere born after May 1946 in Hiroshima or Nagasakito parents one or both of whom had been exposed toA-bomb radiation. The remaining one-half consists ofa control group of children whose parents were stillwithin the city at the time of the bombing or were quitedistally exposed (>3000 m from the hypocenter). I donot intend to go into the details of this particular studybut instead strongly recommend interested readers tothe first-rate book published by Neel and Schull [9].

Anyone who is planning to make a serious attemptto evaluate and quantify the health effects of the ion-izing radiation associated with the Hiroshima and Na-gasaki A-bombs must first obtain ready access to anaccurate dosimetry system with which radiation dosesfor individual survivors can be estimated as accuratelyas possible. Achieving this is by no means easy, how-ever. There are numerous reasons for this task beingan extremely difficult one, and I feel it is worthwhilespending a little time discussing some of the issueswhich have emerged in this vitally important area.First, it is important for readers to be aware that theatomic bombs dropped on Hiroshima and Nagasakiwere very different in their composition. The one re-leased over Hiroshima (little boy) was a gun-type ura-nium bomb whereas the one released over Nagasaki(fat man) was an implosion-type plutonium bomb.Second, while the world has had considerable experi-ence in measuring the output of Nagasaki-type bombs,the gun-type uranium bomb used on Hiroshima ap-pears to have been unique; students of dosimetry aretherefore forced to rely on theoretical models and thecalculations derived from them in their attempts toderive the best possible dosimetry estimates that areever likely to be available for use in radiation effectsstudies. It has however been very well established thatboth bombs were responsible for releasing mixtures ofneutron and gamma-rays, and that the radiation dosescontributed by the latter were far larger than those em-anating from the former. One further problem that isof particular relevance to estimation of neutron dosesconcerns the effective absorption of fast neutrons bythe hydrogen ions in water vapor. This would meanthat the neutron doses as one moves away from thecenter of the A-bomb explosion could have been sig-nificantly reduced if the humidity levels were high, asituation that is not at all uncommon in western Japan


in summer. Unfortunately, however, there is no reliableinformation at all about the exact levels of humidityon these 2 fateful days in August 1945.

Over the years there have been many valiant effortsto establish dose estimate systems that are both reli-able and practical. These efforts would not have beennearly as successful as they were had it not been forthe close cooperation of international groups of physi-cists over many years and through many long andcomplicated workshops and conferences. A first, ten-tative, dosimetry system was developed in the 1950s,and has been refined and upgraded on several occa-sions since then. Two early systems, known as T57D(tentative dose system 1957) and T65D (an updatedversion of T57D which was released in 1965), wereadapted for use in studies of the A-bomb population,and remained in use at ABCC/RERF for relativelylong periods of time. The dosimetry system in use atthe time of writing is known as Dosimetry System1986, or more simply DS86, and only became avail-able after an enormous amount of painstaking workhad been conducted over a period of two decades ormore [10]. A thorough revision of DS86 is said to benearing completion as I write, and is believed to in-clude small but potentially important changes to theneutron dose estimates for both cities.

3. Chromosome aberration studies involvingparticipants in the adult health study (AHS)population

3.1. Data obtained using the conventionalGiemsa-staining method

As mentioned previously, a useable methodologyfor conducting cytogenetic analyses of radiation ef-fects became available during the early 1960s. It wasat about this time that the ABCC was beginning toassemble a new group whose task was to initiate alarge-scale cytogenetic study program, and by 1965a new laboratory had been set up in the Departmentof Clinical Laboratories under the overall supervisionof Dr. Howard Hamilton. That same year, Dr. ArthurBloom (who had been assigned to the ABCC as a pub-lic health surgeon) joined the laboratory. Dr. Bloomhad been expertly trained in the latest human cytoge-netic techniques by Dr. J.H. Tjio, one of the two people

whose 1956 report that the true human chromosomenumber is 46 quickly gained universal acceptance.

Two major cytogenetic projects were promptly ap-proved by the ABCC almost immediately followingDr. Bloom’s arrival in Hiroshima. One was a study ofradiation-induced chromosome damage in the somaticcells of A-bomb survivors, while the other focused ona careful evaluation of the genetic effects of atomicbomb radiation on the children of survivors. The latterproject was of course intended to determine whetherit might be possible to observe any germ-line chromo-somal mutations that may have arisen as a result ofparental radiation exposure.

Dr. Bloom and his colleagues then embarked on anew large-scale cytogenetic examination of heavily ex-posed A-bomb survivors whose estimated T65D doseswere in excess of 2 Gy. The matched controls in thisstudy were either exposed or non-exposed local resi-dents with estimated doses of less than 0.01 Gy, andall of the subjects in the study were recruited fromwithin the AHS population. The results and analyseswere summarized within 2 years in two papers in TheLancet, one of which was on the chromosome aberra-tion data obtained from survivors who were less than30 years of age at the time of the bombing (ATB) [11],while the second contained data from those who weremore than 30 ATB [12]. The single most importantfinding was that structural aberrations of the exchangetype, mainly consisting of dicentrics, rings and grossaberrations of the translocation type, seemed to occurat significantly higher frequencies in survivors fromthe exposed groups than in the controls who were ex-posed to <0.01Gy. Similar findings were obtained inboth age groups in both cities.

Soon after this screening program got under way,Bloom recognized that there were some importantlimitations in his existing laboratory facilities and inhis initial research staff. He therefore contacted Pro-fessor Sajiro Makino, a leading Japanese human cyto-geneticist based at Hokkaido University, in December1966 to try and arrange for some joint work withHokkaido University which he believed would helpto strengthen the programs that were currently underway at the ABCC. It was agreed between Drs. Bloomand Makino that Hokkaido University would supportthe ABCCs cytogenetic programs by arranging forseveral of Dr. Makino’s research staff to join theABCC laboratory. Immediately after this agreement



had been reached, four research associates were as-signed as permanent staff to the ABCC. Three ofthem (Takeo Honda, Toshio Sofuni and myself) wereassigned to the Hiroshima laboratory, and MichihiroYoshida went on to Nagasaki.

In the summer of 1968, Dr. Bloom returned to theUS after 3 years on assignment to the ABCC to joinDr. James V. Neel’s famous Department of Human Ge-netics at the University of Michigan Medical Schoolin Ann Arbor. Arthur Bloom’s name will long be re-membered as one of the key people in the early daysof the ABCC/RERF Cytogenetics Laboratory. AfterArthur’s departure for the US, I was invited to takeover responsibility for the Laboratory, and I did so,continuing for some 25 years until my own retirementat the end of 1993. My replacement as Chief of theLaboratory was Dr. Nori Nakamura, who holds theposition to this day.

When I took over as team leader of the cytogeneticprojects, the research staff agreed with an early sug-gestion that we needed to add a great many additionalexposed and non-exposed Hiroshima residents to anew and more comprehensive study which we hopedwould allow us to determine the most likely rela-tionships between radiation dose and chromosomeaberration frequency in the A-bomb survivor popu-lation. We therefore began selecting study subjectsfrom among the entire AHS population, using theT65D dose estimates that had been assigned to eachsurvivor as a way of constructing a suitable rangeof exposure groups. Before embarking on this—tous—very large-scale program of cytogenetic analysis,however, we recognized that neither our lymphocyteculture techniques nor the quality of the finished mi-croscopic slides in our laboratory were as good asthey ought to be. Sometimes, for example, we foundit difficult to score more than 100 metaphases persample in the course of our daily routine. Also, bythis time more than 20 years had elapsed since the Hi-roshima and Nagasaki A-bombings of 1945, and westrongly suspected that the majority of cells carryingunstable chromosome aberrations (dicentrics, rings,and acentric fragments) that had been produced atthe time of bombings might have already disappearedfrom the circulating blood of many (if not most) ofthe survivors who were originally affected.

In 1968, Sasaki and Miyata published a paper thatdealt with chromosome aberration analysis on 51 Hi-

roshima atomic bomb survivors and 11 unirradiatedhealthy donors chosen as controls [13]. These workersmainly scored dicentrics, rings and acentric deletionsoccurring in an average of about 1500 metaphases perindividual. Based on the resulting aberration frequencydata, they estimated radiation doses for the individualsurvivors using a biodosimetric procedure known asthe Qdr method. Their results indicated that there wasa reasonable measure of agreement between their bi-ological end-points and physical dose estimates, withthe latter being heavily dependent upon distances fromthe hypocenter and on the presence and nature of anyshielding materials that could be identified betweenthe A-bomb burst point and an individual survivor.This 1968 study by Sasaki and Miyata was very muchin line with the then internationally-accepted practiceof relying upon asymmetrical exchanges (sometimesalso described as unstable chromosome aberrations)for the detection and quantification of prior radiationexposures [14–17].

Not long afterwards (in 1969), it was pointed out ina report of the United Nations Scientific Committeeon the Effects of Atomic Radiation (UNSCEAR) thatthe “efficiency of scoring symmetrical events (stablechromosome aberrations) in human chromosomes isnot more than 20%. This follows from the fact thatthe frequency of dicentric plus centric-ring aberrationsshould equal the frequency of reciprocal translocationsplus pericentric inversions, whereas dicentric and ringchromosomes are approximately five times as frequentas abnormal monocentric chromosomes” [14].

This statement was undoubtedly correct. However,I felt then, and still feel today, that the major diffi-culty most of us have in the detection of symmetricalexchange (or stable chromosome) aberrations stemsfrom the following:

(1) In the case of symmetrical interchange aberra-tions, exchanges between points equidistant fromthe telomeres of the two chromosomes involvedare frequently not observable. Such symmetricalexchanges do not visibly alter the morphology ofthe affected chromosomes, and so any aberrationsthat result will almost certainly evade scrutiny.

(2) Generally speaking, chromosomes that belongto the same chromosome group are similar withrespect to length and shape (arm ratio). Thus it isalways going to be much more difficult to identify


symmetrical exchanges between chromosomesfrom any one group than between chromosomesfrom different groups. The difficulties are alsolikely to be greatest for exchanges between pairsof C group chromosomes, where there are toomany chromosomes that closely resemble oneanother for any type of visual discrimination tobe even remotely likely. In addition to all ofthis, it is now widely known that the efficiencieswith which symmetrical exchange aberrations aredetected tend to vary quite markedly between ob-servers, presumably because the favored detectioncriteria of any one individual are almost certainto differ from those of any other individual.

Although I had absolutely no doubt that Sasakiand Miyata’s paper was of the highest quality, andthat it would be of considerable practical value inour efforts to carry out effective quantitative evalua-tions of the events that might be expected to occur inlarge numbers of radiation-exposed humans, I was byno means confident that our laboratory staff and re-sources would be adequate for us to incorporate theircarefully polished protocols into our then establishedroutine protocols. My most obvious concern was thatif we continued to obtain relatively small numbers ofanalyzable metaphases per sample, we might neverbe able to generate enough reliable data to detectdose-response relationships for unstable chromosomeaberrations among the large numbers of Hiroshimaand Nagasaki survivors whose blood samples hadbeen so willingly donated on trust.

After a great deal of careful consideration, wedecided to settle for a procedure involving a totalof 100 well-spread metaphases per person (a num-ber we now knew we could definitely achieve on aroutine basis), all of which would be carefully exam-ined for all discernible types of stable and unstablechromosome aberrations. We were therefore mak-ing a conscious decision to move away from usingfrequencies of unstable chromosome aberrations asthe sole biological indicator for the assessment ofdose-response relationships, even though we werewell aware that this would not be in accord with thethen internationally-accepted methodology for radia-tion biodosimetry (cf. [14–17]). We also knew that re-searchers who scored stable chromosome aberrationsin order to evaluate radiation effects were few and far

between at this time, if indeed there were any others.Thus we were well aware that we were going out on alimb, but we were privately convinced that we wouldbe able to demonstrate to the satisfaction of our peersthat careful scoring of stable chromosome aberrationswould prove to be just as effective as careful scoringof unstable aberrations for the purposes of radia-tion dose-response relationship assessment. Based onour experience while preparing for the main project,we knew that we would almost certainly be able todetect an average of around five cells that carriedstable chromosome aberrations per 100 metaphasesper 1 Gy in the A-bomb survivor population. We hadalso noticed that the frequencies of translocationsplus pericentric inversions in A-bomb survivors wereabout 10-fold higher than the frequencies of dicentricplus ring chromosomes that we could detect.Some difficult questions remained to be answered,

however. To my mind the most important of these waswhether we would ever be able to make reasonableestimates of the radiation doses received by individualsurvivors solely by determining the frequencies of sta-ble chromosome aberrations in blood samples takensome 20 or more years after the relevant radiation ex-posure. If we were to answer this question with anyconfidence, I believed it would be essential to validatethe following propositions:

(1) that radiation-induced breaks and illegitimate re-joining of the broken ends occur equally and atrandom along the entire length of every chromo-some, which means that their frequency will beproportional to either a unit length of chromosomeor a unit sequence of DNA;

(2) that asymmetrical (dicentric with accompanyingacentric fragment) and symmetrical (reciprocaltranslocation) interchange events occur with equalfrequency, and that a similar relationship willhold for intrachange events (ring with acentricfragment versus pericentric inversion);

(3) that the chromosome aberrations detected in pe-ripheral lymphocytes decades after exposure toA-bomb radiation were in all probability producedin the relevant lymphoid stem cells;

(4) that because stable chromosome aberrations weregenerally thought unlikely to lead to deficienciesin the host cell’s chromosomal material, theywould remain viable albeit aberrant, and ought



not to encounter any mechanical disturbance insubsequent cell divisions and so should remainpresent through many cell generations after theirinduction by the relevant radiation exposure;

(5) cells with unstable aberrations would almost cer-tainly be lost with the passage of time.

We attempted to confirm each of these assumptionsas we were progressing through the main study, and bythe end we felt that none of them were unreasonable.However, in the course of such a large-scale study, ourmost important task by far was to ensure that everysingle microscopist maintained his or her ability todetect stable chromosome aberrations at a level thatwas as consistent as it was possible to be. We insistedthat inter-observer differences in aberration detectionrates were minimized at all times, and, if humanlypossible, eliminated. Immense efforts were made tomeet this objective.

According to the International System for HumanCytogenetic Nomenclature [18], non-banded humanchromosomes can be classified into seven groups fromA to G when aligned in descending order of size.Even with non-banded Giemsa-stained preparations,the chromosomes in each group can be identified rea-sonably accurately solely on the basis of the lengthsand arm-ratios that characterize each one.

From this point on, though, there is no simple royalroad that can be shown to anyone who hopes to finda short-cut to the accurate and reliable detection ofstable chromosome aberrations. An absolutely funda-mental requirement for the reliable detection of aber-rations is that the microscopists involved be armedwith a thoroughgoing familiarity with the normalmorphological pattern of each of the human chromo-somes; such familiarity can only be acquired in oneway, which is of course by spending long periodsof time looking directly into the microscope. Whenanalyzing a specimen, the observer should start byclassifying the metaphase chromosomes into groupsA to G by eye and determining whether there is anyindication of the loss or gain of one or more chromo-some(s) from each of the chromosome groups. If anyelement is missing from, or added on to, any of thegroups it can be assumed that the metaphase underexamination is more likely than not characterized byat least one aberrant chromosome. This all soundsterribly easy, but the work involved is not nearly as

easy as it might seem at first sight. In practice, it tookalmost a year for all members of our laboratory tobecome accustomed to the rigors of this system ofanalysis and to make consistently correct decisionsabout whether or not a metaphase under study couldconfidently be labeled “normal”.

In the course of our routine work, we took enormouscare to minimize the numbers of metaphases that werewrongly declared normal despite having aberrationspresent, i.e. the numbers of false-negatives. Althoughhumans are only human, and mistakes are inevitablein the course of any long-term large-scale screeningprogram such as ours, our assembled data would veryquickly have highlighted any excessive inter-observervariation if in fact too many aberrations were beingmissed through observer misjudgment. There can beno doubt that our credibility would have suffered enor-mously had this been occurring to any great extent.One way of overcoming this important problem whichwe identified and quickly adopted required us to takephotographs of every single metaphase in which def-inite or suspected structural aberrations had been de-tected by direct microscopy so that we could thenconduct a careful karyotype analysis on each one. Inthe course of the thousands of karyotype analyses thatwe undertook, every effort was made to determinethe origins of the chromosomes involved in every sin-gle exchange aberration that we reviewed. The proto-cols we developed have been condensed into a guidefor new players entitled: A manual for detecting sta-ble chromosome aberrations by (non-banded) conven-tional Giemsa-staining method. This manual is freelyavailable, and can be accessed via the RERF web siteat http://www.rerf.jp.

The essential items in our shared (and in prac-tice mandatory) scoring procedure were as follows:(a) ID number of the subject; (b) slide number(s);(c) cell numbers; (d) the location of the cells on theX–Y-axis of the microscopic stage; (e) the serial pho-tograph numbers; and (f) comments, if any, regardingthe observer’s judgment. When one considers the costof a frame of 35mm negative film, the addition of sucha reliable record to our routine procedure added al-most nothing to overall operating costs. As soon as webegan to use photographs, the ability of our observersto detect aberrations improved dramatically, and re-mained at a consistently high level throughout the en-tire screening program. By December 1993, we had


accumulated almost 400,000 frames of metaphases inthe 35mm film collection that is maintained in ourDepartmental film cabinet. I hope that it is obviousfrom what I have just explained that we all put a greatdeal of time and effort into improving our routine lab-oratory techniques, and that we were eventually ableto guarantee abundant supplies of metaphases withwell-delineated chromosomes from a vast majority ofthe thousands of samples we had to test.

Our chromosome slides were all coded with sequen-tial case numbers so that the entire investigation wasconducted without anyone in the laboratory knowingthe exposure status of any individual examinee. Weonly began to decode the information on individualdoses after the complete set of slides had been exam-ined. The results were entered into our institute’s maincomputer immediately after the relevant microscopicexamination was finished. Thus it was only at the verylast stage that we were able to conduct a thoroughanalysis of the data we had collected and to begin try-ing to relate our findings to radiation dose.Several important findings emerged from the inves-

tigations we carried out in the course of this study[19–21]. The main ones are as follows:

(1) The frequency with which cells with stable chro-mosome aberrations could be detected appeared toincrease with increasing radiation dose. The fre-quency of aberrations per unit dose always tendedto be somewhat higher in Hiroshima than in Na-gasaki.

