STUDIES ON THE INDUCTION OF THYMINE DEFICIENCY AND ON THEEFFECTS OF THYMINE AND THYMIDINE ANALOGUES

IN ESCHERICHIA COLIPSEYMOUR S. COHEN AND HAZEL D. BARNER

Department of Pediatrics, The Children's Hospital of Philadelphia, and the Department of Biochemistry,School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania

Received for publication September 26, 1955

We have shown that a thymine-requiring strainof Escherichia coli strain 15T loses the power tomultiply, or "dies" when it is permitted tometabolize and grow in the absence of exogenousthymine (Barner and Cohen, 1954). The effectis ascribed to a kind of "unbalanced growth," inthis instance continuing cell syntheses in the ab-sence of a balanced synthesis of normal deoxyri-bose nucleic acid (DNA). Evidence has beensummarized to suggest that many types of bac-tericidal treatments produce their effects in thisway (Barner and Cohen, 1956) and more par-ticularly we have demonstrated that some lethaleffects of ultraviolet radiation are consequencesof unbalanced growth (Cohen and Barner, 1954).

In this paper are presented the results of studieson the use of inhibitors in the blocking of thyminesynthesis and utilization. The phenomenon ofthymineless death is of considerable chemothera-peutic interest if it can be extended to organismsother than 15r-. Therefore, experiments weredesigned to determine whether unbalancedgrowth and death could be induced in bacteriawhich do not normally have an exogenous thy-mine requirement. It also appeared desirable todetermine whether thymine analogues wouldprevent normal DNA synthesis in the presenceof exogenous thymine.We have previously found that bacterial strain

(15r) specifically blocked with respect to nuclearsynthesis as a result of thymine deficiency canafford very useful experimental materials. Forexample, it has been shown with this system thatthe induction of enzyme biosynthesis may pro-ceed in the absence of normal DNA synthesis

1 This research was aided by a grant from theCommonwealth Fund. The authors also wish toacknowledge with gratitude the hospitality of Dr.Edward A. Adelberg of the Department of Micro-biology of the University of California, in whoselaboratory were done some of the experimentsdescribed in this paper.

(Cohen and Barner, 1955). Also it is possible tosynchronize bacterial division in mass cultures of15T. by withholding thymine and later returningit to the complete medium (Barner and Cohen,1955). It was evident that the extension of suchphenomena to other organi would be impor-tant in the development of experimental biologi-cal systems.

MATERIAS AND METHODS

Bacterial strains. Four strains of Escherichiacoli were employed. These included the thymine-requiring mutant, strain 15r, and its nonauxo-trophic parent, strain 15, as well as a cytosine- oruracil-requiring strain WC, and strain B. Thecultivation and properties of these organiminbroth and in some synthetic media have beendescribed (Barner and Cohen, 1954).

Estimation of growth and division. Growth wasfollowed turbidimetrically (Cohen and Barner,1955). Viable count was determined by spreadingsmall aliquots of diluted cultures on broth agarplates.

In all experiments employing antagonists,strain 15r was grown to the appropriate cellconcentration in the presence of thymine. The cul-ture was then chilled and centrifuged and thesediment was washed in the absence of thymine.Finally the bacteria were resuspended and incu-bated in media containing the compounds underexamination.Compounds. Thymidine was prepared in this

laboratory and was found to be identical with asample obtained from the California BiochemicalFoundation. Thymine, uracil, xanthine, and hy-poxanthine were products of Schwarz, Inc.Orotic acid was kindly given to us by Dr. D.Wright Wilson of the University of Pennsylvania.5-Bromouracil was obtained from the KrishellCo. Azathymine, 5-nitrouracil, 5-hydroxyuracil,and thymine ribofuranoside were generously sup-plied by Dr. Jack Fox of the Sloan-Kettering

588

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

INDUCTION OF THYMINE DEFICIENCY

Institute for Cancer Research; 5-hydroxyuracildeoxyriboside and 5-bromouracil deoxyribosidewere kindly given to us by Dr. Donald Visser ofthe University of Southern California, andspongothymidine by Dr. Werner Bergmann ofYale University. Amino acids and vitamins wereobtained from Nutritional Biochemicals.

Growth in mldfanilamidecontaining media. Sulf-anilamide inhibits the growth and multiplicationof E. coli by preventing the incorporation ofp-aminobenzoic acid into folic acid. A folic aciddeficiency prevents the bacterium from synthesiz-ing a large number of essential metabolites whichcontain the one-carbon fragments normallyhandled by the folic acid-containing coenzymes.It was shown that a number of these metabolites(serine, methionine, valine, xanthine and thy-mine) added to the medium containing sulfanila-mide will permit the growth and multiplicationof E. coli (Rutten, Winkler, and DeHaan, 1950).The sole pyrimidine which appeared to be re-quired under these conditions was thymine. Themedium of Rutten et al. was modified as follows:a double strength mineral medium (Cohen andArbogast, 1950) containing 4 mg sulfanilamideper ml was autoclaved and, on cooling, was sup-plemented with glucose and the other essentialmetabolites. The final solution was adjusted tothe normal mineral medium concentration con-taining 2 mg sulfanilamide per ml. The finalconcentrations of the metabolites are given intable 1.

