7 food mutagens - lawrence livermore national … · genetic toxicity of food mutagens like phip...

7

Science & Technology Review September 1995

Food Mutagens

they are chemically changed bymetabolizing enzymes present in animaltissues, such as the liver. When the bodyencounters foreign substances with anaffinity for fats (known as lipophilicxenobiotics), it tries to make them moresoluble in water so they can be excreted.Most of the intermediate compoundsthat are formed in this process are

investigators performing research on foodmutagens have at their disposal severaladvanced techniques and tools withexquisite sensitivity, such as acceleratormass spectrometry, that are providinganswers to daunting questions.

Facts and Basic Questions

Biologists use the term “mutagenicactivity” to describe the potency of amutagen known to cause structuraldamage to the molecular units that makeup the genes. The mutagenic activity ofthe heterocyclic amines found in cookedfood is strongly affected by severalfeatures of their molecular structure.Even small structural changes can havelarge effects on mutagenic activity.

The imidazole ring is a commonfeature in all of the heterocyclic amines(Figure 1). We know that mutagenicactivity is increased when a methyl group(CH3) is present on the imidazole ring.Both the position and number of methylgroups have an effect. Thus, as shown inthe illustration, the mutagenic activity

of one highly potent food mutagen, IQ,increases with the addition of a methylgroup at the number-4 position of themolecule (to make 4-MeIQ). Conversely,mutagenic activity can be decreased bythe addition of a methyl group to otherpositions, such as the number-5 position.The numbers and positions of doublebonds and aromatic rings also have alarge effect. However, the variations inmutagenic activity associated withchanges in chemical structure are notalways consistent in tests using differenttypes of cells, tissues, or animals. Thatis, one species may be more susceptibleto colon tumors for one mutagen,whereas a different species may bemore susceptible to liver tumors forthe same mutagen or a different one.

How does the potency of foodmutagens generally compare with thepotency of other biological toxins asmeasured by standard tests usingbacteria? Benzo[a]pyrene is a widelystudied carcinogen and commonenvironmental pollutant that has beenisolated in cigarette smoke, diesel

6


UR diets expose us to manysubstances that can be beneficial in

maintaining health or harmful by causingdisease. Of all the substances known tobe produced during cooking, we nowthink that the most genetically toxiccompounds are the heterocyclic amines.

The role of these potent foodmutagens in the human diet has been thesubject of ongoing research at LLNL for17 years. A previous report in the July1995 issue of Science and TechnologyReview provided an overview of the wayswe identify and quantify mutagens incooked food. Although isolating thetoxic compounds and determining theiramounts in various protein-containingfoods are major efforts in themselves,they tell only part of our research. The other part concerns our efforts tounderstand how food mutagens can lead

to genetic damage and, ultimately, tocancer—at least in laboratory animalsthat have received very high doses ofmutagens.

Research on the genetic damage thatcan be caused by food mutagens beginswith a paradox. Why does the humanbody make a cancer-causing substanceout of certain trace compounds in foodthat, on ingestion, are virtually inertbiologically?

Researchers have now identified morethan a dozen heterocyclic amines incooked foods commonly found in theWestern diet. Five of these heterocyclicamines were first identified at LLNL.All are lipophilic (they have a strongaffinity for fats). However, they are not,in themselves, either mutagenic orcarcinogenic when eaten. Rather, thecompounds become harmful only after

Figure 1. (a) The imidazolering is a common featurein all of the heterocyclicamines. (b) IQ is a highlypotent food mutagen thathas been widely studied.(c) When a methyl group(CH3) is added to thenumber-4 position of themolecule to make 4-MeIQ,the mutagenic activity isincreased.

further metabolized and harmlesslyeliminated, primarily in the urine.However, some intermediate compounds,including those derived from cookedfood, are highly reactive, binding toDNA (deoxyribonucleic acid) andpotentially resulting in genetic damage.

The sequence of events leadingfrom eating mutagenic precursors(promutagens) to DNA interactions andcancer is highly complex. The principalunknowns in the disease process arethe reactions within cells and amongmolecules, and it is these events thatdrive our research efforts on foodmutagens and the induction of cancer.