(2) The dose-response curve that we obtained forthe survivor population was essentially linearfor Hiroshima residents, but much closer todose-squared for Nagasaki residents. Our find-ings that there were clear inter-city differences inboth the shapes of the dose-response curves andin chromosome aberration frequencies per unitdose have yet to be explained, although a possibleexplanation involving differences in the relativecontributions of neutron and gamma-rays to thetotal radiation doses received by residents of thetwo cities is favored by some. However, the plau-sibility of this explanation is open to review, andit is almost certain to be re-examined just as soonas ongoing efforts to update DS86 (the dosimetrysystem currently being used by RERF) has beenfinalized by an international review panel and

made available to RERF staff for application toindividual survivors.

(3) We observed that stable aberrations (recipro-cal translocations plus pericentric inversions)were more frequent than unstable aberrations(dicentrics plus rings) by nearly an order of mag-nitude, and that this was true for all dose rangesunder investigation. This clearly demonstratesthat the contribution of stable chromosome aber-rations to the radiation dose-response relationshipis of great importance. Having made this point,it is important to note that we also observed apositive dose-response relationship for unstablechromosome aberrations.

(4) We were able to identify a few A-bomb survivorswhose blood contained clones of lymphocyteswith precisely identical karyotypic changes inlarge numbers of cells. All of the aberrations inthe relevant cells were of the stable type. Al-though rare, these clonal aberrations were mostlikely to be found among survivors in the highestdose range (1 Gy and above). We believe that thegenetic alterations responsible for these clonalaberrations were most likely to have been pro-duced in lymphopoietic stem cells which wenton to proliferate and differentiate into matureT lymphocytes, each one of which would thencarry the specific aberration that characterized theindividual survivor.

(5) Interestingly, we were also able to identify a fewsurvivors in the high dose range whose aberrationfrequencies were unusually low; there were alsoa few in the low dose range whose aberration fre-quencies were unusually high. We described thesesurvivors as “ cytogenetic outliers”, and tended toregard them as over-dispersed cases that lay faroutside the confidence limits of the dose-responsecurve. We believe that many of these anomalouscases represent inadvertant errors in dose esti-mates. Certainly we were unable to find evidencein support of the most obvious alternative hypoth-esis, which was of course that they were the resultof individual-to-individual differences in radiationsusceptibility.

RERFs detailed chromosome survey of the AHSpopulation by conventional analysis was completedin 1993 after 25 years of painstaking work. The data



obtained from a total of 3042 survivors (1980 casesin Hiroshima including 1329 proximally exposed and651 controls, and 1062 cases in Nagasaki with 599exposed and 463 controls) were then analyzed to de-termine the relationship between stable chromosomeaberration frequency and DS86 radiation dose [22].The shape of the dose response turned out to be con-cave upward for doses below 1.5 Sv, but exhibits someleveling off at higher doses; this curvature is muchthe same for both cities. The slopes for the two citiesdiffered at the lower dose levels, however, being sig-nificantly higher in Hiroshima than in Nagasaki. Thisinter-city difference is smaller (but does not disappear)when comparisons are limited to survivors who wereexposed in houses. A new finding in the final report isthat Nagasaki survivors who were exposed in factoriestend to show a lower dose response than people whowere outside in Nagasaki or Hiroshima and hence hadlittle or no shielding, or even people who were exposedin typical Japanese houses. Our calculations indicatethat the doses for Nagasaki survivors who were work-ing in factories at the time of exposure may have beenoverestimated by approximately 60% using DS86.

3.2. Data obtained by G-banding

As time went by, we incorporated newly devel-oped cytogenetic techniques into our routine workand hence into our survey. When a burst of develop-ment of various banding techniques occurred in theearly 1970s, we added Q-, C-, and G-banding tech-niques to our routine laboratory studies. After makingsome modifications to the then standard techniques,we found that the trypsin-G-banding method promisedto be a very powerful tool for the detection of chro-mosome aberrations. In particular, we found that it al-lowed us to detect chromosome aberrations in all ofthe human chromosomes. We were also able to useaberration analysis by the G-banding method to val-idate the aberration frequency data obtained by theconventional Giemsa-staining method.

In one paper, for example, we reported on a de-tailed comparison of the types and frequencies of sta-ble chromosome aberrations that could be detected insamples from 23 Hiroshima survivors using both thenon-banded and G-banding methods in parallel [23].This study revealed that our conventional, non-banded,methods enabled us to detect approximately 78% of

the total numbers of aberrations that were detectedby the G-banding method at its best (i.e. when vir-tually all chromosome aberrations of the stable typewere being detected in the cells under examination). Inthis study, the numbers of aberrations that had previ-ously been detected by non-banded analysis were care-fully scrutinized. We were then able to establish thatthe aberration frequencies derived in routine screen-ing of non-banded preparations tended to be lowerthan those obtained when G-banding was used, andwe estimated that routine scoring of aberrations bynon-banded methods was approximately 70% as ef-fective as scoring by G-banding. Later comparativestudies using other techniques have demonstrated thatthis estimate is about right.

Problems inherent in the G-banding technique in-clude the fact that it takes a considerable time for ob-servers to become familiar with the banding patternsthat characterize individual chromosomes. It is alsotrue that scoring individual samples is both tedious andtime-consuming. Nevertheless, there can be no doubtthat G-banding is both accurate and inexpensive (pro-viding the costs of labor are excluded).

3.3. Data obtained by the FISH method(chromosome painting)

In 1989, we established a close collaboration withJoe Gray and Joe Lucas of the Lawrence LivermoreNational Laboratory (LLNL) in California. As a re-sult, a new molecular cytogenetic technique, fluores-cence in situ hybridization (FISH; also referred to as‘chromosome painting’) was introduced to our chro-mosome aberration study program. This techniquefacilitates the easy, rapid and accurate identificationof structural changes to particular chromosomes thathave been painted by hybridizing with appropriatewhole chromosome probes. One limitation of thismethod for studies such as ours in which stable chro-mosome aberrations are of primary interest is that thechanges which can be observed will be limited to re-ciprocal translocations and insertions, with inversionsescaping detection. After an extensive trial period,we decided to use three pairs of chromosomes (1, 2,and 4) as the targets for painting in the course of ourday-to-day aberration screening programs. Any struc-tural alteration to a chromosome then became apparentas a two-color chromosome consisting of painted and


unpainted segments. Only aberrations that affectedchromosomes 1, 2 or 4 were detectable, of course, andin practice this meant that the total amount of DNAbeing screened was roughly equivalent to one quarterof the human genome. It was therefore necessary toscale the aberration frequencies that we observed forchromosomes 1, 2, and 4 (f(1,2,4)) so as to provide uswith an estimate of the (whole) genome-equivalenttranslocation frequency (fG). To do this, the observedfrequency was multiplied by a coefficient of 2.77 formales and 2.81 for females (for further details referto the paper by Nakano and coworkers [24]).

A small but intensive study of samples from 20 Hi-roshima survivors indicated that the genomic translo-cation frequencies derived from FISH measurementswere almost identical to the stable chromosome aber-ration frequencies that we were deriving by G-banding[24]. Once again these results attested to the relevanceof our earlier findings on dose-aberration relation-ships.

Recently, Nakano et al. [25] reported the resultsof FISH-based analysis of translocation frequenciesfor 230 Hiroshima A-bomb survivors for whom chro-mosome aberration measurements had previouslybeen obtained by classical Giemsa-staining methods.Their main finding was that approximately 70% ofthe translocation-type aberrations detected by FISHhad also been detected in our earlier karyotypic anal-ysis by conventional Giemsa-staining methods, thusproviding strong support for our previous estimate.Regression analysis showed that there was goodagreement between the conventional and the FISHmethods, with a scoring efficiency of 0.7 for analysisusing non-banded preparations, always assuming thatour FISH technique was detecting genomic translo-cation frequencies with an efficiency of 100%.

3.4. Data on electron spin resonance (ESR)

An interesting new approach adopted by Naka-mura and his associates involved the collection ofteeth when they became available from 69 Hiroshimasurvivors for whom chromosome aberration data wasalready on file. The teeth were then used to estimatethe gamma-ray doses received by these individualson the basis of electron spin resonance (ESR) mea-surements of tooth enamel sections from the lingualportions of individual extracted molars. The results of

such ESR measurements proved to be in remarkableagreement with the accumulated cytogenetic findingsfrom our laboratory.

There were also some cases whose DS86 dose es-timates and ESR measurements were grossly discor-dant, with the ESR data corresponding much moreclosely to the chromosome aberration data [26]. It isto be expected that complex shielding situations at thetime of the bombing will have made accurate physicaldose estimation for some of the survivors extremelydifficult. In addition, some survivors’ memories of thecircumstances of their A-bomb exposures are likely tohave faded with the passage of time. It therefore seemsreasonable to assume that many of the discrepanciesbetween DS86 dose estimates and measurements ofbiological end-points will prove to be a result of errorsin dose estimates rather than errors (or inconsistency)in our cytogenetic measurements.

Before concluding this memoir of the AHS cytoge-netic study, I would like to recall a long-standing hopethat I have had for the cytogenetic analysis of A-bombsurvivors which can best be expressed in the followingquestion: will there ever be a time when we will beable to use a system of cytogenetic dosimetry basedon stable chromosome aberration frequencies as a sur-rogate for physical radiation dose estimates for thosesurvivors whose doses have never been determined orcould have been erroneously calculated because of thecomplexity of the only available shielding informa-tion? An example might be someone who was in theinterior of a large reinforced concrete building at thetime of the bombing, but whose precise location inthe building is undetermined. I firmly believe that thegreat bulk of our cytogenetic findings, and especiallythe newer ones, provide strong support for my visionthat a system of cytogenetic dosimetry is by no meansas fanciful as it sometimes seemed in the thick of somany important individual cases [27].

I was fortunate in being able to attend many interna-tional conferences at which I became acquainted witha large number of research friends. To name just avery few, I very much enjoyed the company of DavidLloyd (UK), A.T. Natarajan (the Netherlands), Man-fred Bauchinger (Germany), Gayle Littlefield (USA)and Julian Preston (USA). On at least three sepa-rate occasions, David Lloyd presented his paper im-mediately before mine, almost certainly because ourresearch areas were so very close. This was always



very embarrassing for me because David invariablyaddressed the audience in his immaculate Queen’s En-glish and led them to expect a very high standardof presentation. When I stood up to speak, I wouldbegin by asking for the audience’s indulgence whileI addressed them in my fairly typical Japanese ac-cent, “moderated” (which is a very excessively politeway of describing the effect I am referring to) by myversion of the Hiroshima dialect—which would onlyserve to make their task more difficult! I do not knowwhether these few words helped or not, but they cer-tainly seemed to.

4. Cytogenetic study of the children of A-bombsurvivors

In parallel with the AHS survey, RERF initiated along-term cytogenetic study of the children of A-bombsurvivors in 1967 and continued it until it was com-pleted in 1984. For most of the time, this unusuallylong-running project was conducted as a close work-ing collaboration with Dr. James V. Neel, who wasa Professor in Human Genetics at the University ofMichigan Medical School. Dr. Neel had been an Act-ing Director at the inception of the ABCC in 1947,and was one of the founders of the Genetics StudyProgram (untoward pregnancy outcome, sex ratio, andinfantile mortality) that ABCC/RERF had conductedfrom 1948 through 1954. In 1971, Dr. Neel suggestedthat it might be feasible to use a technique involv-ing starch-gel electrophoresis of blood plasma andserum proteins to detect variant proteins, and in par-ticular to detect any that might be present in childrenof A-bomb survivors. This project (known locally asthe Biochemical Genetics Study) began in 1972. Forthe first 3 years we conducted a feasibility test of thetechnology, and the search for variant proteins thenwent into continuous operation until it was terminatedin 1984, the same year as the cytogenetic study wascompleted. Interested readers can find a more detailedaccount of this survey in the monograph by Neel andSchull [9].

Our main objective in the cytogenetic survey was toacquire the ability to evaluate the radiation sensitivityof human germ-cell chromosomes by finding out howmany of the F1 children were carrying chromosomechanges that may have been induced in parental germ

cells as a result of parental exposure(s) to A-bombradiation in Hiroshima or Nagasaki in 1945. The re-sults of a preliminary chromosome study of survivors’children (the F1 study) appeared in 1968 [28]. Be-cause of the relatively small number of subjects in thisfirst study, a larger-scale program had to be designed;this study included many more children whose parentswere survivors selected from the F1 mortality studycohort [29].

During the period from 1967 to 1984, we studied atotal of 8322 children born to A-bomb survivors fromHiroshima and Nagasaki (the exposed group) and, ascontrols, 7976 of the children of parents from the sametwo cities who had either not been exposed at all or hadreceived (estimated) T65D doses of less than 0.005 Sv.After being fully briefed on the purpose and contentsof the survey by ABCC/RERF staff (including socialworkers and clinical genetic counselors), all partici-pants were able to decide for themselves whether theywere willing to cooperate in the study by donatingtheir blood samples on request. Full clinical exami-nations were also provided whenever requested. Thevery large numbers of blood samples that were will-ingly donated were all subjected to detailed cytoge-netic and biochemical genetic analyses.

We examined ten well-spread metaphases for eachsubject, and subjected three of them to karyotypeanalysis. When mosaicism was suspected, we scoredan additional 30–100 metaphases. All basic karyotypeanalysis was done using non-banded Giemsa-stainedpreparations. The newer banding techniques such asQ-, C-, and trypsin-G-banding were incorporated intoour routine screening when they became available.Family studies were conducted whenever possibleto determine whether any chromosome anomaliesthat we observed had been inherited from identifiedparents.

Ethical issues were and remain a very impor-tant consideration in the conduct of research of thistype, but they are also difficult and tend to changedepending upon the specific question that is beingasked. From a very early stage I felt that many ofthe survivors and their children would be sufferingfrom considerable anxiety about their future health,especially in the longer term. It also seemed morelikely than not that they would be very much afraidof their children having acquired genetic alterationsthat might be attributed to parental A-bomb radiation.


I also knew that it was going to be difficult to explaingenetics to lay-people using plain, simple wordswithout the “assistance” of unhelpful and often in-comprehensible technical terms, and yet I felt it wasextremely important to try and remove as much ofthe psychological pressure that the examinees werelikely to be experiencing as I possibly could.

I therefore decided to meet with as many childrenand parent groups as possible, and to explain to themthe overall content and purpose of the examinationthey were about to undergo. In doing this, I met manyhundreds of people in Hiroshima (unfortunately I donot remember exactly how many) to whom I offereda form of genetic counseling. I tried to spend as muchtime as necessary with each of them to ensure thatthey understood what our intentions were and whatthe study was all about. I also provided them with anopportunity to tell us about their health problems. I lis-tened carefully and patiently to what they said. Thesemeetings appeared to give many of them a useful op-portunity to air the troubles and worries that they hadkept buried deep in their heads for a very long time.I honestly believe that many of them felt greatly re-lieved by the opportunity to share their deepest con-cerns in this way. Bilateral communications of thissort appeared to be very successful, and seemed to fillthe knowledge gap between participants and examin-ers. It also helped to smooth the operations of our rou-tine surveys. We also ensured that the report of eachexamination was mailed to each individual examineewithin 2 or 3 weeks.

After completion of the survey, the results were an-alyzed statistically. Unfortunately, dose estimates us-ing the new DS86 dosimetry were not available for alarge number of the exposed parents at this time, andso our results did not include any analysis related toparental radiation doses. Thus analysis of the data wasrestricted to subjects whose parents were proximallyexposed (the exposed group) and subjects whose par-ents were either distally exposed or were not in thecity at the time of the bombing (the control group).Inborn chromosome abnormalities identified in the

survey were categorized into four groups: (a) sexchromosome aneuploidy (including sex chromosomemosaicisms); (b) structural rearrangements of chro-mosomes, most of which were in autosomes (bothreciprocal and Robertsonian translocations, and alsopericentric inversions), plus a few cases involving

sex chromosomes (inverted Y and a ring X); (c)autosomal trisomy (Down syndrome), and (d) otherabnormalities such as minute fragments.

When the data from the two cities were combined,the frequencies with which the children of exposedindividuals were found to carry inborn chromosomalerrors were: 19 with sex chromosome anomalies, 23with structural rearrangements, and 1 with trisomy ina total of 8322 children. The corresponding values forthe 7976 children in the control group were 24, 27,and 9. Thus after a great deal of work we had found nostatistically significant differences in the frequenciesof children with abnormal karyotypes in the exposedand control groups [30].

The results of this survey do not necessarily implythat there is no genetic effect to be detected in theA-bomb exposed human population. The best inter-pretation may be that the study was simply not largeenough to merit statistical analysis. In the future,perhaps the most sensitive mutation assays involvingthe A-bomb exposed human population will be thoseinvolving well-designed studies at the DNA sequencelevel. Unfortunately we may have to wait for quitesome time yet until such studies become manageableon a large enough scale.

5. Rogue cells

As my story nears its end, I would like to share withyou one episode in my scientific life that I will neverforget. This relates to our encounter with “rogue” cells.A more detailed account of our experience with thesecells is available in a book by Dr. Neel [31].

In 1968, Bloom et al. [32] described the firstcells of the type that I wish to discuss. This initialencounter occurred while Bloom et al. were conduct-ing a long-term cytogenetic study of the indigenousYanomama people in Venezuela, and the cells in ques-tion were of the type that were later to be describedas rogue cells. Their characteristics included complexexchange aberrations involving many chromosomes(multiple dicentrics and polycentrics, acentrics andminute rings), together with complicated abnormalmonocentric chromosomes.

A little later, two independent research groups inthe UK published papers on their own observationsof other types of cells that also appeared to be car-



rying multiple chromosome aberrations. The first ofthese papers was by Fox et al. [33] and appeared in1984, while the second by Tawn et al. appeared in1985 [34]. Three points are of interest here. First, themorphologies of the complex aberrations described inthese two publications were remarkably similar. Sec-ond, none of the examinees in either study had a his-tory of radiation exposure. Third, there appeared to beno rogue cells in evidence when later samples fromexaminees who had given positive results in the Tawnet al. study were examined very carefully for evidenceof their continued presence.