Strain B was grown overnight in mineral me-dium containing 1 mg glucose per ml. The glucosewas exhausted from the medium and viable countwas ca. 109 per ml. The sulfanilamide-metabolitemedium (SM) was inoculated at a 1:50 dilutionand incubated overnight with aeration to a viablecount approaching 109 per ml. On dilution intoSM medium, after a 20- to 40-minute lag, thedivision time, determined by viable count, was2.5 to 3 hours and the mass doubling time, deter-mined turbidimetrically, tended to be slightlylonger. Dilution into the sulfanilamide mediumwithout metabolites other than thymine and glu-cose resulted in a very long lag and a very slowincrease in viable count.Thus the division time in a complete SM me-

dium, although markedly improved by the pres-ence of metabolites, is several times longer thanin the absence of sulfanilamide. It can be con-cluded therefore that the supplement does notfulfill the complete requirement of strain B for

TABLE 1Supplementa to sulfanilamide medium (SM)

Final Con-Compound centration

ug. per ml

DL-Methionine..................... 30DL-Serine.......................... 30DL-Valine.......................... 30L-Hlistidine ........................ 30Hypoxanthine..................... 30Xanthine.......................... 30Ca Pantothenate ........... ....... 1Pyridoxine........................ 1Thiamin........................... 1Thymine.......................... 30Glucose.......................... 1000

essential metabolites containing one-carbonfragments or provides these metabolites in a formwhich is not optimal for their utilization in theformation of polymers.

Omission of thymine from the SM mediumresulted in a rapid decline of viable count. Omis-sion of any one or all of the other metabolites inthe presence of thymine inhibited multiplicationbut did not result in death. A typical experimentis presented below:The overnight culture from SM was centri-

fuged, diluted in SM and permitted to go throughone division. At this point the culture was washedonce with mineral medium containing 2 mg sul-fanilamide per ml and resuspended in SM mediumlacking thymine, hypoxanthine, and xanthine.In figure lb can be seen the increase of turbidityin cultures to which have been added either thy-mine, the purines, or both. The data show a con-siderably greater and more rapid effect in tur-bidity increase as a result of the absence of purinethan that due to omission of thymine.

Figure la presents the viable counts of the samecultures. In the absence of exogenous purine, theviable count increases slightly and rapidly reachesa plateau. In the presence of all metabolites withthe single exception of thymine, the viable countwas constant for about 40 minutes and then felloff rapidly. In all comparable experiments theviable counts fell to about 10 per cent of theoriginal in the time necessary for one division, arate of killing comparable to that observed withthymine-deficient 15T_2. A comparison of figures

2With l&r-, extended incubation under condi-tions of thymineless death permits the survival ofone bacterium per 104 cells. Survival of strain B in

1956] 589

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COHEN AND BARNER

MINUTES

Figure 1. Growth of strain B in complete sulf-anilamide-containing medium (SM) and its re-

sponse upon deletions of purines or thymine.a (top): viable counts. b (bottom): turbidity data.

la and lb demonstrates that the phenomenon ofthymineless death in this organism is also accom-panied by cell growth, as indicated by an increasein turbidity of the culture comparable to thatobserved with cultures of 15r dying in the ab-sence of thymine.

If an amino acid such as methionine, as well as

thymine, was omitted from the SM medium, itwas observed that only about half the cells were

killed in one hour and no death occurred there-after. If methionine alone was omitted from theSM medium, a very slow rate of division was

observed. Thus the imposition of even an incom-

SM lacking thymine was of the order of 100 to 1000times that observed with 1&r-. Analysis of thesurvivors in the latter instance have indicatedthat these bacteria are still thymineless, but growat a much slower rate than the bulk of the originalculture. It appears reasonable to suppose that theproportion of very slow growing B is considerablyincreased under conditions of growth in the SMmedium, as compared to the rate of growth inmedia lacking sulfanilamide. This would limit theextent of killing in SM medium devoid of thymine.

plete amino acid deficiency upon the thyminedeficiency markedly reduced the killing observed.A marked reduction of the rate of killing inthymine deficiency with strain 15T- has also beenobserved by the use of the tryptophan analogue,5-methyl tryptophan.Growth in the presence of excretion product8.

One hypothesis suggests that thymineless deathmay be due to the accumulation of a soluble toxicproduct. To test this hypothesis, strain 15,r- wasgrown in thymine-containing media until thyminewas exhausted at 109 bacteria per ml. When thecelLs were removed from the exhausted medium,it was found that the medium containing prod-ucts excreted during growth did not kill 15Tr butinstead protected 15r from death in the absenceof thymine (Barner and Cohen, 1954). The na-ture of this protective effect is not clear but it isevident that the hypothesis of a toxic productcannot be tested easily in these circumstances.

It has been shown that 15r. excretes a varietyof metabolic products during utilization of glu-cose. These include uracil, as a major ultravioletabsorbing component, and orotic acid and hypo-xanthine as lesser components (Cohen and Barner,1954). When 15r, is incubated for two hours with109 bacteria per ml in a glucose medium lackingthymine, the viable count falls in two hours to1 per cent of the original value. In this intervaluracil is excreted to give about 4 ,ug per ml andorotic acid to about 1 ,ug per ml. The addition ofthese metabolites to a glucose-thymine mediumat 10 times these concentrations does not affectthe division time of 15r- nor do these excretionproducts significantly increase the rate of dyingof these bacteria on incubation in the absence ofthymine.When uracil is added at 100 ,ug per ml to 15r-

in media containing thymine at 2 ,ug per ml, thedivision time was observed to increase from 45to 55 minutes. At 1 Mug of thymine, 100 Mg ofuracil increased the division time by 60 to 75 percent and prevented more than a doubling of thebacteria. Despite this inhibition of multiplicationthe turbidity of such cultures increased about7- to 8-fold in a 3-hour interval. The presence of100 Mug uracil did not affect the rate of dying of15r in the absence of thymine. Uracil thus ap-pears to act as a thymine competitor at a suffi-ciently high uracil-to-thymine ratio.The effects of high concentration of uracil are

superficially similar to those obtained with

590 [VOL. 71

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

INDUCTION OF THYMINE DEFICIENCY

5-bromouracil (BrU) whose properties will bediscussed in greater detail below. At this pointwe may present some data comparing the effectsof uracil and 5-bromouracil on 15r in the presenceof thymine. In figure 2 it can be seen that 100;sg uracil is somewhat less active than 10 pg of5-bromouracil as a thymine analogue and thateach compound appears to potentiate the other'sactivity.

The effect of thymine analogues. Azathymine,5-nitrouracil, and 5-bromouracil (BrU) weretested as thymine analogues with 15r . Azathy-mine and 5-nitrouracil at 20 ug per ml had noeffect on division in the presence of 0.4 ,ug ofthymine or on the rate of dying in the absence ofthymine. 5-Bromouracil is an effective thymineanalogue (Hitchings et al., 1950a and 1950b) andappears to replace thymine in the synthesis ofDNA (Dunn and Smith, 1954; Zamenhof andGriboff, 1954).When 15r was incubated without thymine in

the presence of 3 pg bromouracil per ml, theturbidity of the culture usually increased five- tosixfold before growth ceased. Nucleic acids weredetermined (Schneider, 1945) and both DNA andribose nucleic acid (RNA) were found to increasein proportion to the turbidimetric increase. De-spite this apparent growth, the viable count ofthe culture was observed to double very slowlyand then fell off as in thymineless death, albeitat a somewhat slower rate. A typical experimentis presented in figure 3. In many experimentswith 5-bromouracil, viable count never increasedmore than twofold.In the presence of thymine, 5-bromouracil was

also effective in provoking cell death, as deter-mined by the reduction of viable count. Turbidityincrease was extensive, despite the loss of thepower to multiply and examination of the culturerevealed tremendously long bacteria. The lack ofrelation of turbidity to viable count is clearlydemonstrated in figure 4. In this figure, a weightratio of 5-bromouracil to thymine of 5 to 10:1 isseen to permit a single division, while a largerratio did not permit even this. At a ratio of 10of Br to thymine the bacteria were killed aftera single division. Larger ratios increased the deathrate. In order to completely overcome the in-hibitory action of 5-bromouracil at 10 ,ug per ml,it was necessary to add 200 Mg of thymine per mlof culture of 15T . In contrast to the extraordinarysensitivity of this organism to 5-bromouracil,

c

z0

45;

HOURS

Figure B. The eflect of thymine analogues onstrain 1ST- in the presence of limiting thymine.Medium contained 1 ,ug/ml thymine, 10 ;pg 5-bromouracil, and 100 g uracil added as indicated.

amounts up to 500 pg per ml were inactive inreducing the division time of the parent strain 15,strain B, and strain Wo.

Thymidine analogues in the presence of thymine.Bromouracil is far less effective in killing cells in

8.

*BrU

z

LId80 48W -1

cc~~~~~~~~~~D60~~~~~~~~~

HOURS

Figure S. The response of strain 1ST- to 5-bromo-uracil (BrU) as sole pyrimidine source in syntheticmedium. Control contained no pyrimidine.