Now for our own experience with “rogue” cells. Itall began in the summer of 1968, when Dr. Toshio So-funi, in those days one of my collaborators, discovereda very odd metaphase while analyzing routine chro-mosome samples from a particular AHS subject. Thecell in question had many abnormal chromosomes, in-cluding dicentrics, tricentrics, polycentrics, centric andacentric rings, double minutes, and greatly rearrangedmonocentrics. After decoding the dose information,we found that this sample was from a non-exposedindividual, indicating that the complex chromosomeaberrations observed by Dr. Sofuni could have hadnothing whatsoever to do with A-bomb radiation. Bythe time we finished our first series of examination ofthe AHS population in Hiroshima, we had encountereda total of 5 of these “rogue” cells in 24,414 metaphases(0.02%) in 263 controls, and 11 in 35,564 metaphases(0.03%) in 386 exposed persons. We were unable toreport this final result until 1978 [21], even though ourfirst example was recorded in 1968, coincidentally thesame year as Bloom et al. published their first paperon cells of this sort.Soon after the discovery of our first rogue cell, we

found a number of other metaphases exhibiting multi-ple chromosome aberrations in blood samples donatedby F1 children. None of the children concerned hada history of exposure to ionizing radiation (other thanthat due to diagnostic chest X-rays), although theirparents could have been exposed to A-bomb radia-tion. Once again, therefore, there appeared to be noobvious correlation with direct exposure to ionizingradiation.

Dr. Toshio Sofuni, who as a result of this work wasone of the discoverers of rogue cells, joined the ABCCin 1967 and continued to work with us as a seniorresearch staff member of ABCC/RERF from 1967 to

1979. He then left us to join Dr. Motoi Ishidate, whowas at that time Chief of the Department of Geneticsand Mutagenesis at the National Institute of HealthSciences in Tokyo. Dr. Sofuni was later to become aninternationally recognized expert in cytogenetic as-says for chemical mutagens and carcinogens; he wasalso elected President of the Japanese Society of En-vironmental Mutagenesis (JEMS), and if my memoryserves me correctly served a term as Vice-President ofthe International Association of Environmental Mu-tagen Societies. Although he is always smiling, hisattitude towards research was consistent and rigorous,and he became an excellent leader in both researchand management.

When Dr. Neel made a routine annual visit to Hi-roshima in 1984, I showed him our data on roguecells in the F1 population. He was surprised to seehow closely our findings resembled those of his ownVenezuelan study. By this time we had discovered 24“rogue” cells in a total of 102,170 metaphases. Westrongly suspected that this unusual but not altogetheruncommon phenomenon had a viral etiology, but wehad no direct evidence to support our hypothesis. Wetherefore published a paper on the rogue cells we hadseen in which we included epidemiological data ob-tained in the course of RERFs detailed study of the F1population [35].

It was around this time that Dr. Neel intimatedto us that we should “call them ‘rogue’ cells, in theclassical biological sense of a marked deviant fromthe typical observation [30]”. I consulted several dic-tionaries, and found that the word “rogue” generallyimplies something “bad”. To me the rogue cell wasa poor cell rather than a bad cell, and so my originalfeeling was that the term “rogue” did not do justice toour observations. However, once Dr. Neel had fullyexplained the reasoning in the quotation above I washappy to accept that the word “rogue” was not sucha misleading one after all.

In 1990, we were asked to examine chromosomesin the lymphocytes of people who could have beenexposed to radioactive fallout from the Chernobylnuclear power plant accident in the former SovietUnion in 1986. This was part of a collaborative studyunder the auspices of the IAEA International Project.Once again we encountered many rogue cells in chro-mosome preparations obtained from people who hadbeen residing in uncontaminated control areas. These


data were published promptly [36]. At about the sametime, Sevan’kaev et al. reported finding rogue cellsin their study of children who had been exposed toradiation from the Chernobyl explosion. After sometheoretical speculation, Sevan’kaev et al. concludedthat these rogue cells could have been induced bysomething other than exposure to radioactive fallout[37]. After the publication of this report, there wasa veritable shower of publications on rogue cells(see [38–44]).

As far as I know, the origins and implications ofthe rogue cells that have now been seen in severaldistinct populations remain a mystery. I have no ideawhere they come from, nor do I know where theygo. One thing I am sure of, though, is that, althoughthe typical “rogue” cell has self-evidently experiencedheavy chromosome damage, it continues to respondnormally to mitogenic stimulation by PHA, to undergomitosis, and to reach metaphase without delay. Thusthe “rogue” cell should still be competent to respondto immunological challenge, a fact which leaves mewondering greatly about the mysteries of cell survivalin living organisms.

Acknowledgements

The Radiation Effects Research Foundation(RERF), Hiroshima and Nagasaki, Japan, is a privatefoundation, and since 1975 it has been funded by theJapanese Ministry of Health, Labor, and Welfare andthe United States Department of Energy through theUS National Academy of Sciences. The forerunnerof RERF was known as the Atomic Bomb CasualtyCommission (ABCC); this organization was operatedand funded by the US Department of Energy throughthe US National Academy of Sciences between 1947and 1975.

I wish to thank Dr. Nori Nakamura and his asso-ciates in the Department of Genetics, RERF, for theircontinued support extended over the decades of myassignment to RERF. Special gratitude is also due toDr. Donald MacPhee for his invaluable advice and ed-itorial help in the preparation of this manuscript.

I am deeply grateful to the citizens of Hiroshimaand Nagasaki for their willingness to participate in thesurveys I have described. Without their cooperationthese studies would not have been possible.

References

[1] J.D. Watson, F.H.C. Crick, Molecular structure of nucleicacids, Nature 171 (1953) 738–739.

[2] J.H. Tjio, A. Levan, The chromosome number of man,Hereditas 42 (1956) 1–6.

[3] M.A. Bender, X-ray induced chromosome aberrations innormal diploid human cultures, Science 126 (1957) 974–975.

[4] T.C. Hsu, C.M. Pomerat, Mammalian chromosomes in vitro.Part II. A method for spreading the chromosomes of cells intissue culture, J. Hered. 44 (1953) 23–29.

[5] P.S. Moorhead, P.C. Nowell, W.J. Mellman, D.M. Battips,D.A. Hungerford, Chromosome preparations of leukocytescultured from human peripheral blood, Exp. Cell Res. 20(1960) 613–616.

[6] I.M. Tough, K.E. Buckton, A.G. Baikie, W.M. Court Brown,X-ray-induced chromosome damage in man, Lancet 2 (1960)849–851.

[7] K.E. Buckton, P.A. Jacobs, W.M. Court Brown, R. Doll,A study of the chromosome damage persisting after X-raytherapy for ankylosing spondylitis, Lancet 2 (1962) 676–682.

[8] M. A Bender, P.C. Gooch, Persistent chromosome aberrationsin irradiated human subjects, Radiat. Res. 16 (1962) 44–53.

[9] J.V. Neel, W.J. Schull, The Children of Atomic BombSurvivors: A Genetic Study, National Academy Press,Washington, DC, 1991.

[10] W.C. Roesch, US-Japan Joint Reassessment of Atomic BombRadiation Dosimetry in Hiroshima and Nagasaki, vol. 1,Radiation Effects Research Foundation, Hiroshima, Japan,1987.

[11] A.D. Bloom, S. Neriishi, N. Kamada, T. Iseki, R.J. Keehn,Cytogenetic investigation of survivors of the atomic bombingsof Hiroshima and Nagasaki, Lancet 2 (1966) 672–674.

[12] A.D. Bloom, S. Neriishi, A.A. Awa, T. Honda, P.G. Archer,Chromosome aberrations in leucocytes of old survivors ofthe atomic bombings of Hiroshima and Nagasaki, Lancet 2(1967) 802–805.

[13] M.S. Sasaki, H. Miyata, Biological dosimetry in atomic bombsurvivors, Nature 220 (1968) 1189–1193.

[14] United Nations, Report of the United Nations ScientificCommittee on the Effects of Atomic Radiation, Annex C,Radiation-Induced Chromosome Aberrations in Human Cells,General Assembly, Official Records: Twenty-fourth Session,Supplement No. 13 (A/7613), 1969, pp. 98–155.

[15] K.E. Buckton, H.J. Evans, Methods for the Analysis of HumanChromosome Aberrations, World Health Organization,Geneva, 1973.

[16] International Atomic Energy Agency, Biological dosimetry:Chromosome Aberration Analysis for Dose Assessment,IAEA Technical Reports Series No. 260, 1986.

[17] M. A Bender, A.A. Awa, A.L. Brooks, H.J. Evans, P.G. Groer,L.G. Lttlefield, C. Pereira, R.J. Preston, B.W. Buchholz,Current status of cytogenetic procedures to detect and quantifyprevious exposure to radiation, Mutat. Res. 196 (1988) 103–159.

[18] ISCN 1995, An International System for Human CytogeneticNomenclature (1995), Cytogenetics and Cell Genetics.



[19] A.A. Awa, S. Neriishi, T. Honda, M.C. Yoshida, T. Sofuni,T. Matsui, Chromosome-aberration frequency in culturedblood-cells in relation to radiation dose of A-bomb survivors,Lancet 2 (1971) 903–905.

[20] A.A. Awa, Cytogenetic and oncogenic effects of theionizing radiations of the atomic bombs, in: J. German(Ed.), Chromosomes and Cancer, Wiley, New York, 1974,pp. 637–674.

[21] A.A. Awa, T. Sofuni, T. Honda, M. Itoh, S. Neriishi, M. Otake,Relationship between the radiation dose and chromosomeaberrations in atomic bomb survivors of Hiroshima andNagasaki, J. Radiat. Res. 19 (1978) 126–140.

[22] Y. Kodama, D. Pawel, N. Nakamura, D. Preston, T. Honda,M. Itoh, M. Nakano, K. Ohtaki, S. Funamoto, A. Awa, Stablechromosome aberrations in atomic bomb survivors: Resultsfrom 25 years of investigations, Radiat. Res. 156 (2001) 337–346.

[23] K. Ohtaki, H. Shimba, A.A. Awa, T. Sofuni, Comparisonof type and frequency of chromosome aberrations byconventional and G-staining methods in Hiroshima atomicbomb survivors, J. Radiat. Res. 23 (1982) 441–449.

[24] J.N. Lucas, A. Awa, T. Straume, M. Poggensee, Y. Kodama,M. Nakano, K. Ohtaki, H.-U. Weier, D. Pinkel, J. Gray, G.Littlefield, Rapid translocation frequency analysis in humansdecades after exposure to ionizing radiation, Int. J. Radiat.Biol. 62 (1992) 53–63.

[25] M. Nakano, Y. Kodama, K. Ohtaki, M. Itoh, R. Delongchamp,A.A. Awa, N. Nakamura, Detection of stable chromosomeaberrations by FISH in A-bomb survivors: comparison withprevious solid Giemsa staining data on the same 230individuals, Int. J. Radiat. Biol. 77 (2001) 971–977.

[26] N. Nakamura, C. Miyazawa, S. Sawada, M. Akiyama, A.A.Awa, A close correlation between electron spin resonance(ESR) dosimetry from tooth enamel and cytogeneticdosimetry from lymphocytes of Hiroshima atomic-bombsurvivors, Int. J. Radiat. Biol. 73 (1998) 619–627.

[27] Committee on Dosimetry for the Radiation Effects ResearchFoundation, Status of the Dosimetry for the Radiation EffectsResearch Foundation (DS86), 2001, National Academy Press.

[28] A.A. Awa, A.D. Bloom, M.C. Yoshida, S. Neriishi, P.G.Archer, Cytogenetic study of the offspring of atomic bombsurvivors, Nature 218 (1968) 367–368.

[29] H. Kato, W.J. Schull, J.V. Neel, A cohort-type study ofsurvival in the children of parents exposed to atomicbombings, Am. J. Hum. Genet. 18 (1966) 339–373.

[30] A.A. Awa, T. Honda, S. Neriishi, T. Sofuni, H. Shimba, K.Ohtaki, M. Nakano, Y. Kodama, M. Itoh, H.B. Hamilton,Cytogenetic study of the offspring of atomic bomb survivors,Hiroshima and Nagasaki, in: G. Obe, A. Basler (Eds.),Cytogenetics, Springer, Berlin, 1987, pp. 166–183.

[31] J.V. Neel, Physician to the Gene Pool—Genetic Lessons andOther Stories, Wiley, New York, 1994.

[32] A.D. Bloom, J.V. Neel, K.W. Choi, S. Iida, N. Chagnon,Chromosome aberrations among the Yanomama Indians, Proc.Natl. Acad. Sci. U.S.A. 66 (1968) 92–927.

[33] D.P. Fox, R.W. Robertson, T. Brown, A.R. Whitehead,J.D.M. Douglas, Chromosome aberrations in divers, UnderseaBiomed. Res. 11 (1984) 193–204.

[34] E.J. Tawn, C.L. Cartmel, E.M.T. Pyta, Cells with multiplechromosome aberrations on control individuals, Mutat. Res.144 (1985) 247–250.

[35] A.A. Awa, J.V. Neel, Cytogenetic “rogue” cells: what is theirfrequency, origin, and evolutionary significance? Proc. Natl.Acad. Sci. U.S.A. 83 (1986) 1021–1025.

[36] J.V. Neel, A.A. Awa, Y. Kodama, M. Nakano, K. Mabuchi,“Rogue” lymphocytes among Ukrainians not exposed toradioactive fall-out from the Chernobyl accident the possiblerole of this phenomenon in oncogenesis, teratogenesis, andmutagenesis, Proc. Natl. Acad. Sci. U.S.A. 89 (1992) 6973–6977.

[37] A.V. Sevan’kaev, A.F. Tsyb, D.C. Lloyd, A.A. Zhloba,V.V. Moiseenko, A.M. Skrjabin, VM. Climov, ‘Rogue’ cellsobserved in children exposed to radiation from the Chernobylaccident, Int. J. Radiat. Biol. 63 (1993) 361–367.

[38] M.A. Pilinskaya, A.M. Shemetin, S.S. Diblsky, D.V. Redko,Concerning the problem of appearance of multiaberrantcells in a man, Rep. Acad. Sci. Ukraine 3 (1993) 180–183.

[39] W. Scheid, J. Weber, S. Petrenko, H. Traut, Chromosomeaberrations in human lymphocytes apparently induced byChernobyl fallout, Health Phys. 64 (1993) 531–534.

[40] L. Verschaeve, E.V. Domracheva, S.A. Kuznetsov, V.V.Nechai, Chromosome aberrations in inhabitants of Belorussia:consequence of the Chernobyl accident, Mutat. Res. 287(1993) 253–259.

[41] N.P. Bochkov, L.D. Katosova, Analysis of multiaberrant cellsin lymphocytes of persons living in different ecologicalregions, Mutat. Res. 323 (1994) 7–10.

[42] Y.R. Ahuja, G. Obe, Are rogue cells an indicator of cancerrisk due to the action of bacterial restriction endonucleases?Mutat. Res. 310 (1994) 103–112.

[43] N.B. Rivkind, L.P. Staritsina, G.V. Skvortsova, Multiaberrantcells in liquidators in Chernobyl, in: N. Takeichi (Ed.),Chernobyl Accident: Thyroid Abnormalities in Children,Cogenital Abnormalities and Other Radiation RelatedInformation—The Firtst 10 Years, Hiroshima–Nagasaki PeaceFoundation, 1996, pp. 215–216.

[44] J.R. Lazutka, Chromosome aberrations and rogue cells inlymphocytes of Chernobyl clean-up workers, Mutat. Res. 350(1996) 315–329.


Pitfalls of enzyme-based molecular anticancer dietary manipulations: food for thought ☆Moreno Paolini a *, Marion Nestle b

a Department of Pharmacology, Biochemical Toxicology Unit, Alma-Mater Studiorum, University of Bologna, Via Irnerio 48, I-40126 Bologna, Italy

b Department of Nutrition and Food Studies, New York University, New York, NY, USAAccepted 25 November 2002



Pitfalls of enzyme-based molecular anticancer dietarymanipulations: food for thought�

Moreno Paolini a,∗, Marion Nestle b

a Department of Pharmacology, Biochemical Toxicology Unit, Alma-Mater Studiorum, University of Bologna,Via Irnerio 48, I-40126 Bologna, Italy

b Department of Nutrition and Food Studies, New York University, New York, NY, USA

Accepted 25 November 2002

Abstract

Dietary approaches to cancer chemoprevention increasingly have focused on single nutrients or phytochemicals to stim-ulate one or another enzymatic metabolizing system. These procedures, which aim to boost carcinogen detoxification orinhibit carcinogen bioactivation, fail to take into account the multiple and paradoxical biological outcomes of enzymemodulators that make their effects unpredictable. Here, we critically examine the scientific and medical evidence for theidea that the physiological roles of specific enzymes may be manipulated by regular, long-term administration of iso-lated nutrients and other chemicals derived from food plants. Instead, we argue that consumption of healthful diets ismost likely to reduce mutagenesis and cancer risk, and that research efforts and dietary recommendations should be redi-rected away from single nutrients to emphasize the improvement of dietary patterns as a principal strategy for public healthpolicy.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Cancer; Chemoprevention; Mutagenesis; Phytochemicals; Metabolizing enzymes; Dietary patterns

1. Introduction

Approaches to cancer prevention necessarily focuson eliminating cigarette smoking or improving dietand exercise patterns, both of which are believed tocontribute to about one-third of annual cancer deaths[1,2]. Dietary factors, for example, have been esti-mated to account for up to 80% of cancers of the largebowel, and breast, prostate, and even lung cancer may

� This article is part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, readers shouldcontact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).

∗ Corresponding author. Tel.: +39-051-2091783;fax: +39-051-248862.E-mail address: [email protected] (M. Paolini).

have a dietary component; to a variable extent, eatingand drinking habits can be said to have some role toplay in many if not all cancers. Many epidemiolog-ical and animal studies suggest that consumption offood plants significantly lowers the risk of cancer [3],and recent recommendations, such as those outlined inTable 1 [4], assign the highest priority to plant-baseddiets [3–6].