59119561

.;j

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COHEN ANDBARNER[

HOURSFigure 4. The effect of 5-bromouracil (BrU) on

strain 15T- in synthetic medium containing 1 ,ugthymine/ml.

N-C-OH N-C-OH N-C-OH N-C-OH N-C-OHo

I I IC j-CHs O-C C-OH O-C C- -C C-CH5 OSC C-CH,L I I tN-&H N-CH N- N-CH N-CH

I

HCHI HCH HCH HCOH HOCHCOI HUH HCOH

&.OH &40H COH CHOH

Thymidine S-OH Urccil 5-Bromouracil Thymine SpongothwIdiudeonyriboside deo.yriboeide nbosie

Figure 5. The structures of various thymidineanalogues tested.

the presence of thymidine. This result was ex-pected since free pyrimidines are rarely used to asignificant extent in bacteria and animals, whichdo however appear to incorporate pyrimidinenucleosides. Although there are some indicationsthat plants can incorporate pyrimidines, evenhere thymidine appears to be used to a greaterextent than is thymine (McQuade et al., 1955).Therefore it would be desirable to have thymidineanalogues since the more ready utilization of thedeoxyriboside suggests that competition with thismetabolite would be potentially more effectivein inhibiting division than is competition withthe free pyrimidine. Until very recently prepara-tive methods were inadequate to supply com-pounds of this type but a number have now be-

come available. Thymidine analogues wereobtained inwhichthepyrimidinewas altered at the5 position without alteration of the deoxyribosylmoiety. These were 5-hydroxyuracil deoxyribo-side and 5-bromouracil deoxyriboside (Beltz andViser, 1955). Two others were obtained in whichthymine was present but in which the sugarmoiety was altered. These were thymine ribo-furanoside and spongothymidine (Bergman andFeeney, 1951). The latter is a natural productisolated from sponges, in which the sugar is re-ported to be D-arabinose. The structures ofthese compounds are given in figure 5.

Strain 15T-wasincubated inmedia containing 7.9X 1O6 mi thymine and 5-bromouracil, 5-hydroxy-uracil deoxyriboside, or 5-bromouracil deoxyri-boside at a molar ratio of inhibitor to thymine of6.6. In figure 6, it can be seen that the hydroxy-uracil deoxyriboside had no effect on the divisiontime of 15r. A tenfold increment of this com-pound similarly had no effect on the growth anddivision of 15r-. 5-Bromouracil at this concentra-tion permitted one division, which was then fol-lowed by a slow decline in viable count. However,the deoxyriboside of 5-bromouracil produced aprecipitous killing of 85 per cent of the celLs,

Li-ico

4;

OR

HOURS

Figure 6. The effect of various analogues onstrain 15T- in medium containing 1 jug thymine/ml.All inhibitors tested at a molar ratio of 6.6 tothymine.

592 [VOL. 71

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

INDUCTION OF THYMINE DEFICIENCY

followed by a bacteriostasis of the rest of thepopulation.When the three compounds were tested at these CONTRO

concentrations on strains 15, WG, and B grownin the absence of exogenous thymine, no effect BRMOUR,CILwas obtained on the growth or division of theseorganisms. SPONGOTHYMIDINE

Strain 15T- was incubated in media lacking freethymine but containing thymine ribofuranoside dsor spongothymidine. It was observed that the F.former was slowly able to yield thymine for Dgrowth. However, death occurred with spongo- 8thymidine, albeit less rapidly than in the absenceof the compound. These results are presented < BROMOURACILin figure 7. > DEOXYRIBOSIDE

In the presence of thymine in the medium,thymine ribofuranoside nevertheless inhibitedgrowth, apparently until such a time as it wasdecomposed. Spongothymidine, however, wasonly slightly effective in the presence of exogenousthymine. 107

Thymidine anlogues in presence of thymidine. I 2 3Strain 15T- was grown in thymine, centrifuged, HOURS

resuspended and incubated in 7.9 X 10-1 M Figure 8. The effect of various analogues onthymidine and various concentrations of the strain 1ST- in medium containing 1.9 pg

thymidine/ml as pyrimidine source. Molar ratio9.0 of inhibitors to thymidine 26.4.

WITH THYMINE- NO THYMINE analogues effective in the presence of thymine.

thymine __ £ Thymine ribofuranoside was essentially ineffec-thymnine +~,....*@ __ ..___ tive at a molar ratio of 33. Spongothymidine-pnohye'- 5--inhibited division significantly in three hours.

8.0 ,__. t mine Molar ratios of inhibitor to thymidine of 6.6, 33,irLSLde and 66 permitted 63, 42, and 36 per cent, re-thymine 4thymine riboside spectively, of the control viable count. At com-

z parable molar concentrations, the inhibition af-8 \ spongothymidine forded by spongothymidine is only slightlygreaterw \ than that produced by 5-bromouracil in theco 70 presence of thymidine.