The evident protective effect of consuming foodplants raises the theoretical possibility that their spe-cific micronutrient or phytochemical constituentsmight have beneficial effects as chemopreventiveagents, either as naturally occurring dietary consti-tuents or pharmaceuticals that could be used to controlcancer incidence [7–10]. In the absence of knowl-edge of the specific mechanisms of action of many

1383-5742/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.doi:10.1016/S1383-5742(02)00092-3


182 M. Paolini, M. Nestle / Mutation Research 543 (2003) 181–189

Table 1Guidelines for diet and cancer prevention from [4]

Choose most of the food you eat from plant sources:Eat five more servings of fruits and vegetables each dayEat other foods from plant sources, such as breads,

cereals, grain products, rice, pasta, or beans severaltimes each day

Limit your intake of high-fat foods, particularly from animalsources:

Choose foods low in fatLimit consumption of meats, especially high-fat meats

Be physically active: achieve and maintain a healthy weightBe at least moderately active for 30 min or more on most

days of the weekStay within your healthy weight range

Limit consumption of alcoholic beverages, if you drink at all

phytochemicals, proponents of their use as chemopre-ventive agents speculate that they could manipulatethe activity of enzymes that break down mutagensand carcinogens to reduce lifetime cancer risk. Theysuggest that phase-II post-oxidative enzymes, such asglutathione S-transferase, UDP-glucuronosyl trans-ferase and acetyl transferase, promote health by detox-ifying xenobiotics, while phase-I oxidative enzymes,mainly cytochrome P450 (CYP) and FAD-containingmonooxygenases, raise cancer risk by activating car-cinogens. This rather simplistic dichotomy has in turnsuggested that plant foods rich in key nutrients or

Fig. 1. Schematic representation of biological outcomes derived in theory from enzyme-based anticancer strategies.

phytochemicals might be used to reduce the risk ofcancer through two enzyme-based strategies: boost-ing the “good” detoxifying enzymes, or inhibiting the“bad” activating enzymes [9]. Fig. 1 illustrates thesestrategies. In this scheme, phytochemical-containingfruits and vegetables such as grapes, cauliflower, kale,and broccoli stimulate phase-II induction, whereas theones contained in tea, garlic, and onion cause phase-Iinhibition.

These strategies were extrapolated from epidemio-logical observations on populations consuming dietsvarying in quantity and type of plant foods contain-ing large numbers of chemical components capableof modulating the activity of metabolizing enzymes.They have been popularized by the media, and ex-ploited by marketers of supplements of phytochemi-cals and desiccated vegetables labeled as containingsubstantial amounts of enzyme modulators [6]. Whatthese accounts fail to do is to address the complex-ity of the interactions between dietary componentsand xenobiotic metabolism. As this paper reveals,dietary magic bullets can produce health benefitsor harmful outcomes, depending on circumstancesthat cannot yet be predicted (see Fig. 1). Given thissituation, the effects of single nutrient or phytochem-ical components isolated from whole plant foodson xenobiotic metabolism and cancer risk are alsouncertain.

M. Paolini, M. Nestle / Mutation Research 543 (2003) 181–189 183

2. Enzyme manipulation strategies

The boosting strategy involves large-scale induc-tion of phase-II metabolizing enzymes that detoxifyxenobiotics, thereby accelerating the elimination oftoxicants and protecting cells against mutagenesis andneoplasia [9]. The potential benefits of this strategyhave stimulated active investigations of the chemi-cal specificity of inducers and their molecular mech-anisms in many laboratories [10–24]. Much attentionhas focused on resveratrol, a phytoalexin found ingrapes and other food products that boosts phase-IIlinked activities [25], and cruciferous (mustard fam-ily) vegetables of the genus Brassica, such as broc-coli, kale, cabbage, Brussels sprouts and cauliflower.Brassica vegetables contain considerable quantities ofglucosinolates; these are precursors of isothiocyanates[26,27], which are potent inducers of phase-II en-zymes [28–30]. US health authorities have recom-mended consumption of these vegetables for cancerprevention since the early 1980s [31]. An alternativeanticancer hypothesis is to inhibit the typical phase-Ibioactivating enzymes [9,22]. Such strategies now per-meate both the scientific literature and the media, assuggested by numerous reports urging regular con-sumption of garlic and onions rich in diallyl sulfide orgreen or black tea containing catechins [32–34].The difficulty with these strategies is that they

ignore the complexity of metabolizing enzyme sys-tems. Although consumption of food plants, whichcontain hundreds of phytochemicals, is associatedwith reduced cancer risk, stimulation of xenobioticmetabolism by one specific component may alsostimulate unwanted formation of active mutagenicmetabolites [35,36]. The use of isolated naturally oc-curring dietary constituents such as isothiocyanatesor individual drugs such as disulfiram, for example,also elicit contrary effects that can be highly unde-sirable. As discussed below, both proposed strategiesalso must be considered in the context of geneticpolymorphisms, which may differentially modulatethe effects of any one dietary factor on individuals.

3. Limitations of the boosting strategy

Enzyme upregulators are already consumed by hu-mans as food additives such as BHA [2(3)-tert-butyl-4-

hydroxyanisole], medicines such as oltipraz, or natu-ral constituents of vegetables such as glucoraphanin,bioprecursor of sulforaphane. That such compoundsmight confer protection against cancer by raising theactivity of post-oxidative enzymes has been widelyaccepted during the last two decades [9,10], somuch so that researchers have created hybrid plantsspecifically to produce higher amounts of singlephytochemical-inducers [37]. Such efforts ignore ev-idence suggesting that each phase-II enzyme is alsoinvolved in electrophilic species generation and, there-fore, should be considered as an “activating system”for specific chemical classes: halogenated hydrocar-bons by glutathione S-transferases, for example, orpolycyclic aromatic hydrocarbons (PAHs) by epoxidehydrolases or sulphotransferases [38]. In other words,the bioactivation or bioinactivation of a specific com-pound depends on the nature of the compound itself.

In general, manipulation of the activity of oneor more post-oxidative enzymes can either increaseor reduce the bioactivation of specific compounds.Whereas induction increases the detoxification ofcertain protoxics and promutagens/procarcinogens,thereby favoring chemoprevention, it also increasesthe bioactivation of countless other foreign chemicals.Since humans are exposed to a myriad of potentiallyharmful molecules, any modification of the activityof these enzymes could actually lead to an increasein toxicological risk [39].

Thus, we should not be surprised if molecules thatin certain experiments appear to possess anticancerproperties actually turn out to have unexpectedlydetrimental effects in humans. For example, crucifer-ous isothiocyanates such as sulforaphane, most oftenconsidered as beneficial phase-II detoxifying systeminducers, turn out to be genotoxics or strong pro-moters of urinary bladder and liver carcinogenesisas well as inducing cell cycle arrest and apopto-sis [40–42]. Hepatic metabolic S9 fractions isolatedfrom rodents treated with BHA (monofunctionalphase-II booster) paradoxically have been proposedas metabolizing systems to bioactivate promuta-gens in short-term genotoxicity tests [43]. Similarly,engineered Salmonella typhimurium TA1535 trans-fected with the plasmid vector pKK233-2 containingrat glutathione S-transferase 5-5 cDNA has beenshown to activate many genotoxicants, whereas thenon-transfected counterpart does not [44].



The difficulty of determining how isolated dietaryfactors might affect metabolizing enzymes is illus-trated by inconsistencies in studies on cruciferous veg-etables. Although consumption of such vegetables is,on balance, associated with reduced cancer risk [3],epidemiological data show that a high intake of theseplant foods in the form of vegetable mixtures [45]or single plants (e.g. broccoli, cabbage or Brusselssprouts) [46], can exert cancer-enhancing effects dueto their content of enzyme inducers that activate pro-carcinogens such as polycyclic aromatic hydrocarbonsand aromatic amines in tobacco. Furthermore, the in-dole carbinol, a well-known promoter of chemicalcarcinogenesis and an inducer of dioxin-metabolizing(CYP1A2) enzymes, is also present in large amountsin broccoli and other brassicas [13,47,48]. On this ba-sis, the authoritative refusal of a former President ofthe US, George Bush, to eat broccoli may be under-stood to have little effect on his cancer risk.

4. Drawbacks of the inhibitory approach

Similar considerations affect the inhibitory strat-egy. The inhibitory approach has stimulated recom-mendations to increase consumption of green andblack teas, because they contain catechins and otherchemicals that determine the inhibition of phase-Ienzymes and reduce the generation of mutagens andcarcinogens such as N-nitroso compounds [9,49].Likewise, the extensively documented inhibition ofdimethylhydrazine-induced colon cancer by diallylsulfide, a flavor component of garlic (Allium sativum),or by drugs like disulfiram, has encouraged recom-mendations to increase garlic consumption [50,51].In addition, the flavonoid naringin, the most abundantphytochemical of grapefruit and related citrus fruits,has been found to be a highly effective blocker of afla-toxin B1 activation and of carcinogens activated byCYP3A4 [52]. Such examples from cell and animalmodels seem to support the idea that vegetable-basedenzyme modulators may inhibit carcinogenesis.

This concept, however, ignores evidence that, due tothe benefical but also detrimental nature of cytochromeenzyme systems, the reduced activation of certain xe-nobiotics occurs simultaneously with reduced detoxi-fication of other environmental toxicants. Such unde-sirable side effects are highly unpredictable, however.

Because humans are unable to select safe, person-alized exposures to chemicals in such a way as tosystematically avoid harmful compounds, the perpet-ual inhibition of phase-I enzymes might actually leadto an increase in cancer risk. Investigations on theeffects of tea on cancer risk are an example of thisproblem. Whereas some studies in different popula-tions have shown a protective effect of tea consump-tion against certain types of malignancy, others haveindicated a negative effect [49,53]. Similarly, somelaboratory studies have shown anticancer properties ofgreen and black tea preparations and others the op-posite effect [54]. In this context, it is notable that arecent prospective cohort study failed to support thehypothesis that consumption of tea protects againstcancer at four major sites among elderly people [55].

An inhibitor that selectively affects one CYP en-zyme may be an inducer of other CYPs; for example,phenethyl isothiocyanate derived from Brassica, anddiallyl sulfide from garlic, inhibit CYP2E1 but induceCYP2B1 and CYP1A2 [56]. Also, because of the pres-ence in humans of multiple CYP isoforms, each ableto activate specific compounds, an inhibitory approachnecessarily requires the use of a “cocktail” of enzy-matic inhibitors (or inducers, for the boosting strat-egy), one for each CYP to be manipulated, thus leadingto complex and unpredictable biological outcomes.Unhealthy consequences from supplementation withenzyme-activity manipulators could also stem from al-teration of endogenous metabolism (linked to catalystssuch as arachidonic acid derivatices, nitric oxide, al-dosterone, cholesterol, or vitamins, or the pharmacoki-netics of co-administrated drugs). Phase-I enzymes areupstream in the regulatory cascade of numerous trans-duction signal pathways involved in the maintenanceof steady-state levels of specific endogenous ligandsin cells. Thus, xenobiotics that mimic these ligands,after binding with specific cytosolic receptors, can actas agonists/antagonists in activating/inhibiting genes,thereby affecting growth, differentiation, apoptosis,homeostasis and neuroendocrine functions [57].

5. The influence of genetic polymorphisms

The paradoxical effects of isolated dietary com-ponents on metabolizing enzymes are further com-plicated by genetic polymorphisms that lead to the


occurrence of high- or low-metabolizer phenotypesin the population, each at increased toxicologicalrisk from exposure to specific chemicals [58]. Forexample, the high susceptibility of fast acetylators topathologies such as colorectal cancer and type I dia-betes, and of slow acetylators to bladder cancer, lupuserythematous, liver disease and drug-induced neuro-toxicity, has been widely reported [59]. The extensivedebrisoquine metabolizer phenotype is associatedwith a disproportionately high risk of lung canceramong smokers, as well as of liver and gastroin-testinal carcinoma; the poor-metabolizer phenotypeis associated with an acute idiosyncratic responsein hypotension to adrenergic blocking agents and achronic response in Parkinson’s disease [60]. In addi-tion, the multiple polymorphisms (e.g. occurrence ofhigh or low-metabolizers for any of oxidative or post-oxidative isoenzymes) characterizing the “individual”metabolic fingerprint, further complicate the issue.Thus, the possibility of manipulating enzyme activ-ity, which already in its “constitutive” diversity maydetermine genetic disorders, raises further questionsabout the effectiveness of chemical-based enzymaticmodulation of cancer risk [61,62]. Such questionssuggest the need for considerable caution before con-sidering any form of enzyme-activity manipulation forgeneralized chemoprevention such as that indicatedin Fig. 1.

6. A broader perspective: recapturing theforest from the trees

At issue is the clinical significance of modulationof such enzymatic systems by single phytochemicalsand the need to retain sight of the larger “forest” con-text. Both phase-I and phase-II enzymes are highlymultifunctional and can be induced or inhibited bya wide variety of dietary compounds. Plants haveevolved over millennia in such a way as to producethousands of natural pesticides against infection bymicrobes and predation by animals, and humans mayconsume as many as 10,000 of these chemicals andtheir metabolic products when eating vegetables [63].Of the dozens of such compounds that have beenidentified in cabbage for example, several have beenfound to be mutagenic and carcinogenic in bioassays[64]. Thus, cruciferous and other vegetables contain

some phytochemicals that are anticarcinogenic, alongwith others that are carcinogenic.

The dual activating and detoxicating nature of en-zymatic systems, the impressive number of chemicalsthat can modulate them, the presence in vegetablesof chemicals that induce both activation and inhi-bition of carcinogenesis, the genetically determinedinter-individual variability that may moderate the ef-fects of specific dietary factors, and the complexity ofthe interactions among food constituents and enzymesystems feed ongoing debates as to whether glucosi-nolates or other phytochemicals can alone explain thecancer-protective ability of many vegetables [6].

It is highly unlikely that single-constituent foodsupplements would offer an advantage, since a varietyof fruit and vegetables seems necessary to provide themixture of vitamins and minerals (including the es-sential enzyme cofactors iron, niacin and riboflavin),carotenoids, folic acid, fibers, and phytochemicals thatappear to favor protection against carcinogenesis [65].Indeed, in contrast to the uncertainty surroundingthe precise roles of specific single-nutrient compo-nents, the overall anti-mutagenic/carcinogenic prop-erties of vegetables strongly outweigh any adverseeffects of their constituent carcinogens or carcinogen-modulators. It is difficult to see how the beneficialeffects of consuming vegetables, in which enzyme-modulating components appear in varying amountsand proportions, and in which unpredictable syn-ergistic and antagonistic interactions occur amongthousands of different chemicals in their natural ma-trix, could be reproduced by supplements of singlecomponents [66].

The unexpected results of cancer chemopreven-tion trials of antioxidant provitamins and vitaminsconstitute an exemplary warning about the hazardsof single-nutrient approaches [67–72]. Beta-caroteneadministered alone or in combination with VitaminsA, E or C for the prevention of lung cancer and othercancers in heavy smokers or asbestos workers failedto reduce cancer risk and, in some cases, actuallyincreased the risk, raising the suspicion that single-nutrient supplements may have harmful as well asbeneficial effects [72–79]. Detrimental effects of beta-carotene supplementation seem to be linked to itsability to stimulate metabolizing enzymes such asactivators of polycyclic aromatic hydrocarbons andto generate an oxidative stress [80] and, therefore, to



alter tumor suppressor genes [81]. Likewise, excessivesupplementation with Vitamin C has induced seriousoxidative DNA damage in human lymphocytes [82],probably by means of the ability of this vitamin toproduce oxygen centered radicals linked to phase-Iinduction [83]. The finding that the antioxidant ac-tivity of synthetic Vitamin C is much lower than thatof extracts of fresh apples [84] supports this con-cern; the decomposition of lipid hydroperoxydes toDNA-reactive bifunctional electrophiles by VitaminC has also been recently documented [85].

7. Concluding remarks

In marketing, claims of simple solutions to com-plex problems in the form of dietary supplements havebeen demonstrated to have considerable mass appeal.The idea of a “magic bullet,” as conceived by PaulEhrlich for antibacterial substances, engages the imag-ination of the public and scientists alike. In the fieldof cancer prevention, a magic-bullet approach can beseen as a reductive search for a long-life elixir on themolecular level. Given the evident marketing attrac-tions of such implicit promises, advertising campaignsfor broccoli and its derivatives provide a rich sourceof scientific ambiguity for the commercial purveyorsof several mass chemoprevention strategies.

Such approaches also attract scientists. Studies ofcancer prevention involving the use of single nutrientsor phytochemicals are demonstrably easier to conduct,analyze, and report, and funding agencies much pre-fer them to messier examinations of dietary patternsand cancer risk. From the standpoint of cancer re-search policy, the role of single dietary constituentsis of pivotal interest. Basic information about the roleof metabolizing enzyme systems, however, makes itclear that the role of any single anticarcinogenic phy-tochemical, no matter how well characterized, cannotbe understood except in the context of broader dietarypatterns. The level of scientific controversy surround-ing the effects of single micronutrients or phytochem-icals on cancer risk should provide a salutary warningfor health policymakers.

Despite the strong attractions exerted by chemo-prevention, the strategies involving enzymatic activitymanipulations require reconsideration in the contextof public health policies. The interactions of the count-

less molecular dietary constituents within human bio-logical systems can best be understood as analogous tothe global balance found in ecosystems. This analogyhelps explain the disappointing results so far obtainedfrom phytochemical approaches to cancer preven-tion, and why recommendations to consume singlesubstances are unlikely to improve cancer risk [3,4].

Providing that other lifestyle factors are also takeninto account, educational campaigns encouraging theconsumption of fruit, fiber, and greens can be wel-comed. In general, we need to gain a clear understand-ing of the relationship between the health-promotingeffects of certain dietary patterns and their behavioral,economic, environmental, and cultural determinants,as well as of the molecular basis of these relationships[6].

Acknowledgements

We are indebted to Professor Pier Luigi Lollini (In-stitute for Cancer Research, University of Bologna,Italy) and Dr. Giuseppe Remuzzi (Mario Negri In-stitute, Bergamo, Italy) for critical comment on themanuscript. We thank Robin M.T. Cooke for editing.

References

[1] J.C. Bailar III, H.L. Gornik, Cancer undefeated, New Engl.J. Med. 336 (1997) 1569–1574.

[2] J.M. McGinnis, W.H. Foege, Actual causes of death in theUnited States, J. Am. Med. Assoc. 270 (1993) 2207–2212.

[3] World Cancer Research Foundation and American Institute forCancer Research, Food nutrition and the prevention of cancer:a global perspective, World Cancer Research Foundation andAmerican Institute for Cancer Research, Washington, DC,1997.

[4] American Cancer Society Advisory Committee on Diet,Nutrition and cancer prevention, Ca Cancer J. Clin. 46 (1996)325–341.

[5] M. Nestle, Dietary recommendations for cancer prevention:public policy implementation, J. Natl. Cancer Inst. Monogr.12 (1992) 153–157.

[6] M. Nestle, Broccoli sprouts as inducers of carcinogen-detoxi-fying enzyme systems: clinical, dietary, and policy implica-tions, Proc. Natl. Acad. Sci. U.S.A. 94 (1997) 11149–11151.

[7] R. Peto, R.J. Doll, J.D. Buckley, M.B. Sporn, Can dietarybeta-carotene materially reduce human cancer rates? Nature290 (1981) 201–208.