In contrast to the relative lack of effectivenessn fh mine

of 5-bromouracil, the bromouracil deoxyriboside is3 n thymine fairly active in inducing death of 15r in the pres-

ence of thymidine. These results are presentedin figure 8.

6.0DISCUSSION

It had been shown that a multiple deficiency@|3including that for thymine can be imposed on

HOURS non-thymine-requiring strains of bacteria, e.g.,Figure 7. The response of strain iT- to thymi- E. coli strain B, by growth in the presence of

dine analogues in media with or without thymine. sulfanilamide (Rutten et al., 1950). The experi-Molar ratio of inhibitors to thymine (when pres- ments presented above demonstrate by meanent) 6.6. of this technique that thymineless death can be

59319561

- -1 ----

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COHEN AND BARNER

induced in strain B as well as in 15T-. Also wehave shown that a single deficiency with respectto compounds other than thymine, e.g., purineor methionine, induces bacteriostasis even as ob-served in most auxotrophic mutants of E. coli.Furthermore, multiple deficiencies which includea thymine requirement tend to protect the organ-ism against thymineless death. These results areconsistent with the theory of "death by unbal-anced growth," since a deficiency other than thatfor thymine will usually prevent both nuclearand cytoplasmic growth. For example, purine isorganized into both RNA and DNA, and its lackproduces an inhibition of cell synthesis in general.A similar effect is noted under methionine defi-ciency and in the absence of methionine andthymine, the cell is unable to synthesize cyto-plasm in excess of its normal nuclear complement.

It has been customary to distinguish the actionof sulfanilamide from that of the antibiotics, theformer being classed as "bacteriostatic" in con-trast to the "bactericidal" penicillin or strepto-mycin. The effect of sulfanilamide is clearly toprevent both cytoplasmic and nuclear synthesesas a result of the multiple deficiencies produced.These experiments illustrate that the bacterio-static sulfanilamide can become bactericidal ifthe deficiency for metabolites induced by thisdrug is made specific for a nuclear constituent,thymine.

Contrariwise, bactericidal antibiotics, e.g.,penicillin and streptomycin, do not kill if the or-ganism exposed to these drugs fails to grow forlack of essential metabolites (Hobby et al., 1942;Rosenblum and Bryson, 1953). With several ofthe antibiotics, the cells become filamentous,even as under conditions of thymineless death,and it appears possible that the antibiotics act byinducing unbalanced growth; i.e., they inhibitreactions essential to the biosynthesis of specificcritical cell structures. With both penicillin andstreptomycin, available evidence suggests thepossibility of effects on nucleic acid biosynthesisand structure (Park, 1951; Cohen, 1947), anddoes not yet exclude these modes of action. Thedifferentiation of the various inhibitors of bac-terial growth as "bacteriostatic" or "bactericidal"tends to break down under these circumstances.

In recent years a number of chemicals havebeen found to be anti-tumor agents. The actionof the anti-tumor nitrogen mustards when ap-

plied to bacteria has been shown to have an effectsimilar to that of a thymine deficiency, as dis-cussed in an earlier paper (Barner and Cohen,1955, Loveless et al., 1954). The anti-neoplasticantibiotic, azaserine, induces filament formationin E. coli, i.e., cell growth without division, evenas do the nitrogen mustards, or various types ofradiation (Maxwell and Nickel, 1954).Marked antileukemia activity has been ob-

tained with various antifolic acid derivatives,such as aminopterin and A-methopterin. Toxicamounts of the former have been observed toinhibit RNA synthesis somewhat less than DNAsynthesis (Goldthwait and Bendich, 1951). How-ever A-methopterin has been observed to inhibitthymine synthesis in the spleens of the leukemicmice far more than even the synthesis of DNApurine, with very little effect on RNA synthesis(Balis and Dancis, 1955). This result has led tothe suggestion that different folic acid-enzymecomplexes are involved in thymine and purinesynthesis. Thus the use of this compound maylead to a far more specific deficiency for thyminethan does sulfanilamide, which prevents the syn-thesis of all folic acid coenzymes.

Since the lethal action of sulfanilamide re-quires a complete supply of metabolites otherthan thymine, it may be asked if the anti-tumoreffect of the antifolic acid agents can not be im-proved by a supply of essential metabolites otherthan thymine. If these agents do have effects onsystems other than that necessary for thyminesynthesis, it may be anticipated that the mostrapid rate of thymineless death or minimal sur-vival will not be attained until such metabolitesare provided. These considerations may bear onthe survival of apparently resistant cells, a resultwhich conceivably is attributable to slow growthrather than an insensitivity to the antifolic acidagent.