[8] L.W. Wattenberg, Chemoprevention of cancer, Cancer Res.45 (1985) 1–8.


[9] M.A. Morse, G.D. Stoner, Cancer chemoprevention: princi-ples and prospects, Carcinogenesis 14 (1993) 1737–1746.

[10] C. Gerhauser, M. You, J. Liu, et al., Cancer chemopreventionpotential of sulforamate, a novel analogue of sulforaphanethat induces phase 2 drug-metabolizing enzymes, Cancer Res.57 (1997) 272–278.

[11] T. Prestera, W.D. Holtzclaw, Y. Zhang, P. Talalay, Chemi-cal and molecular regulation of enzymes that detoxify carci-nogens, Proc. Natl. Acad. Sci. U.S.A. 90 (1993) 2965–2969.

[12] G.J. Kelloff, C.W. Boone, J.A. Crowell, et al., New agentsfor cancer chemoprevention, J. Cell. Biochem. 26 (Suppl.)(1996) 1–28.

[13] C. Gerhauser, M. You, J. Liu, et al., Chemopreventivepotential of a novel sulforaphane analog, Proc. Am. Assoc.Cancer Res. 37 (1996) A189.

[14] Y. Zhang, T.W. Kensler, C.G. Cho, G.H. Posner, P. Talalay,Anticarcinogenic activities of sulforaphane and structurallyrelated synthetic norbornyl isothiocyanates, Proc. Natl. Acad.Sci. U.S.A. 91 (1994) 3147–3153.

[15] K. Maheo, F. Morel, S. Langouet, et al., Inhibition ofcytochromes P-450 and induction of glutathione S-trans-ferases by sulforaphane in primary human and rat hepatocytes,Cancer Res. 57 (1997) 3649–3652.

[16] Y. Zhang, P. Talalay, Mechanism of differential potenciesof isothiocyanates as inducers of anticarcinogenic phase 2enzymes, Cancer Res. 58 (1998) 4632–4639.

[17] T.A. Shapiro, J.W. Fahey, K.L. Wade, K.K. Stephenson, P.Talalay, Human metabolism and excretion of cancer chemo-protective glucosinolates and isothiocyanates of cruciferousvegetables, Cancer Epidemiol. Biomarkers Prev. 7 (1998)1091–1100.

[18] H.L. Bradlow, N.T. Telang, D.W. Sepkovic, M.P. Osborne,Phytochemicals as modulators of cancer risk, Adv. Exp. Med.Biol. 472 (1999) 207–221.

[19] S.S. Hecht, Chemoprevention of cancer by isothiocyanates,modifiers of carcinogen metabolism, J. Nutr. 129 (1999)768S–774S.

[20] R. Yu, W. Lei, S. Mandlekar, et al., Role of a mitogen-activated protein kinase pathway in the induction of phase IIdetoxifying enzymes by chemicals, J. Biol. Chem. 274 (1998)27545–27552.

[21] G.J. Kelloff, J.A. Crowell, V.E. Steele, et al., Progress incancer chemoprevention: development of diet-derived chemo-preventive agents, J. Nutr. 130 (Suppl. 2S) (2000) 467S–471S.

[22] L. Gamet-Payrastre, P. Li, S. Lumeau, et al., Sulforaphane,a naturally occurring isothiocyanate, induces cell cycle arrestand apoptosis in HT29 human colon cancer cells, CancerRes. 60 (2000) 1426–1433.

[23] V.P. Kelly, E.M. Ellis, M.M. Manson, et al., Chemopreventionof aflatoxin B1 hepatocarcinogenesis by coumarin, a naturalbenzopyrone that is a potent inducer of aflatoxin B1-aldehydereductase, the glutathione S-transferase A5 and P1 subunits,and NAD(P)H: quinone oxidoreductase in rat liver, CancerRes. 60 (2000) 957–969.

[24] A.T. Dinkova-Kostova, W.D. Holtzclaw, R.N. Cole, et al.,Direct evidence that sulfhydryl groups of Keap1 are thesensors regulating induction of phase 2 enzymes that protect

against carcinogens and oxidants, Proc. Natl. Acad. Sci.U.S.A. 99 (2002) 11908–11913.

[25] M. Jang, L. Cai, G.O. Udeani, et al., Cancer chemopreventionactivity of resveratrol, a natural product derived from grapes,Science 275 (1997) 218–220.

[26] D.T. Verhoeven, H. Verhagen, R.A. Goldbohm, et al., Reviewof mechanisms underlying anticarcinogenicity by brassicavegetables, Chem. Biol. Interact. 10 (1997) 79–129.

[27] D.T. Verhoeven, R.A. Goldbohm, G. van Poppel, et al.,Epidemiological studies on brassica vegetables and cancerrisk, Cancer Epidemiol. Biomarkers Prev. 5 (1996) 733–748.

[28] H. Steinkellner, S. Rabot, C. Freywald, E. Nobis, G.Scharf, M. Chabicovsky, S. Knasmuller, F. Kassie, Effectsof cruciferous vegetables and their constituents on drugmetabolizing enzymes involved in the bioactivation of DNA-reactive dietary carcinogens, Mutat. Res. 480–481 (2001)285–297.

[29] S. Barcelo, J.M. Gardiner, A. Gescher, J.K. Chipman,CYP2E1-mediated mechanism of anti-genotoxicity of thebroccoli constituent sulforaphane, Carcinogenesis 17 (1996)277–282.

[30] M.A. Kall, O. Vang, J. Clausen, Effects of dietary broccolion human in vivo drug metabolizing enzymes: evaluation ofcaffeine, Carcinogenesis 17 (1996) 793–799.

[31] National Research Council Diet, Nutrition and Cancer,National Academy Press, Washington, DC, 1982.

[32] P.P. Tadi, R.W. Teel, B.H.S. Lau, Organosulfur compounds ofgarlic modulate mutagenesis, metabolism, and DNA bindingof aflatoxin B1, Nutr. Cancer 15 (1991) 87–95.

[33] J.Y. Hong, T. Smith, M.J. Lee, et al., Metabolism of carci-nogenic nitrosamines by rat nasal mucosa and the effect ofdiallyl sulfide, Cancer Res. 51 (1991) 1509–1514.

[34] J. Hernaez, M. Xu, R. Dashwood, Effects of tea and chloro-phyllin on the mutagenicity of N-hydroxy-IQ: studies ofenzyme inhibition, molecular complex formation, and degra-dation/scavenging of the active metabolites, Environ. Mol.Mut. 30 (1997) 468–474.

[35] M. Paolini, Brussels sprouts: an exceptionally rich source ofambiguity for anticancer strategies, Toxicol. Appl. Pharmacol.152 (1998) 293–294.

[36] M. Paolini, M.S. Legator, Healthy broccoli? Nature 357(1992) 448.

[37] K. Faulkner, R. Mithen, G. Williamson, Selective increaseof the potential anticarcinogen 4-methylsulphinylbutyl gluco-sinolate in broccoli, Carcinogenesis 19 (1998) 605–609.

[38] F.P. Guengerich, Enzymatic activation of chemicals to toxicmetabolites, CRC Crit. Rev. Toxicol. 14 (1985) 259–307.

[39] G. Cantelli-Forti, P. Hrelia, M. Paolini, The pitfall of detoxi-fying enzymes, Mutat. Res. 402 (1998) 179–183.

[40] M. Hirose, T. Yamaguchi, N. Kimoto, et al., Strong promot-ing activity of phenylethyl isothiocyanate and benzyl isothio-cyanate on urinary bladder carcinogenesis in F344 male rats,Int. J. Cancer 77 (1998) 773–777.

[41] F. Kassie, B. Pool-Zobel, W. Parzefall, S. Knasmuller, Geno-toxic effects of benzyl isothiocyanate, a natural chemopreven-tive agent, Mutagenesis 14 (1999) 595–604.

[42] L. Gamet-Payrastre, P. Li, S. Lumeau, et al., Sulforaphane,a naturally occurring isothiocyanate, induces cell cycle arrest



[75] X.D. Wang, R.M. Russell, Procarcinogenic and anticarci-nogenic effects of beta-carotene, Nutr. Rev. 57 (1999) 263–272.

[76] I.M. Lee, N.R. Cook, J.E. Manson, J.E. Buring, C.H.Hennekens, Beta-carotene supplementation and incidence ofcancer and cardiovascular disease: The Women’s HealthStudy, J. Natl. Cancer Inst. 91 (1999) 2102–2106.

[77] J.R. Marshall, Beta-carotene: a miss for epidemiology, J. Natl.Cancer Inst. 91 (1999) 2068–2069.

[78] H. Vainio, Chemoprevention of cancer: lessons to be learnedfrom beta-carotene trials, Toxicol. Lett. 15 (2000) 513–517.

[79] G.E. Goodman, Prevention of lung cancer, Crit. Rev. Oncol.Hematol. 33 (2000) 187–197.

[80] M. Paolini, G. Cantelli-Forti, P. Perocco, G.F. Pedulli, S.Z.Abdel-Rahman, M.S. Legator, Co-carcinogenic effect ofbeta-carotene, Nature 398 (1999) 760–761.

[81] X.D. Wang, C. Liu, R.T. Bronson, D.E. Smith, N.I. Krinsky,M. Russell, Retinoid signaling and activator protein-1expression in ferrets given beta-carotene supplements andexposed to tobacco smoke, J. Natl. Cancer Inst. 91 (1999)60–66.

[82] I.D. Podmore, H.R. Griffiths, K.E. Herbert, N. Mistry, P.Mistry, J. Lunec, Vitamin C exhibits pro-oxidant properties,Nature 392 (1998) 559.

[83] M. Paolini, L. Pozzetti, G.F. Pedulli, E. Marchesini, G.Cantelli-Forti, The nature of prooxidant activity of vitaminC, Life Sci. 64 (1999) 273–278.

[84] M.V. Eberhard, C.Y. Lee, H.L. Liv, Antioxidant activity offresh apples, Nature 405 (2000) 903–904.

[85] S.H. Lee, T. Oe, I.A. Blair, Vitamin C-induced decompositionof lipid hydroperoxides to endogenous genotoxins, Science292 (2001) 2083–2086.


and apoptosis in HT29 human colon cancer cells, CancerRes. 60 (2000) 1426–1433.

[43] M. Paolini, L. Pozzetti, P. Silingardi, C. Della Croce, G.Bronzetti, G. Cantelli-Forti, Isolation of a novel metabo-lizing system enriched in phase-II enzymes for short-termgenotoxicity bioassays, Mutat. Res. 413 (1998) 205–217.

[44] R. Thier, M. Muller, J.B. Taylor, S.E. Pemble, B. Ketterer,F.P. Guengerich, Enhancement of bacterial mutagenicity ofbifunctional alkylating agents by expression of mammalianglutathione S-transferase, Chem. Res. Toxicol. 8 (1995) 465–472.

[45] S.R. Tannenbaum, Predictors of dietary heterocyclic amineintake in three prospective cohorts, Cancer Epidemiol.Biomarkers Prev. 7 (1998) 1153–1154.

[46] J.W. Lampe, I.B. King, S. Li, et al., Brassica vegetablesincrease and apiaceous vegetables decrease cytochrome P4501A2 activity in humans: changes in caffeine metabolite ratiosin response to controlled vegetable diets, Carcinogenesis 21(2000) 1157–1162.

[47] G.S. Bailey, J.D. Hendricks, D.W. Shelton, J.E. Nixon,N.E. Pawlowski, Enhancement of carcinogenesis by naturalanticarcinogen indole-3-carbinol, J. Natl. Cancer Inst. 78(1987) 931–934.

[48] R.H. Dashwood, Indole-3-carbinol: anticarcinogen or tumorpromoter in brassica vegetables? Chem. Biol. Interact. 110(1998) 1–5.

[49] C.S. Yang, Tea and cancer, J. Natl. Cancer Inst. 85 (1993)1038–1049.

[50] E.S. Fiala, G. Bobotas, C. Kulakis, L.W. Wattenberg, J.H.Weisburger, Effects of disulfiram and related compoundson the metabolism in vivo of the colon carcinogen, 1,2-dimethylhydrazine, Biochem. Pharmacol. 26 (1977) 1763–1768.

[51] M.J. Wargovich, Diallyl sulfide, a flavor component of garlic(Allium sativum), inhibits dimethylhydrazine-induced coloncancer, Carcinogenesis 8 (1987) 487–489.

[52] F.P. Guengerich, D.H. Kim, In vitro inhibition of dihydro-pyridine oxidation and aflatoxin B1 activation in human livermicrosomes by naringenin and other flavonoids, Carcino-genesis 11 (1990) 2275–2279.

[53] IARC Working Group on the Evaluation of CarcinogenicRisk to Humans, Coffee, tea, mate, methylxanthines andmethylglyoxal. Lyon, 1990, IARC Monographs on theEvaluation of Carcinogenic Risks to Humans. No. 51, Inter-national Agency for Research on Cancer, Lyon, 1991,pp. 207–271.

[54] A. Bu-Abbas, M.N. Clifford, R. Walker, C. Ioannides,Modulation of hepatic cytochrome P450 activity andcarcinogen bioactivation by black and decaffeinated black tea,Env. Toxicol. Pharmacol. 7 (1999) 41–47.

[55] A. Goldbohm, M.G.L. Hertog, H.A.M. Brandts, G. vanPoppel, P. van den Brandt, Consumption of black tea andcancer risk: a prospective cohort study, J. Natl. Cancer Inst.88 (1996) 93–100.

[56] C.S. Yang, T.J. Smith, J.Y. Hong, L. Chen, Cancer preventionthrough modulation of xenobiotic metabolising enzymes:opportunities and limitations, Proc. Annu. Meet. Am. Assoc.Cancer Res. 37 (1996) 665–666.

[57] D.W. Nebert, Drug metabolizing enzymes in ligand-modu-lated transcription, Biochem. Pharmacol. 47 (1994) 25–27.

[58] D.W. Nebert, Role of genetics and drug metabolism in humancancer risk, Mutat. Res. 247 (1991) 267–281.

[59] A.K. Daly, S. Cholerton, W. Gregory, J.R. Idle, Metabolicpolymorphisms, Pharmacol. Ther. 57 (1993) 129–160.

[60] L.W. Wormhoudt, J.N.M. Commandeur, N.P.E. Vermeulen,Genetic polymorphisms of human N-acetyltransferase, cyto-chrome P450, glutathione-S-transferase, and epoxide hydro-lase enzymes: relevance to xenobiotic metabolism andtoxicity, Crit. Rev. Toxicol. 29 (1999) 59–124.

[61] M. Paolini, Acetylator genotype and Parkinson’s disease,Lancet 351 (1998) 141–142.

[62] M. Paolini, G.L. Biagi, G. Cantelli-Forti, Glutathione trans-ferase polymorphism and Parkinson’s disease, Lancet 353(1999) 71–72.

[63] M. Paolini, G.L. Biagi, C. Bauer, G. Cantelli-Forti, Enzymeevolution and cancer: a hypothesis to explain why naturalcarcinogens are more potent than synthetic ones, Life Sci. 63(1998) 2141–2146.

[64] B.N. Ames, M. Profet, L.S. Gold, Dietary pesticides (99.99%all natural), Proc. Natl. Acad. Sci. U.S.A. 87 (1990) 7777–7781.

[65] S.A. Gerrior, C. Zizza, Nutrient content of the US food supply1909–90, US Department of Agriculture, Washington, DC,1994.

[66] A.R. Collins, B. Olmedilla, S. Southon, F. Granado, S.Y.Duthie, Serum carotenoids and oxidative DNA damage inhuman lymphocytes, Carcinogenesis 19 (1998) 2159–2162.

[67] C.H. Hennekens, J.E. Buring, J.E. Manson, et al., Lack ofeffect of long-term supplementation with beta carotene on theincidence of malignant neoplasms and cardiovascular disease,New Engl. J. Med. 334 (1996) 1145–1149.

[68] G.S. Omenn, G.E. Goodman, M.D. Thornquist, et al., Effectsof a combination of beta carotene and vitamin A on lungcancer and cardiovascular disease, New Engl. J. Med. 334(1996) 1150–1155.

[69] P.M. Rowe, Beta-carotene takes a collective beating, Lancet347 (1996) 249.

[70] G.S. Omenn, G.E. Goodman, M.D. Thornquist, et al., Riskfactors for lung cancer and for intervention effects in CARET,J. Natl. Cancer Inst. 88 (1996) 1550–1559.

[71] T.S. Hinds, W.L. West, E.M. Knight, Carotenoids andretinoids: a review of research, clinical, and public healthapplications, J. Clin. Pharmacol. 37 (1997) 551–558.

[72] M. Paolini, S.Z. Abdel-Rahman, G. Cantelli-Forti, M.S.Legator, Chemoprevention or antichemoprevention? Asalutary warning from the beta-carotene experience, J. Natl.Cancer Inst. 93 (2001) 1110–1111.

[73] S.T. Omaye, N.I. Krinsky, V.E. Kagan, S.T. Mayne, D.C.Liebler, W.R. Bidlack, Beta-carotene: friend or foe? Fund.Appl. Toxicol. 40 (1997) 163–174.

[74] D.L. Baker, E.S. Krol, N. Jacobsen, D.C. Liebler, Reactionsof beta-carotene with cigarette smoke oxidants. Identificationof carotenoid oxidation products and evaluation of theprooxidant/antioxidant effect, Chem. Res. Toxicol. 12 (1999)535–543.


2 E.H.Y. Chu / Mutation Research 566 (2004) 1–8

The first order of business was to set up a primitivecell culture facility within our laboratory in whichthe principal research activity was the biochemicalgenetics of Neurospora. With a scientific backgroundin agriculture and prior experience in genetic studiesof Drosophila, Nicotiana and Neurospora, I was in agood position to take on this new challenge. At thattime on the Yale campus only four laboratories wereapplying cell or tissue culture techniques; the investi-gations ranged from chick embryonic fibroblasts fordevelopmental studies to primary cultures of monkeykidney epithelial cells for poliovirus research. Whilepracticing cell culture methodology, we publishedtwo articles that dealt with, respectively, the chromo-some complements and DNA contents of a number ofOld World primates [13,14]. Up to that time, the onlyknown chromosome number of nonhuman primateswas that of the Rhesus monkey (Maccaca mulatta,2n = 42). Our subsequent interest in, and reportson, the cytotaxonomy and evolution of Prosimianand Old World monkeys are beyond the scope ofthis article.

To search for suitable human somatic cell materialfor genetic analyses, established cell lines, includingmany derived from presumably normal tissues, wereobtained from many investigators. We examined over35 cell lines, and all turned out to be heteroploids. At-tempts were then made to develop strains of normalhuman skin fibroblasts that would be useful in definingthe human karyotype [15,16] and in quantitative anal-ysis of chromosome aberrations induced by X-rays[17]. The limited life span and low cloning efficiencyof human normal diploid fibroblasts in culture posedsevere restrictions for extensive genetic analyses.In order to initiate diploid human fibroblast strains,

we depended on surgeons to save us biopsy materi-als. I had to go occasionally to the Yale Hospital towait outside operating rooms, and I regularly went tothe Pediatrics Department for foreskin samples. Thesamples were kept in sterile 10 cm glass Petri dishes,which were packed in a metal can, marked outsidewith a sign saying ‘foreskin for Botany,’ and storedin a refrigerator in the staff room. One day when Iwent to retrieve the samples, a young resident physi-cian was taking a break in the staff room. After seeingme take the marked sample containers, he questionedme intensely about my intentions. He turned out to beOrlando J. Miller, a good friend ever since, who was

on his way for postdoctoral training with L.S. Penroseof the Galton Laboratory in London.

Herbert Lubs was another Yale resident physicianat the time, and he came to my laboratory often. Westarted short-term cultures of leucocytes from periph-eral blood. However, there was little mitotic activityin our cultures because, by following the literature,we used dextrin instead of phytohemagglutinin formitotic stimulation. We missed the boat. At a recentannual meeting of the American Society of HumanGenetics, Herb recalled the many holes he poked intomy arms. While visiting New Haven, Albert Levanshowed us the method of aceto-orcein squash prepa-ration of human chromosomes and the fine art ofdrawing with the aid of a camera lucida. Charles Fordcame to demonstrate meiotic chromosomes. We ob-tained a male rat instead of a mouse from the ZoologyDepartment. To sacrifice the animal, Ford grabbedits tail and smashed it down on the laboratory bench,cracking a glass plate in the process.

Having been awarded a research grant, I was sent toreceive a ‘proper education’ on tissue culture technol-ogy by attending a 4-week course in Denver in June,1956. The course was sponsored by the Tissue CultureAssociation and held in the Department of Microbi-ology of the University of Colorado Medical Center.The training course included laboratories and was wellorganized and taught under the direction of CharlesPomerat and John Paul. A parade of distinguishedspeakers came to address us every morning, coveringa wide variety of subjects but little genetics. TheodorePuck gave one lecture on survival analysis by the sin-gle cell plating technique. However, a visit to his lab-oratory was discouraged. With persistence on my part,I was finally able to obtain a 15min interview withPuck late one Friday afternoon. When the opportunitycame, I started by saying that the ‘mutants’ isolated bycell cloning [18] might well be pre-existing variants inthe heterogeneous HeLa cell line. Puck agreed to sendus the HeLa cells and several representative clonalderivatives for cytological examination. Our findings,confirming our assumption, were published later [19].

As collaborators now, Puck and I agreed to teacheach other certain relevant techniques that were usedin our laboratories. Accordingly, I went to his labo-ratory the next morning to make cytological prepara-tions of HeLa cells. It took us a long while becausewe had to go to different departments in the Medical


Early days of mammalian somatic cell genetics:the beginning of experimental mutagenesis ☆Ernest H.Y. Chu

Department of Human Genetics, University of Michigan, Ann Arbor, MI 48109-0618, USAAccepted 14 March 2003



Early days of mammalian somatic cell genetics:the beginning of experimental mutagenesis�

Ernest H.Y. ChuDepartment of Human Genetics, University of Michigan, Ann Arbor, MI 48109-0618, USA

Accepted 14 March 2003

Keywords: Somatic cell genetics; Mammalian cytogenetics; V79 cells; Chinese hamster cells; Mutagen assays; History of science

Although genetic variation had been known to oc-cur in mammalian somatic cells both in vivo and invitro [1], experimental induction of gene mutationswas not realized until 1968 when reports were madeindependently by three laboratories [2–6]. These andsubsequent studies demonstrated that the frequencyof mutation affecting a variety of phenotypes canbe significantly increased by treatment of cells withchemical and physical agents [7–9]. Thus, the processof mutagenesis in mammalian cells could now bestudied and compared with that in microorganisms.The ability to induce mutations in mammalian somaticcells greatly increases the genetic variability availablefor analysis in these cells. Cell cultures in vitro alsooffered new bioassay systems for mutagenicity testingof agents present in the human environment. Quan-titative determination of spontaneous and inducedsomatic mutations has led to a more precise evalu-ation of genetic hazards of radiation and chemicals.Furthermore, somatic mutation itself plays importantroles in development, immune response, ageing andother cellular phenomena. Hence, mutation studies in

�This article is part of the Reflections in Mutation Researchseries. To suggest topics and authors for Reflections, readers shouldcontact the series editors, G.R. Hoffmann ([email protected]) or D.G. MacPhee ([email protected]).E-mail address: [email protected] (E.H.Y. Chu).

cultured mammalian cells are shedding light on theconsequences of somatic mutation in vivo.

In this article, I recall my experience with regardto the first demonstration of induced gene mutationsin cultured mammalian somatic cells. The work wascarried out in the Biology Division, Oak Ridge Na-tional Laboratory, Oak Ridge, TN. Before doing so,however, I must provide the background and expressmy gratitude to several individuals who inspired andencouraged me in my entry into the new field of so-matic cell genetics.

During the period from 1954 to 1959 in the De-partment of Botany, Yale University, New Haven, CT,Norman Giles encouraged me to explore approachesto the genetic analysis of human and other mam-malian cells in culture, patterning on the techniques ofmicrobial genetics. Single cell plating of certain mam-malian cell lines had been demonstrated [10,11]. Thechance discovery that treatment of mammalian cellsin tissue culture with hypotonic saline could spreadchromosomes [12] marks the dawn of modern mam-malian and human cytogenetics. Mammalian cellspossessing distinct and observable chromosomes areamenable to studies of genotype–phenotype correla-tions. It was reasoned that somatic cells in vitro couldpermit direct experimental manipulation for geneticanalyses that are difficult or impossible to do in vivoin man.

1383-5742/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S1383-5742(03)00030-9


E.H.Y. Chu / Mutation Research 566 (2004) 1–8 5

cells after treatment with MNNG were summarizedin an abstract [4]. It must have been submitted tothe 12th International Congress of Genetics to meetthe February 1 acceptance deadline. The full pa-per was contributed by Puck and appeared in theProceedings of the National Academy of Sciencesof the United States of America in August of thatyear [5].

In the present article I recall my personal experi-ence in exploring the possibility of mutagenesis andgenetic analyses in mammalian somatic cells in vitrowhen this type of investigation was at its buddingstage. Progress at the beginning was necessarily slowas there had been many steps of trial and error. Fromthose early years onward, newer techniques, newerchallenges and remarkable accomplishments by manyinvestigators made the field of mammalian somaticcell genetics rich and interesting. These advances laida foundation for studies in cell hybridization, genemapping, linkage analysis, genomics, the molecularbasis of differentiation and development, carcinogen-esis, and medical genetics. Hsu [45] wrote a delightfulpersonal account of the history of human and mam-malian cytogenetics, including the story of the DenverConference. Readers who are interested in a history ofsomatic cell genetics, starting with the growth of so-matic cells outside of the body and continuing throughcytogenetics, cell fusion, parasexual genetics, and thedevelopment of molecular biology, are referred to thetreatise by Harris [46].In the field of genetic toxicology, mammalian cells

in culture have been used to great advantage fordiverse investigations. One of the earliest demonstra-tions that ultraviolet light induces pyrimidine dimersin the DNA of mammalian cells was accomplished inChinese hamster V79 cells by James Trosko, WilliamCarrier and me at Oak Ridge [47]. My former asso-ciates Chia Cheng Chang and James Trosko, togetherwith their collaborators and students [48], further ap-plied the Chinese hamster cell system to detect epige-netic toxicants. They co-cultivated wild type Chinesehamster V79 cells with 6-thioguanine-resistant mu-tant derivatives in the presence of 6-thioguanine, withor without nonmutagenic chemicals that were knownor suspected of being teratogens, tumor promoters,reproductive toxicants or neurotoxicants. This assaybecame known as the “metabolic cooperation” assayto detect epigenetic toxicants [49].

The experimental induction of gene and chro-mosomal mutations in V79 cells has continued toyield information of fundamental and practical im-portance. A sample of recent literature illustrates thebreadth and depth of these accomplishments, e.g.: (i)isolation and characterization of mutagen-sensitiveand DNA-repair mutants of Chinese hamster V79cells [50]; (ii) Chinese hamster V79 UV-sensitiveand hypermutable aphidicolin-resistant DNA poly-merase � mutants [48,51]; (iii) the effect of damagedDNA-binding protein on UV resistance in V79 cells[52]; (iv) molecular and cellular influences of buty-lated hydroxyanisole on MNNG-treated V79 cells[53]; (v) clastogenic, DNA-intercalative and topoi-somerase II-interactive properties of bioflavonoids inV79 cells [54]; (vi) chromosome aberration yields inV79 cells after high LET radiation [55]; and (vii) min-isatellite and HPRT mutations in V79 cells irradiatedwith helium ions and � rays [56].

It has been gratifying to observe how other re-searchers have used the V79 mutagencity assay sys-tem through the years. In a little more than a decadeafter our description of the assay in 1968, it was usedto evaluate many dozens of chemicals for mutagenic-ity, and the methods were refined for application inlarge-scale genotoxicity testing [57]. That use has con-tinued through the years, such that the National Li-brary of Medicine’s PubMed database now containsmore than 2500 citations to studies in V79 cells, manyhundreds of which concern chemical mutagenesis. Iam especially pleased as I read the current scientific lit-erature to see that a new generation of scientists is find-ing ever more creative ways to take advantage of thefoundation laid years ago. An exciting development inthe V79 mutation system is the incorporation of cy-tochrome P450 activity for metabolizing xenobiotics[58]. Additional screening tests have also been assim-ilated into the V79 mutagenicity assay, such as an invitro transformation assay [59]. Thus, we can surelylook forward to even more exciting developments inthe V79 Chinese hamster system in the years ahead.

I was fortunate from the very beginning to have hadmentors, colleagues and collaborators who were pa-tient and generous in giving me advice and support.Coworkers, students and visitors in my laboratoriesin Oak Ridge and Michigan have often offered sug-gestions and assistance, because I have always beenopen in sharing ideas, materials and information. My


Eva Klein at the Institute of Tumor Biology, Karolin-ska Institutet, Stockholm, Sweden. A paper on im-munological variation and interaction between normaland tumor human cells in culture resulted from thiscollaboration [37].

Upon returning to Oak Ridge, I tried to induceboth chromosomal and gene mutations experimen-tally in aneuploid cell lines derived from the Chinesehamster. The cell material was chosen because of itslow, near diploid chromosome number (2n = 22),distinct chromosome morphology and almost per-fect cell plating efficiency. In the spring of 1967,repeated tries were made to induce gene mutationsin Chinese hamster V79 lung cells. The origin, cul-ture procedures and properties of V79 cells havebeen described elsewhere [38]. Two genetic markerswere chosen: one controlling sensitivity and resis-tance to 8-azaguanine, the other glutamic acid andglutamine utilization. Although we studied both for-ward mutation and reversion at both loci, we basedout studies of chemical mutagenesis on forward mu-tations from azaguanine sensitivity to resistance andreversion from glutamine auxotrophy to prototrophy.The chemical mutagens used were ethyl methane-sulfonate (EMS), methyl methanesulfonate (MMS)and N-methyl-N�-nitro-N-nitrosoguanidine (MNNG).These mutagens were used because of their potent mu-tagenicity and their use in parallel experiments withNeurospora [39,40], mice [41] and other organisms insister laboratories within the Biology Division. It wasno accident that the Environmental Mutagen Societyof the USA was founded in Oak Ridge in 1969 andthat the Environmental Mutagen Information Centerwas established there soon thereafter.

In several repeat experiments, the number of vari-ants in mutagen-treated cultures was not significantlydifferent from that in the untreated controls. In frustra-tion, I went to chat with Heinrich Malling, as I oftendid for consultation with him and other colleagues inthe Biology Division. On a small blackboard in histiny office, I sketched the design of a typical muta-genesis experiment. Malling immediately spotted thecritical mistake I had committed. He simply remindedme about the process of ‘error replication’ in cellsafter their exposure to a mutagen [42]. Biochemicalevidence indicates that alkylating agents, such as EMSand MMS, react with the four nucleotide bases ofDNA in the following order: guanine (G) > adenine

(A) > cytosine (C) > thymine (T) [43]. The pre-dominant mutagenic effect of alkylation is typicallybase-pair transition from G:C to A:T [44]. The pair-ing error theory [44] requires that before the inducedmutant phenotype can appear in the cell population, atleast two rounds of DNA replication must occur afteralkylation. In other words, if a nucleotide base in theDNA of the treated cell was modified by the mutagen,the cell phenotype still remains wild type, and the cellwill be killed when exposed to a selective agent. Theexperimental error of subjecting mutagen-treated cellsto selection too early was probably also committedby other investigators [28].

Immediately upon the realization of this mistake inmy earlier experiments, I jumped up, thanked Malling,and rushed back to do more work. The essential resultswere first presented at the 12th International Congressof Genetics in Tokyo [2], and they were publishedin full in December of the same year [3]. When thedraft of the manuscript was completed, I showed it toMalling and asked his consent to be my co-author, notbecause of his participation in actual experimentation,but because of his critical remark that led to its success.I was most pleased that he agreed to lend his name tothe authorship.

Publication of the article in the Proceedings of theNational Academy of Sciences of the United States ofAmerica was delayed, in part due to comments madeby one reviewer. According to this reviewer, our ex-periments were carefully performed, and the resultsappeared to be valid, well analyzed and presented.Nevertheless, the paper was thought to have only lim-ited significance because chemical mutagenesis hadbeen demonstrated in other organisms. We could notreally disagree with these comments, but prevention ofpublication of our results might postpone the progressof this young science of somatic cell genetics. Follow-ing another round of reviews, Alexander Hollaender,Director of the Biology Division, agreed to communi-cate the article for publication in the Proceedings [3].

In November 1967, I went to Denver to attendthe annual meeting of the American Society of CellBiology. I visited Puck and shared with him andhis coworkers our results on chemical mutagene-sis in Chinese hamster V79 cells. They were soonable to confirm our findings in another line of Chi-nese hamster cells (CHO). Their results showing theisolation of nutritionally deficient mutants in CHO


E.H.Y. Chu / Mutation Research 566 (2004) 1–8 7

[24] T.T. Puck, The mammalian cell, in: J. Cairns, G.S. Stent, J.D.Watson (Eds.), Phage and the Origins of Molecular Biology,Cold Spring Harbor Laboratory Press, Cold Spring Harbor,New York, 1966, pp. 275–286.

[25] F.J. Ryan, Applicability of approaches of microbial geneticsto characterization of mammalian cells, in: D.J. Merchant,J.V. Neel (Eds.), Approaches to the Genetic Analysis ofMammalian Cells, University Michigan Press, Ann Arbor,MI, 1962, pp. 1–11.

[26] R. De Mars, J.L. Hooper, A method of selecting auxotrophicmutants of HeLa cells, J. Exp. Med. 111 (1960) 559–572.

[27] R.S. Krooth, A. Weinberg, Properties of galactosemic cells inculture, Biochem. Biophys. Res. Commun. 3 (1960) 518–524.

[28] W. Szybalski, E.H. Szybalska, G. Ragni, Genetic studies withhuman cell lines, in: Analytic Cell Culture, National CancerInstitute Monograph No. 7, 1962, pp. 75–79.

[29] K. Paigen, The genetic control of enzyme activity duringdifferentiation, Proc. Natl. Acad. Sci. U.S.A. 47 (1961) 1641–1649.

[30] L.W. Law, Origin of resistance of leukemic cells to folic acidantagonists, Nature 169 (1952) 628–629.

[31] I. Lieberman, P. Ove, Estimation of mutation rates withmammalian cells in culture, Proc. Natl. Acad. Sci. U.S.A. 45(1959) 872–877.

[32] K.C. Atwood, S.L. Scheinberg, Some variation in humanerythrocyte antigens, J. Cell. Physiol. 52 (Suppl.) (1958) 97–123.

[33] K.C. Atwood, The presence of A2 erythrocytes in A1 blood,Proc. Natl. Acad. Sci. U.S.A. 44 (1958) 1054–1057.

[34] W. Szybalski, Genetics of human cell lines. II. Methods fordetermination of mutation rates to drug resistance, Exp. CellRes. 18 (1961) 588–591.

[35] S. Sorieul, B. Ephrussi, Karyological demonstration ofhybridization of mammalian cells in vitro, Nature 190 (1961)653–654.

[36] R.S. Krooth (Ed.), Somatic Cell Genetics, UniversityMichigan Press, Ann Arbor, MI, 1964.

[37] E.H.Y. Chu, J. Stjernwärd, P. Clifford, G. Klein, Relativityof human lymphocytes against autochtonous and allogeneicnormal and tumor cells in vitro, J. Natl. Cancer Inst. 39(1967) 595–617.

[38] E.H.Y. Chu, P. Brimer, K.B. Jacobson, E.V. Merriman,Mammalian Cell Genetics. I. Selection and characterizationof mutations auxotrophic for l-glutamine or resistant to8-azaguanine in Chinese hamster cells, Genetics 62 (1969)359–377.

[39] H.V. Malling, F.J. de Serres, N-Methyl-N′-nitro-N-nitro-soguanidine (MNNG) is a mutagenic agent for Neurosporacrassa, Genetics 56 (1967) 575.

[40] H.V. Malling, F.J. de Serres, Identification of geneticalterations induced by ethyl methanesulfonate in Neurosporacrassa, Mutat. Res. 6 (1968) 181–193.

[41] U.H. Ehling, R.B. Cumming, H.V. Malling, Chemicallyinduced dominant lethals in male mice, Genetics 56 (1967)557.

[42] J.W. Drake, The Molecular Basis of Mutation, Holden Day,San Francisco, London, Cambridge, Amsterdam, 1970.

[43] P.D. Lawley, P. Brookes, The action of alkylating agents ondeoxyribonucleic acid in relation to biological effects of thealkylating agents, Exp. Cell Res. Suppl. 9 (1963) 512–520.