Antifolic acid compounds produce markedeffects on mitosis (Himes and Leuchtenberger,1950; Hughes, 1950). In a recent study of theantifolic acid compounds, the 1 ,2-dihydro-s-triazines, in plant material (Rudenberg et al.,1955) it was observed that these compounds, likeaminopterin, produced an accumulation of pro-phase cells. On incubation in the absence of theagent the cells were often found to be irreversiblydamaged with aberrant mitoses characterized by"sticky" anaphase bridges. Of particular interest

594 [VOL. 71

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

INDUCTION OF THYMINE DEFICIENCY

in this connection is the ability of thymidine toreverse the inhibition produced by the triazine.Of comparable interest in this connection is thereport that thymine prevents chromosomalbreaks in urethane-treated animals and therebyreduces the toxic effects of this anti-tumor agent(Boyland and Koller, 1954).

Mitotic abnormalities as consequences of apathological thymine metabolism have been par-ticularly observed in plant material. The thymineanalogue, 5-aminouracil, has been used to inducesuch effects in onion root tips, in a manner com-parable to an antifolic acid agent (Duncan andWoods, 1953). The cells of treated roots accumu-late in interphase. Cells which had started mitosisprior to the start of treatment continue throughdivision and then stop in interphase. Thus thy-mine utilization would seem to be blocked by theanalogue at interphase, when it is apparentlyrequired for the doubling of DNA, which has beenshown to occur at this stage of development inplant and animal cells (Vendrely, 1955) and inbacteria (Barner and Cohen, 1955). The effect of5-aminouracil on plant cell mitosis is completelyreversed by thymine. Although the antithymineeffect of uracil described in this paper does notappear to have been previously reported for bac-teria, the incubation of onion root tip in uradilsolutions has been observed to result in chromo-some breaks in mitosis (Deysson, 1952). Morerecently the incorporation of highly radioactivethymidine or thymine in the nuclei of plantmaterial has also been observed to producechromosome breaks (McQuade et al., 1955).In the case of E. coli, strain 15T- in the absence

of exogenous thymine, a new type of DNA ismade in small amounts. The new DNA lacksthymine but contains a new base, 6-methyla-minopurine (Dunn and Smith, 1955). The methylgroup which is normally combined to uracil or toa derivative then appears to be transferred toadenine. The thymine analogues, 5-amino uracilor 2-thiothymine, are not themselves incorporatedinto DNA but are stated to stimulate the syn-thesis of the DNA contaig methylaminopurine. It is not known whether a similar effectis observed in plant material.

It is not known if a new type of DNA is madein thymine deficiency induced by sulfanilamideor an antifolic acid agent. The results withA-methopterin (Balis and Dancis, 1955) suggest

the synthesis of a DNA lacking thymine. How-ever, it does not appear likely that a methylpurine would be contained in such a DNA underconditions of a folio acid deficiency, since thegeneration and transfer of methyl groups appearto involve folic acid enzymes. Whether a deficientDNA is made at all in sulfanilamide-treated strainB which dies by unbalanced growth is of consider-able interest with respect to the precise mecha-nism of thymineless death.

In contrast to 5-aminouracil, bromouracil hasbeen found to be incorporated into DNA in strain15r-. Since 5-bromouracil is found in such DNA asthe deoxyriboside, it appears reasonable to sup-pose that exogenous bromouracil deoxyribosideis merely incorporated more readily than is free5-bromouracil. However, unlike the methylamino-purine-containing DNA which is only synthesizedto the extent of about 20 per cent of the DNAoriginallypresent, 55-bromouracil-containing DNAmay be made to the extent of 500 per cent of theDNA originally present.Thus four types of effects on DNA synthesis

appear possible as a consequence of a specificthymine deficiency or antagonism. There maybe, (1) a complete inhibition of DNA synthesis,(2) a synthesis of thymine-deficient DNA, inwhich the missing thymine deoxynucleotide is notreplaced by another nucleotide, (3) a synthesisof a thymine-deficient DNA, in which a methyl-aminopurine replaces thymine, or (4) a synthesisof a thymine-deficient DNA, in which a thymineanalogue replaces thymine. One may wonderwhich category is produced by high concentra-tions of uracil or whether spongothymidine pro-duces yet another type of DNA, in which a thy-mine pentoside replaces thymidine.