[44] E. Freese, Molecular mechnisms of mutation, in: J.H. Taylor(Ed.), Molecular Genetics, Part I, Academic Press, New York,London, 1963, pp. 207–269.

[45] T.C. Hsu, Human and Mammalian Cytogenetics. A HistoricalPerspective, Springer-Verlag, New York, Heidelberg, Berlin,1979.

[46] H. Harris, The Cell of the Body. A History of Somatic CellGenetics, Cold Spring Harbor Laboratory Press, Cold SpringHarbor, New York, 1995.

[47] J.E. Trosko, E.H.Y. Chu, W.L. Carrier, The induction ofthymine dimers in ultraviolet-irradiated mammalian cells,Radiat. Res. 24 (1965) 667–672.

[48] C.C. Chang, J.A. Boezi, S.F. Warren, C.K. Sobourin, P.K.Liu, L. Glazer, J.E. Trosko, Isolation and characterizationof a UV-sensitive hypermutable aphidicolin-resistant Chinesehamster cell line, Somat. Cell Genet. 7 (1982) 235–253.

[49] L.P. Yotti, C.C. Chang, J.E. Trosko, Elimination of metaboliccooperation in Chinese hamster cells by a tumor promoter,Science 206 (1979) 1089–1091.

[50] R.A. Schultz, J.E. Trosko, C.C. Chang, Isolation and partialcharacterization of mutagen-sensitive and DNA repair mutantsof Chinese hamster fibroblasts, Environ. Mutagen. 5 (1981)53–64.

[51] P.K. Liu, C.C. Chang, J.E. Trosko, D.K. Dube, G.M.Martin, L.A. Loeb, Mammalian mutator mutant with anaphidicolin-resistant DNA polymerase alpha, Proc. Natl.Acad. Sci. U.S.A. 80 (1983) 797–801.

[52] N.K. Sun, H.P. Lu, C.C. Chao, Over expression of damagedDNA-binding protein 2 (DDB2) potentiates UV resistance inhamster V79 cells, Chang Gung Med. J. 25 (2002) 723–733.

[53] D. Slamenova, E. Horvathova, S. Robichova, L. Hrusovska,A. Gabelova, K. Kleibl, J. Jakubikova, J. Sedlak, Molecularand cellular influences of butylated hydroxyanisole onChinese hamster V79 cells treated with N-methyl-N′-nitro-N-nitrosoguanidine: antimutagenicity of butylated hydroxyani-sole, Environ. Mol. Mutagen. 41 (2003) 28–36.

[54] R.D. Snyder, P.J. Gillies, Evaluation of the clastogenic, DNAintercalative, and topoisomerase II-interactive properties ofbioflavonoids in Chinese hamster V79 cells, Environ. Mol.Mutagen. 40 (2002) 266–276.

[55] S. Ritter, E. Nasonova, E. Gudowska-Nowak, M. Scholz, G.Kraft, Integrated chromosome aberration yields determinedfor V79 cells after high LET radiation, Int. J. Radiat. Biol.78 (2002) 1063–1064.

[56] R. Cherubinit, S. Canova, S. Favaretto, V. Bruna, P. Battivelli,L. Celotti, Minisatellite and HPRT mutations in V79 cellsirradiated with helium ions and gamma rays, Int. J. Radiat.Biol. 78 (2002) 791–797.

[57] M.O. Bradley, B. Bhuygan, M.C. Francis, R. Langenbach,A. Peterson, E. Huberman, Mutagenesis by chemical agentsin V79 Chinese hamster cells: a review and analysis of theliterature. A report of the Gene-Tox Program, Mutat. Res. 87(1981) 81–142.


involvement in scientific research has afforded me theopportunity to travel to all corners of the world whereI can appreciate the similarities and differences of thepeople and make friends. I have had a lot of fun doingwhat I like to do while making a living. I feel fortu-nate and am enormously indebted to so many individ-uals. In particular, I thank Norman Giles who set meoff in an exciting field of investigation and providedpersonal example as an educator and researcher withvision, dedication and honesty.

Acknowledgements

It is my pleasure to thank George Hoffmann, Hein-rich Malling and James Trosko, all colleagues andfriends since the Oak Ridge days, for advice duringthe preparation of this article. I record here the smallway in which I was fortunate to have participated inthe early history of mammalian somatic cell genetics.Lewontin [60] in a book review commented: “Scien-tists have written memoirs, and a few have consideredthe philosophical implications of scientific discoveriesof which their own work has been a part. They have notwritten of their work as part of an historical movementnor have they preened themselves self-consciously be-fore the mirror of history.” They do so because “theyare, after all, as vain as other men, but their concernfor their place in history has not been flaunted.” Inthis article, on the contrary, it is not my intention toevaluate my contributions but to sincerely acknowl-edge the guidance and inspiration I have received fromso many individuals in the course of my own learn-ing process and to indicate my delight of interactionwith them.

References

[1] M. Harris, Cell Culture and Somatic Variation, Holt, Rinehartand Winston, New York, 1964.

[2] E.H.Y. Chu, H.V. Malling, Chemical mutagenesis in Chinesehamster cell in vitro, in: Proceedings of the 12th InternationalCongress of Genetics, vol. 1, 1968, p. 102.

[3] E.H.Y. Chu, H.V. Malling, Mammalian cell genetics. II.Chemical induction of specific locus mutations in Chinesehamster cells in vitro, Proc. Natl. Acad. Acad. Sci. U.S.A.61 (1968) 1300–1312.

[4] F.T. Kao, T.T. Puck, Isolation of nutritionally deficientmutants of Chinese hamster cells, in: Proceedings of the 12thInternational Congress of Genetics, vol. I, 1968, p. 157.

[5] F.T. Kao, T.T. Puck, Genetics of somatic mammalian cells.VII. Induction and isolation of nutritional mutants in Chinesehamster cells, Proc. Natl. Acad. Sci. U.S.A. 60 (1968) 1275–1281.

[6] N.I. Shapiro, O.N. Petrova, A.E. Khalizev, Induction of genemutations in mammalian cells in vitro, Technical ReportIAE-1782, Institute of Atomic Energy, Moscow, USSR, 1968.

[7] E.H.Y. Chu, Mammalian Cell Genetics. III. Characterizationof X-ray-induced forward mutations in Chinese hamster cellcultures, Mutat. Res. 11 (1971) 23–34.

[8] E.H.Y. Chu, 1972. Mutagenesis in mammalian cells, in:Environment and Cancer, Williams & Wilkins, Baltimore,pp. 198–213.

[9] T.T. Puck, The Mammalian Cell as a Microorganism, HoldenDay, San Francisco, London, Cambridge, Amsterdam, 1972.

[10] T.T. Puck, P.I. Marcus, S.J. Cieciura, Clonal growth ofmammalian cells in vitro. Growth characteristics of coloniesfrom single HeLa cells with and without a ‘feeder’ layer, J.Exp. Med. 103 (1956) 273–284.

[11] T.T. Puck, Quantitative studies on mammalian cells in vitro,Rev. Mod. Phys. 31 (1959) 433–448.

[12] T.C. Hsu, C.M. Pomerat, Mammalian chromosomes in vitro.II. A method for spreading the chromosomes of cells in tissueculture, J. Hered. 44 (1953) 23–29.

[13] E.H.Y. Chu, N.H. Giles, A study of primate chromosomecomplement, Am. Nat. 41 (1957) 273–282.

[14] R. Kleinfeld, E.H.Y. Chu, DNA determinations of kidney cellcultures of three species of monkeys, Int. J. Cytol. 23 (1958)452–459.

[15] E.H.Y. Chu, N.H. Giles, Chromosome complement andnucleoli in normal human somatic cells in culture, in:Proceedings of the 10th International Congress of Genetics,vol. II, 1958, p. 50.

[16] E.H.Y. Chu, N.H. Giles, Human chromosome complementin normal somatic cells in culture, Am. J. Hum. Genet. 11(1959) 63–79.

[17] E.H.Y. Chu, N.H. Giles, K. Passano, Types and frequenciesof human chromosome aberrations induced by X-rays, Proc.Natl. Acad. Sci. U.S.A. 47 (1961) 830–839.

[18] T.T. Puck, H.W. Fisher, Genetics of somatic mammalian cells.I. Demonstration of the existence of mutants with differentgrowth requirements in a human cancer cell strain (HeLa), J.Exp. Med. 104 (1956) 427–434.

[19] E.H.Y. Chu, N.H. Giles, Comparative chromosome studies onmammalian cells in culture. I. The HeLa strain and its mutantclonal derivatives, J. Natl. Cancer Inst. 20 (1958) 383–401.

[20] J.H. Tjio, A. Levan, The chromosome number of man,Hereditas 42 (1956) 1–6.

[21] J.H. Tjio, T.T. Puck, The chromosomes of man and othermammals, in: Proceedings of the 10th International Congressof Genetics, 1958, p. 61.

[22] J.H. Tjio, T.T. Puck, The somatic chromosomes of man, Proc.Natl. Acad. Sci. U.S.A. 44 (1959) 1229–1237.

[23] Denver Report, A proposed standard system on nomenclatureof human mitotic chromosomes, J. Hered. 51 (1960) 221–241.



[58] B. Oesch-Bartlomowicz, F. Oesch, Cytochrome P450phosphorylation as a functional switch, Arch. Biochem.Biophys. 409 (2003) 228–234.

[59] A. Sakai, Y. Iwase, Y. Nakamura, K. Sasaki, N. Tanaka, M.Umeda, Use of a cell transformation assay with establishedcell lines, and a metabolic cooperation assay with V79 cells

for the detection of tumour promoters: a review, AlternativesLab. Anim. 30 (2002) 33–59.

[60] R.C. Lewontin, in: J. Cairns, G.S. Stent, J.D. Watson (Eds.),Essay Review of Phage and the Origins of Molecular Biology,Cold Spring Harbor Laboratory Press, Cold Spring Harbor,New York; J. Hist. Biol. 1 (1968) 155–161.


Incorporation of mammalian metabolisminto mutagenicity testing ☆Heinrich V. Malling *

Mammalian Mutagenesis Group, Laboratory of Toxicology, Environmental Toxicology Program, National Institute of Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709-2233, USAAccepted 16 November 2003



Incorporation of mammalian metabolisminto mutagenicity testing�

Heinrich V. Malling∗Mammalian Mutagenesis Group, Laboratory of Toxicology, Environmental Toxicology Program, National Institute of

Environmental Health Sciences, P.O. Box 12233, Research Triangle Park, NC 27709-2233, USA

Accepted 16 November 2003

Abstract

In the 1950’s and 1960’s it became obvious that many chemicals in daily use were mutagenic or carcinogenic, but thereseemed to be little relation between the two activities. As scientists were debating the cause of this discrepancy, it washypothesized that mammalian metabolism could form highly reactive intermediates from rather innocuous chemicals andthat these intermediates could react with DNA and were mutagenic. This commentary presents the historical developmentof metabolic activation in mutagenicity tests, beginning with Udenfriend’s hydroxylation system, which mimics aspectsof mammalian metabolism in a purely chemical mixture, and extending through procedures that moved closer and closerto incorporating actual mammalian metabolism into the test systems. The stages include microsomal activation systems,host-mediated assays, incorporation of human P450 genes into the target cells or organisms, and detecting mutations in singlecells in vivo. A recent development in this progression is the insertion of recoverable vectors containing mutational targetsinto the mammalian genome. Since the target genes of transgenic assays are in the genome, they are not only exposed to activemetabolites, but they also undergo the same repair processes as endogenous genes of the mammalian genome.© 2003 Elsevier B.V. All rights reserved.

Keywords: Metabolic activation; Mutagen assays; Cytochrome P450; Mutagen–carcinogen correlation; Host-mediated assay; Somatic cellmutagenesis; Transgenics

1. Introduction

Mutations in mammalian somatic cells or germcells often lead to a decrease in the fitness of the or-ganism or its offspring. Mutations in dividing somaticcells may cause cancer, whereas mutations in germcells may result in defective offspring. We can spec-


∗ Tel.: +1-919-541-3378; fax: +1-919-541-4634.E-mail address: [email protected] (H.V. Malling).

ulate that somatic mutations in non-dividing cells,such as neurons, may be a cause of sporadic mentaldisorders such as Kuru or Alzheimer’s disease [1,2].Somatic cell mutagenesis has also been implicatedin arteriosclerosis [3]. The frequency of mutationsin somatic cells is likely to be high enough that anyhuman being is a mosaic of cells with slightly differ-ent genotypes. Humans experience both spontaneousmutations and induced mutations, and we are exposedto mutagens throughout life. Sources of mutagenexposure include toxicants arising in nature, dietaryconstituents, pollutants, industrial chemicals, drugs,household products, and others. While most chemicalcompounds are relatively benign, strong mutagens

1383-5742/$ – see front matter © 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.mrrev.2003.11.003


184 H.V. Malling / Mutation Research 566 (2004) 183–189

are found among chemicals from all of the abovesources.

There is a great discrepancy between the impor-tance of mutations and the adequacy of the methodsthat we have developed to detect and study them.Despite decades of research on mechanisms of mu-tagenesis and mutation assays, our ability to estimatespontaneous and induced mutation frequencies in vivoremains primitive. It is simply not possible to screenfor mutations representing the whole genome in asufficient number and diversity of cells to obtain reli-able mutation frequencies. We are therefore forced torely on target genes in specific cells. The ideal targetswould be two dream genes that lend themselves to theselection of mutations and that exhibit the most com-mon features of genes in humans. One of the genesshould be transcribed in all tissues, and the othershould be almost always silent. Present systems fordetecting somatic or germinal mutations are greatlyinferior to dream systems of this sort, and it seemsmost unlikely that one will emerge in the foreseeablefuture. For the time being at least we will need to becontent with systems that are a long way short of ideal.

2. Non-mutagenic carcinogens and mammalianmetabolism

In reflecting on the state of mutation assays andtheir development, I am struck by the influence of ar-guments raised by Burdette [4] a half century ago.In his extensive review of carcinogenic mechanisms,Burdette stated that there was no conclusive evidencefor a correlation between mutagenicity and carcino-genicity. One of the cornerstones in his argument wasthe lack of mutagenicity of potent carcinogens suchas dimethylnitrosamine (DMN) and benzo[a]pyrene(BP). The clonal nature of tumors as proliferations ofdeviant cells made it difficult for scientists studyingmutagenesis to accept Burdette’s argument. In the late1950’s and early 1960’s it became clear that detoxica-tion mechanisms in organisms, rather than leading ex-clusively to compounds of lesser toxicity, sometimesproduced reactive metabolic intermediates that couldform adducts with DNA and cause mutations.

Fig. 1 reviews the sequence of events in the devel-opment of experimental approaches for incorporatingmammalian xenobiotic metabolism into chemical

mutagenesis testing. I was involved in much of thiswork, ranging from the initial in vitro metabolic ac-tivation of carcinogens into mutagens, through theformation of mutagens by microsomal metabolic acti-vation systems and host-mediated assays, to the studyof mutagens in transgenic animals. In this Reflectionsarticle, I relate the accomplishments that my col-leagues and I made during our research into the broadsubject of linkages between mammalian metabolismand mutagenesis.

Even at the early stages of this research, themetabolism issue could be divided into two questions:(1) Can mammalian metabolism form mutagenicintermediates from non-mutagenic compounds? (2)Does treatment of a mammal with carcinogens causean increase in somatic mutations? The first of thesequestions proved easier to tackle than the second.

2.1. Metabolic activation in vitro

Several scientists addressed the first question in theearly 1960’s. Among them were James and Eliza-beth Miller, who isolated active intermediates formedmetabolically from several carcinogens [5]. A diffi-culty encountered in studying reactive derivatives ofcarcinogens is that they can be extremely unstable,such that their chemical structures can be predicted,but they cannot be isolated. One of the first steps inthe metabolic pathway of many xenobiotics is oxi-dation or hydroxylation by various cytochrome P450enzymes. The isolation of active microsomes was akey step in studying these reactions.

In the 1960’s I worked in the Biology Divisionof Oak Ridge National Laboratory, where NormanAnderson had pioneered the use of ultracentrifugesfor biological use. ORNL was a natural place for thisdevelopment to occur, given that the idea of puttingultracentrifuges to work on biological problems wasa direct consequence of their use in the productionof enriched uranium, one of the routine tasks thatbecame familiar to a great many of the ORNL staff.I shudder to think about my many unsuccessful at-tempts to make active microsomes. That was not theonly problem facing us. We also faced questions likethe following: (1) Are the metabolic intermediatesof DMN sufficiently stable to penetrate into a testorganism? (2) If an active metabolite can penetrate,will it react with DNA? and (3) Could the mutation

H.V. Malling / Mutation Research 566 (2004) 183–189 185

Fig. 1. Flow diagram of the incorporation of mammalian metabolism into the testing of carcinogens for mutagenicity.

systems existing at the time detect the specific geneticalterations that result from the DNA damage induced?

By that time Udenfriend et al. [6] had constructedan in vitro system for mimicking oxidative processesin metabolism. It occurred to me that his system of-fered a means of approaching some of the thornyquestions surrounding carcinogens considered to benon-mutagenic. I applied the Udenfriend system andwas able to show that the non-mutagenic carcinogenDMN formed mutagenic intermediates by hydroxyla-tion [7]. At about the same time, Harry Gelboin andothers were making great progress in purifying mi-crosomes, and they established the co-factor require-

ments for their metabolic activity [8]. A visit to Gelboin’s laboratory was a great help. With the rightco-factors added to the microsome suspension we wereable to show using a bacterial mutation assay that mi-crosomes could form mutagenic intermediates fromnon-mutagenic chemicals [9]. In the 1970’s, the prin-ciple of mixing a microsomal metabolic activation sys-tem with a genetic assay for mutagens was extendedfrom bacteria [9] to Neurospora [10] and culturedmammalian cells [11].It was not all smooth sailing; our first attempts at

microsomal activation with mammalian cells were atotal failure. The microsomes were toxic. Accidently,



we discovered that freezing took care of that problem.Science progresses incrementally. While my bacterialstudies clearly showed metabolic activation, it was notuntil Bruce Ames got the idea of incorporating the mi-crosomes into agar medium in his bacterial assay, thuscreating the well-known bacterial plate test, that muta-gen testing with microsomes became routinely useful[12]. This line of study was later carried to the next log-ical stage by the incorporation of human cytochromeP450 genes into the indicator cells of mutagen as-says, both in bacteria [13,14] and in mammalian cells[15].

3. Host-mediated assays

The host-mediated assay [16] was developedconcurrently with the development of microsomalmetabolic activation systems. In this method, the mi-croorganisms of a mutation assay were moved fromthe test tube and petri dish into the intact mammal.An indicator organism (e.g. bacteria or Neurospora)was injected into the animal. When the animal wastreated with a chemical, the indicator organism wasexposed, in principle, not only to the chemical itselfbut also to its mammalian metabolites. The indicatororganism could then be removed from the animaland assayed for induced mutations. Host-mediatedassays had a theoretical advantage over microsomalmetabolic activation, in that it was not feasible toincorporate all the metabolic pathways for metabolicactivation of chemicals in an intact animal into thetest tubes or petri dishes of an in vitro system.

In the initial version of the host-mediated assaydescribed by Gabridge and Legator [16], the indica-tor organism was placed in the peritoneal cavity. Al-though ingenious, the technique was limited by thefact that the peritoneal cavity was too remote fromthe site of formation of active intermediates, princi-pally the liver. To overcome this problem the intra-venous host-mediated assay was developed [17–19].The indicator organisms were injected into the bloodstream and later recovered from several different or-gans. Animals treated with carcinogens now showedorgan-specific mutagenesis [17,19]. Though a usefulresearch tool, the host-mediated assay did not fulfillthe most optimistic expectations that it could mimic anatural in vivo mammalian system.

As is so often the case, nature was able to throwa monkey wrench into the wheels of progress. Theenvironment in a mammal could be highly mutagenicto foreign organisms, leading to high mutation fre-quencies in the controls [20]. The highly mutagenicenvironment inside laboratory mice and rats was soastonishing to us that John Wassom and I did morethan 100 experiments measuring ad3 mutants in bigjugs of Neurospora to convince ourselves that ourobservation was correct. After 36 h of incubationin intact mammals, the spontaneous mutation fre-quencies of Neurospora conidia were found to haveincreased by 60-fold in rats and 10-fold in miceabove the typical spontaneous frequencies. Follow-upstudies revealed that there are at least two agents inrodents that are implicated in mutagenesis in intraperi-toneally injected Neurospora conidia. The majoragent requires cellular contact or is a macromolecule,and the minor agent has low molecular weight, inthat it can penetrate through dialysis tubing. Thismutation-induction phenomenon may be an importantconsideration in many different types of experiments[21].

4. Mutagenesis in vivo and transgenic systems

Neither in vitro metabolic activation systems nor thehost-mediated assay encompassed the mammalian ca-pacity for DNA repair. Most DNA damage is repairedcorrectly and therefore unimportant for the fitness ofthe organism, although increased levels of DNA dam-age can result in increased numbers of fixed mutations.To incorporate the repair capacity of the mammal intothe test system, the next logical step was to incorpo-rate the target gene of an indicator organism into themammalian genome. The ultimate host-mediated as-say sprang out of these thoughts—transgenic systems[22,23].

Transgenic systems based on recoverable vectorswere developed in several laboratories for the studyof mutagenesis in mammals, and I had the pleasure ofbeing involved in the earliest stages of this endeavor[22]. Shortly after I moved from Oak Ridge to theNational Institute of Environmental Health Sciencesin Research Triangle Park, NC, I met Dr. Phil Chen ina cafeteria on the NIH campus in Bethesda. He askedme what I hoped to accomplish at NIEHS. My answer


was to develop a system that can be used to measuremutations in any tissue of a mouse.In 1980, when I finally came up with the idea that

transgenic systems were the way to go, I discussedthe idea with my NIEHS colleagues Chuck Langley,who recommended that I use phage lambda, and BurkJudd, who suspected that it could not be done, ow-ing to the possible presence of toxic sequences inphages. At that time, viable phage could only be re-covered with very low efficiency from lambda DNA.However, �X174 could be rescued from �X DNAwith an efficiency of 1–5%. I was lucky enough tohave two renowned �X scientists, Professors Mar-shall Edgell and Clyde Hutchison, in the neighbor-hood at the University of North Carolina in ChapelHill. I contacted them to explore the possibility of myspending a sabbatical at UNC, and thus began a re-warding collaboration. It turned out that they had al-ready developed an interest in transgenics, and a post-doc in their lab, Steve Hardies, had made transgenicmouse L cells. At least that took care of the questionof toxic sequences. I set out to rescue �X from themouse L cells. If I could do that, the road was clearto making the transgenic mice. The rescue succeeded[22,24].

Direct measurement of spontaneous and inducedmutations in the somatic cells of mammals in vivobegan to take form in the early 1970’s with many at-tempts and many failures. Among the dead ends weresystems based on fetal hemoglobin and glucose-6-pho-sphate dehydrogenase, proposed by Sutton [25,26],and on lactic acid dehydrogenase-X, attempted byAnsari et al. [27]. These and other systems failed dueto the presence of phenocopies, which are cells or in-dividuals with apparently mutant phenotypes but nor-mal genotypes. However, the second generation of invivo systems, notably including the glycophorin sys-tem [28] and the hprt system [29], succeeded. Theglycophorin system is based on direct observation ofthe M/N serotypes in red blood cells; the absence ofnuclei in these cells precluded the molecular charac-terization of the genetic alterations. This factor led tothe preeminence of Albertini’s hprt system among so-matic cell mutation assays [29,30]. Despite the powerof the newest somatic cell assays, there are still lim-itations in measuring and characterizing mutations insomatic cells, notably an inability to recover and cul-ture mutant cells from many organs. Transgenic mu-

tation systems, based on mutation detection in recov-ered vectors, do not require growth of the cells in vitroto measure and identify the mutations.

Phages and plasmids have both been used as vec-tors in transgenic mutation systems [22,23,31]. Thetarget can be constructed so as to detect either forwardmutations or reverse mutations. There are advantagesand disadvantages to each, owing in part to the factthat some chemicals are rather specific in their muta-genic action. Forward mutation systems may not de-tect chemicals with low and specific mutagenic activ-ity. Reverse mutation systems will be very sensitiveto chemicals that happen to induce the type of geneticalteration that leads to a revertant but will not detectother genetic alterations. In choosing the mutations onwhich to base our transgenic assay, we followed thelead of the most successful in vitro microbial mutationsystems and planned the assay around the detection ofrevertants.

The transgenic system that we are using is based on�X174 am3, cs70. Bacteriophage �X174 is a smallphage, having a genome of 5386 base pairs [32,33].The am3 mutation is a nonsense mutation that causesincomplete protein synthesis in the lysis gene (geneE), which overlaps with another essential gene (geneD). The am3 allele has the triplet TAG, specifying anamber codon, where the wild-type allele has the nu-cleotide sequence TGG. The genetic alteration in re-verse mutations only involves the center adenine ofthe target triplet. One transition and two transversionscan result in phage particles with wild-type pheno-type. Reverse mutation frequencies are measured byplating on an indicator bacterial strain that only sup-ports the growth of wild type �X174. Fifty tandemcopies of the phage have been incorporated into themouse haploid genome, and a strain homozygous forthis insert was established at NIEHS. The limited tar-get in the mutation system and the ease by whichmutations are detected makes the system suitable forobtaining a precise understanding of the mutagenicefficiency of adducts and for studying the pharma-cokinetics of mutation induction in whole animalsin vivo. A secondary advantage, consistent with theaims of the NIH Revitalization Act of 1993 (section1301, Public Law 103-43), is that the implementationof transgenic systems offers the possibility of reduc-ing the numbers of animals required in toxicologicaltesting.



5. Perspective

My odyssey from microbial mutation assays,through in vitro metabolic activation and host-mediatedassays, to transgenic assays has led me to believe thatthe only general systems presently available for study-ing mutagenesis in vivo in mammals are based ontransgenic animals with recoverable vectors insertedin their genome. That is not to say that transgenicassays are without pitfalls. Most, if not all, transgenicmutation systems have high spontaneous mutationrates compared to endogenous genes. Even more dis-turbing is evidence that most of the spontaneous mu-tations are due to in vitro events and are therefore ofno interest with respect to mutagenesis in mammals.

The assay in mice transgenic for �X174 am3, cs70clearly separates in vivo from in vitro and ex vivomutagenesis [34]. Another advantage is that �X174does not contain the GATC sequence and is thereforenot subject to mismatch repair of DNA damage [35].In Escherichia coli mismatch repair in GATC-bearingconstructs can result in complete mutations rather thanthe original sequence, and these would be an artifac-tual indication of in vivo mutagenesis [36]. Although�X174 is not the only vector that can be used to studymutagenesis in mammals in vivo, it has taught us threeprinciples that I think should be sought in any trans-genic assay: (1) a single burst assay should be used;(2) the bacterial cell used for the rescue of phage frommammalian DNA should release only a minimal num-ber of phages; and (3) the released phages should notbe able to reinfect the bacterial cells used to rescuethe phage DNA.

Systems for testing chemicals for mutagenicity havecome and gone through the years. Older scientists inthis field remember the Allium cepa and Vicia fabaroot-tip assays [37]. Until the 1960’s, hundreds ofchemicals had been tested for clastogenic effects inroot tips, and cytogenetic analysis in these two organ-isms was the most used method of mutagenicity test-ing at the time. Yet many young scientists of todayhave never heard of these assays. The demise of theroot-tip system for use in mutagenicity testing camefrom many sides, not the least being the Mrak report,a comprehensive study of pesticides published in 1969[38]. A summary of the root-tip literature showed thatthere was, in general, no difference in chromosomeaberrations between the solvents and the active compo-

nents of pesticide formulations, including some clearlymutagenic pesticides. When Gary Flamm, a scientistworking at the boundary of mutation research and reg-ulatory policy [39], saw the literature compilation, hesaid: “That finishes the root-tip mutation system.”

Surely transgenic mutation systems based on re-coverable vectors will also disappear. What will notdisappear is the necessity of having a keen eye for ar-tifacts or noise in the systems that we use and a clearability to distinguish between the noise and mutationsfixed in vivo. Only the latter mutations have bearingon multicellular organisms, and we still have much tolearn about scoring them correctly.

References

[1] D. Gajdusek, The transmissible amyloidoses: genetical controlof spontaneous generation of infectious amyloid proteinsby nucleation of configurational change in host precursors:kuru-CJD-GSS-scrapie-BSE, Eur. J. Epidemiol. 7 (1991) 567–577.

[2] C. Lendon, F. Ashall, A. Goate, Exploring the etiology ofAlzheimer’s disease using molecular genetics, J. Am. Med.Assoc. 277 (1997) 825–831.

[3] M.G. Andreassi, N. Botto, M.G. Colombo, A. Biagini, A.Clerico, Genetic instability and arteriosclerosis: can somaticmutations account for the development of cardiovasculardiseases? Environ. Mol. Mutagen. 35 (2000) 265–269.

[4] W.J. Burdette, The significance of mutations in relation to theorigin of tumors: a review, Cancer Res. 15 (1955) 201–226.

[5] J.A. Miller, J.W. Cramer, E.C. Miller, The N- and ringhydroxylation of 2-acetylaminofluorene during carcinogenesisin the rat, Cancer Res. 20 (1960) 950–962.

[6] S. Udenfriend, C.T. Clark, J. Axelrod, B.B. Brodie, Ascorbicacid in aromatic hydroxylation, I. A model system foraromatic hydroxylation, J. Biol. Chem. 208 (1954) 731–739.

[7] H.V. Malling, Mutagenicity of two potent carcinogens,dimethylnitrosamine and diethylnitrosamine, in Neurosporacrassa, Mutat. Res. 3 (1966) 537–540.

[8] H. Gelboin, Carcinogens, enzyme induction, and gene action,Adv. Cancer Res. 10 (1967) 1–76.

[9] H.V. Malling, Dimethylnitrosamine: formation of mutageniccompounds by interaction with mouse liver microsomes,Mutat. Res. 13 (1971) 425–429.

[10] W. Whong, T. Ong, Metabolic activations of dimethylnitrosa-mine by male and female mice and rats to metabolites muta-genic in Neurospora crassa, Mutat. Res. 61 (1979) 393–398.

[11] C.N. Frantz, H.V. Malling, Bromodexyuridine resistanceinduced in mouse lymphoma cells by microsomal activationof dimethylnitrosamine, J. Toxicol. Environ. Health 2 (1976)179–186.

[12] B.W. Ames, W.E. Durston, F. Lee, Carcinogens are mutagens:a simple test system combining liver homogenates for


activation and bacteria for detection, Proc. Natl. Acad. Sci.U.S.A. 70 (1973) 2281–2285.

[13] P.D. Josephy, L.S. DeBruin, H.L. Lord, J.N. Oak, D.H.Evans, Z. Guo, M.-S. Dong, F.P. Guengerich, Bioactivation ofaromatic amines by recombinant human cytochrome P4501A2expressed in Ames tester strain bacteria: a substitute foractivation by mammalian tissue preparations, Cancer Res. 55(1995) 799–802.

[14] P.D. Josephy, P. Gruz, T. Nohmi, Recent advances in theconstruction of bacterial genotoxicity assays, Mutat. Res. 386(1997) 1–23.

[15] C.L. Crespi, R. Langenbach, K. Rudo, Y.T. Chen, R.L. Davies,Transfection of a human cytochrome P-450 gene into thehuman lymphoblastoid cell line, AHH-1, and use of therecombinant cell line in gene mutation assays, Carcinogenesis10 (1989) 295–301.

[16] M.G. Gabridge, M.S. Legator, A host-mediated assay fordetection of mutagenic compounds, Proc. Soc. Exp. Biol.Med. 130 (1969) 831–834.

[17] H.V. Malling, The host-mediated assay, in: A.G. Motulsky,W. Lentz (Eds.), Proceedings of the Fourth InternationalConference, Vienna, Austria, Excerpta Medica, Amsterdam,2–8 September 1973, pp. 207–211.

[18] H.V. Malling, C.N. Frantz, In vitro versus in vivo activationof mutagens, Environ. Health Perspect. 6 (1973) 71–82.

[19] G. Mohn, J. Ellenberger, Mammalian blood-mediatedmutagenicity tests using a multipurpose strain of Escherichiacoli K-12, Mutat. Res. 19 (1973) 257–260.

[20] H.V. Malling, G.E. Cosgrove, The internal level ofmutagens in animals II, in: F. Vogel, G. Röhrborn (Eds.),Chemical Mutagenesis in Mammals and Man, Springer,Berlin-Heidelberg-New York, 1970, pp. 271–278.

[21] H.V. Malling, Mutation induction in Neurospora crassaincubated in mice and rats, Mol. Gen. Genet. 116 (1972)211–222.

[22] H. Malling, J. Burkhart, M. Fater, S. Hardies, C.Hutchison III, M. Edgell, Development towards the ultimatehost-mediated assay, Environ. Mutagen. 5 (1983) 416.

[23] J.A. Gossen, W.J.F. de Leeuw, C.H.T. Tan, E.C. Zwarthoff,F. Berends, P.H.M. Lohman, D.L. Knook, J. Vijg, Efficientrescue of integrated shuttle vectors from transgenic mice: amodel for studying mutations in vivo, Proc. Natl. Acad. Sci.U.S.A. 86 (1989) 7971–7975.

[24] J.G. Burkhart, H.V. Malling, Mutagenesis of �X174 am3cs70 incorporated into the genome of mouse L-Cells, Mutat.Res. 213 (1989) 125–134.

[25] H. Sutton, Workshop on monitoring of human mutagenesis,Teratology 4 (1971) 103–108.

[26] H.E. Sutton, Somatic cell mutations: birth defects, in:Proceedings of the Fourth International Conference, Vienna,Austria, vols. 2–8, 1974, pp. 212–214.

[27] A. Ansari, M. Baig, H. Malling, In vivo germinal mutationdetection with “monospecific” antibodies against lactate deh-ydrogenase-X, Proc. Natl. Acad. Sci. U.S.A. 77 (1980) 7352–7356.

[28] W.L. Bigbee, Detection of somatic mutations in humanerythrocytes—initial hemoglobin results and development ofglycophorin A as a new marker, Environ. Mutagen. 2 (1980)264.

[29] R.J. Albertini, K.L. Castle, W.R. Borcherding, T-cell cloningto detect the mutant 6-thioguanine-resistant lymphocytespresent in human peripheral blood, Proc. Natl. Acad. Sci.U.S.A. 79 (1982) 6617–6621.

[30] R.J. Albertini, HPRT mutations in humans: biomarkers formechanistic studies, Mutat. Res. 489 (2001) 1–16.

[31] H.J. Martus, M.E. Dolle, J.A. Gossen, J. Vijg, Use oftransgenic mouse models for studying somatic mutations inaging, Mutat. Res. 338 (1995) 203–213.

[32] F. Sanger, G.M. Air, B.G. Barrell, N.L. Brown, A.R. Coulsen,J.C. Fiddes, C.A. Hutchison, P.M. Slocombe, M. Smith,Nucleotide sequence of bacteriophage �X174 DNA, Nature265 (1977) 687–695.

[33] F. Sanger, A.R. Coulson, T. Friedmann, G.M. Air, B.G.Barrell, N.L. Brown, J.C. Fiddes, C.A. Hutchison III,P.M. Slocombe, M. Smith, The nucleotide sequence ofbacteriophage �X174, J. Mol. Biol. 125 (1978) 225–246.

[34] H.V. Malling, R.R. Delongchamp, Direct separation of in vivoand in vitro am3 revertants in transgenic mice carrying the�X174 am3 cs70 vector, Environ. Mol. Mutagen. 37 (2001)345–355.

[35] F. Langle-Rouault, G. Maenhaut-Michel, M. Radman, GATCsequences, DNA nicks and the MutH function in Escherichiacoil mismatch repair, EMBO J. 6 (1987) 121–127.

[36] M. Radman, R. Wagner, Mismatch repair in Escherichia coil,Annu. Rev. Genet. 20 (1986) 523–528.

[37] B.A. Kihlman, Actions of Chemicals on Dividing Cells,Prentice-Hall, Englewood Cliffs, NJ, 1966.

[38] H.V. Malling, J.S. Wassom, Tabulation of the mutageniceffect of certain pesticides with supporting bibliographicalreferences, in: Mrak Report: Report of the Secretary’s Com-mission on Pesticides and their Relationship to EnvironmentalHealth, Parts I and II, US DHEW, 1969, pp. 611–653.

[39] W.G. Flamm, Observations at the interface of mutationresearch and regulatory policy, Mutat. Res. 544 (2003) 1–7.

mutation research 488 (2001) 195–209 forging the links...

Documents