Finally we must comment on the differentsensitivities of various organisms to thymine orthymidine analogues. It is evident that the differ-ential sensitivities of various organisms to thesecompounds will determine the chemotherapeuticsignificance of our observations. It is recorded inmost texts of biochemistry that most cells willnot utilize significant amounts of free pyrimidinesbut will incorporate pyrimidine nucleosides. Un-fortunately studies of this type have been per-formed almost entirely on animal cells. The datamentioned in this discussion indicate that freepyrimidines, including thymine, can be used bysome plant cells as well as by 15r , and that this

5951956]

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

COHEN AND BARNER

apparently is a distinguishing metabolic featureof these plant cells in contrast to animals.3 Thusa thymine analogue is conceivably of chemothera-peutic interest in plant and fungal systems, as wellas in the few types of bacteria which possessrequirements for free pyrimidine. It will be ofconsiderable interest to explore the extent towhich thymidine analogues will kill other typesof cells and, as noted above, the incorporation ofthymidine in animal cells may prove to supplybut a minor part of the thymine necessary forDNA synthesis. Thus a route is known for thesynthesis of pyrimidine nucleosides, which in-volves the condensation of orotic acid and 1-pyro-phosphoryl ribose-5-phosphate to form orotidylicacid. It is conceivable that in E. coli strain B orstrain Wc-, the synthesis of thymidylic acid forDNA by-passes thymidine by a similar type ofmechanism. The existence of multiple alternativemetabolic pathways in the pyrimidine series isbarely beginning to be explored. The differencesin these pathways which are beginning to be seenfrom organism to organism may be expected todetermine the range of applicability of any chemo-therapeutic agent designed to operate in this areaof metabolism.

SUMMARY

Escherichia coli strain B grown in the presenceof sulfanilamide has a thymine deficiency, amongother requirements. If the requirements otherthan thymine are met by supplementing the me-dium, the bacteria die at the rate of approxi-mately 90 per cent in the time necessary for onedivision. If an amino acid or purine requirementis not met, then the bacteria do not grow and arenot killed. Prevention of growth by imposing anamino acid requirement on the thymine require-ment markedly reduces death in the absence ofthymine. The significance of these results is dis-cussed in relation to the use of antifolic acid

8 One may speculate that the ability to in-corporate thymine may be associated with a routethrough thymidine by reaction with deoxyribose-1-phosphate. This sugar phosphate is formed bythe condensation of triose phosphate and acetalde-hyde, and the latter compound is most readilygenerated by carboxylase, an enzyme found inplants and not in animals. E. coli strain B whichdoes not incorporate thymine into DNA does notmake deoxyribose during growth by a condensa-tion of triose phosphate and acetaldehyde(Lanning and Cohen, 1954) and lacks carboxylase.

agents in tumor chemotherapy, in the inhibitionof plant growth, and in the induction of chromo-some breaks in mitosis.A number of pyrimidines and their analogues

were tested for their ability to compete withthymine in the multiplication of 15,r. It wasobserved that uracil at high concentrations actsas a thymine analogue. 5-Bromouracil appearsto compete with and to replace thymine in sup-porting growth and nucleic acid synthesis in15r. However, cells incorporating this pyrimi-dine in their DNA were capable of only onedivision and then were killed irreversibly as in-corporation of the analogue continued. Thecompound was relatively ineffective in the pres-ence of thymidine.

In tests on 15T- thymidine analogues, thymi-dine ribofuranoside, spongothymidine, and 5-bro-mouracil deoxyriboside, were inhibitors ofdivision in the presence or absence of thymidine.The compounds were inactive with E. coli strainB and strain Wc. Under conditions of normalgrowth, the riboside was slowly degraded by15r-. Spongothymidine was much less activethan the bromouracil deoxyriboside. The sig-nificance of these results has been discussed.

REFERENCESBALIS, M. E. AND DANCIS, J. 1955 Effects of

A-methopterin on nucleic acid synthesis inleukemic spleen breis. Cancer Research, 15,603-608.

BARNER, H. D. AND COHEN, S. S. 1954 The in-duction of thymine synthesis by T2 infectionof a thymine requiring mutant of Escherichiacoli. J. Bacteriol., 68, 80-88.

BARNER, H. D. AND COHEN, S. S. 1955 Syn-chronization of division of a thyminelessmutant of Escherichia coli. Federation Proc.,14, 177.

BARNER, H. D. AND COHEN, S. S. 1956 Therelation of growth to the lethal damage in-duced by ultraviolet irradiation in Escherichiacoli. J. Bacteriol., 71,149-157.

BELTZ, R. E. AND VISSER, D. W. 1955 Growthinhibition of Escherichia coli by newthymidine analogs. J. Am. Chem. Soc.,77, 736-738.

BERGMANN, W. AND FEENEY, R. J. 1951 Marineproducts XXXII. The nucleosides of sponges.I. J. Org. Chem., 16, 981-987.

BOYLAND, E. AND KOLLER, P. C. 1954 Effectsof urethane on mitosis in the Walker ratcarcinoma. Brit. J. Cancer, 8, 677-684.

COHEN, S. S. 1947 Streptomycin and desoxy-

596 [VOL. 71

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

INDUCTION OF THYMINE DEFICIENCY

ribonuclease in the study of variations in theproperties of a bacterial virus. J. Biol.Chem., 168, 511-526.

COHEN, S. S. AND ARBOGAST, R. 1950 Chemicalstudies in host virus interactions. VII. Acomparison of some properties of three mutantpairs of bacterial viruses, T2r+ and T2r,T4r+ and T4r, T6r+ and T6r. J. Exptl.Med., 91, 619-636.

COHEN, S. S. AND BARNER, H. D. 1954 Studieson unbalanced growth in Escherichia coli.Proc. Natl. Acad. Sci. U. S., 40, 885-893.

COHEN, S. S. AND BARNER, H. D. 1955Enzymatic adaptation in a thymine-requiringstrain of Escherichia coli. J. Bacteriol.,69, 59-66.

DEYSSON, M. 1952 Fragmentation d'un deschromosomes de l'Allium Cepa L. sous l'in-fluence de l'uracile. Compt. rend., 234,650-652.

DUNCAN, R. E. AND WOODS, P. S. 1953 Somecytological aspects of antagonism in synthesisof nucleic acid. Chromosoma, 6, 45-60.

DUNN, D. B. AND SMITH, J. D. 1954 Incorpora-tion of halogenated pyrimidines into thedeoxyribonucleic acids of Bacterium coli andits bacteriophages. Nature, 174, 305.

DUNN, D. B. AND SMITH, J. D. 1955 The oc-currence of a new base in the deoxyribonucleicacid of a strain of Bacterium coli. Nature,175, 336-337.

GOLDTHWAIT, D. A. AND BENDICH, A. 1951 Ef-fect of aminopterin on nucleic acid metabolismin the rat. Federation Proc., 10, 190.

HIMEs, M. AND LEUCHTENBERGER, C. 1950 Theeffects of some folic acid analogues on mitosis.Genetics, 35, 114.

HITCHINGS, G. H., ELION, G. B., FALCO, E. A.,RUSSELL, P. B., SHERWOOD, M. B., AND VAN-DERWERFF, H. 1950a Antagonists of nucleicacid derivatives. I. The Lactobacillus caseimodel. J. Biol. Chem., 183, 1-9.

HITCHINGS, G. H., ELION, G. B., AND FALCO, E. A.1950b Antagonists of nucleic acid derivatives.II. Reversal studies with substances struc-turally related to thymine. J. Biol. Chem.,185, 643-649.

HOBBY, G. L., MEYER, K., AND CHAFFEE, E. 1942Observations on the mechanism of action ofpenicillin. Proc. Soc. Exper. Biol. Med., 50,281-285.

HUGHES, A. 1950 The effect of inhibitory sub-stances on cell division. Quart. J. Microscop.Sci., 91, 251-277.

LANNING, M. C. AND COHEN, S. S. 1954 Themechanism of ribose formation in Escherichiacoli. J. Biol. Chem., 207, 193-199.

LOVELESS, T. E., SPOERL, E., AND WEISMAN, T. H.1954 A survey of effects of chemicals on divi-sion and growth of yeast and Escherichia coli.J. Bacteriol., 68, 637-644.

MAXWELL, R. E. AND NICKEL, V. S. 1954 Fila-ment formation in E. coli induced by azaserineand other antineoplastic agents. Science,120, 270-271.

MCQUADE, H. A., FRIEDKIN, M., AND ATcHISON,A. A. 1955 Chromosome aberrations causedby radioactive thymidine. Nature, 175,1038.

PARK, J. T. 1951 The uridine-5-pyrophosphatecompounds found in penicillin-treatedStaphylococcus aureus cells. Phosphorusmetabolism, Vol. 1, pp. 93-98. Johns HopkinsPress, Baltimore, Md.

ROSENBLUM, E. D. AND BRYSON, V. 1953 Rateof bacterial metabolism and the action ofstreptomycin. Antibiotics and Chemo-therapy, 3, 957-965.

RUDENBERG, L., FOLEY, G. E., AND WINTER, W. D.,JR. 1955 Chemical and biological studieson 1,2-dihydro-s-triazines. XI: Inhibition ofroot growth and its reversal by citrovorumfactor. Science, 121, 899-900.

RUTTEN, F. J., WINKLER, K. C., AND DEHAAN,P. G. 1950 The action of sulfanilamideon bacteriophages T1-T7. Brit. J. Exptl.Pathol., 31, 369-379.

SCHNEIDER, W. C. 1945 Phosphorus compoundsin animal tissues. I. Extraction and estima-tion of desoxypentose nucleic acid and ofpentose nucleic acid. J. Biol. Chem., 161,293-303.

VENDRELY, R. 1955 The deoxyribonucleic acidcontent of the nucleus. The nucleic acids.Vol. II, pp. 155-180. Academic Press, NewYork, N. Y.

ZAMENHOF, S. AND GRIBOFF, G. 1954 Incorpora-tion of halogenated pyrimidines into thedeoxyribonucleic acids of Bacterium coli.Nature, 174, 306.

19561 597

on May 23, 2018 by guest

http://jb.asm.org/

Dow

nloaded from