Major difficulties arise in our attemptsto calculate the dose of food mutagensin the human diet and, therefore, to makerealistic assessments of cancer risk toan individual. We are concerned withtrace levels of exposure at the part-per-billion or even part-per-trillion level.The content of mutagen precursors evenin one type of food, such as a hamburger,can vary widely depending on the detailsof cooking. In addition, some foodmutagens are a thousand times morepotent than others. Moreover, humandietary habits differ appreciably. Aswith most environmental carcinogens,the cancer risks posed to humans are afunction of many variables, andestimating those risks entails makingassumptions. Fortunately, LLNL

Potent mutagens, called heterocyclic amines, areproduced when foods derived from muscle andother protein sources are cooked. We have studiedthe metabolic pathways of these compounds andtheir interactions with DNA. This report, thesecond of two, focuses on the mechanisms bywhich food mutagens may lead to cancer and onthe potential risks associated with theirconsumption.

CH3

NH2

CH3N

CH3

NH2

N

C

H

H

H

HC

N

N

C

CH3

(a) The imidazole ring

(b) The food mutagen IQ (c) The food mutagen 4-MeIQ

Ring in shorthand form plus methyl group (CH3)

NN

N N

N

N

98

76 5

4

3

21

OO

Mutagenic Activity, DNA Mechanisms, and Cancer RiskFood Mutagens:

Mutagenic Activity, DNA Mechanisms, and Cancer Risk

9


Food Mutagens

still visible in six types of tissue,notably the colon, breast, and pancreas.

Whereas this type of research answerssome questions, it raises others. Howmuch confidence can we place inextrapolations made from rats (andother animals) to humans? After all,mammals differ from one another inmany ways, including the expressionof various enzymes. Thus, even aspecies related more closely to humansthan the rat may pose problems whenwe study animals to make humanassessments. To address, in part, thesedifferences and to evaluate the relevanceof such information in human disease,we use many different kinds of modelsystems to estimate risk, including wholeanimals, human and animal tissuefractions, and bacterial assays, coupledwith state-of-the-art research techniques.We have made remarkable discoveriesat the molecular level on specificmechanisms by which food mutagenscan lead to adverse health consequences.

Steps Leading to Cancer

Figure 2 shows some ofthe main steps by whichexposure to heterocyclic

amines may lead to certaintypes of cancer in humans. Of necessity,this scheme is highly simplified. Inreality, many different kinds of chemicalreactions, enzymes, intermediatemetabolites, inhibitors, tissues, andgenes—including tumor-suppressorand DNA-repair genes—play a role inwhether or not humans will developcancer after being exposed to dietarycarcinogens.

Nevertheless, for convenience, wecan break down the complex processinto the following steps:1. Ingestion. Humans eat foods, such asfried meat, containing promutagens thatcan become highly mutagenic whenacted upon by enzymes.

2. Bioactivation. The body attempts toexcrete the ingested toxins. Naturallyoccurring intracellular enzymes catalyzethe formation of intermediate metabolites,which have the potential to react stronglywith DNA.3. Adduction. Certain intermediatefood-mutagen molecules bindcovalently to specific atoms in theDNA macromolecule and form bulkylesions called adducts.4. Mutation. Structural changes in themolecular units that make up the genescan cause DNA replication errors,preventing the gene from functioningproperly in daughter cells. DNA repairmechanisms may determine whether thestructural changes are fixed or not.5. Proliferation. In some cases, themutations occur in genes controllingcell proliferation and replication,leading to tumors. Oncogenes or tumor-suppressor genes are specific examplesof such genes.

Steps 1 and 5 are generally related toour research efforts in dose and riskassessment, respectively. But beforeone can understand how we assess risk,one must understand how food mutagensare biologically transformed into highlyreactive intermediate molecules that arecapable of linking up with anddamaging the genetic material.

Bioactivation Is Key

Once the promutagens in cookedfood are ingested, even in doses thatare one-millionth those used in manyanimal tests, studies have shown thatthey survive the acid in the stomach.After they pass through the stomachand enter the intestine, the compoundsare taken up by the bloodstream andare metabolized by the liver. Locatedwithin the cells of the liver and otherorgans is a family of enzymes calledcytochrome P450s essential for manyfunctions, including the metabolism of

hormones and defense against harmfulenvironmental chemicals.

In the liver, the heterocyclic aminesinteract with cytochrome P450 enzymesthat are involved in defensive reactions.Biologists use the terms metabolicactivation, or bioactivation, to describethe kinds of chemical changes thattake place when cytochrome P450sact on foreign chemicals to convertthem into chemically more polar and,consequently, biologically morereactive forms.

Researchers at LLNL and elsewherehave shown that the mutagenic activityof PhIP, for example, clearly dependson the reactive intermediates that formafter it is acted upon by cytochromeP450 enzymes. As shown in Figure 3,one of the cytochrome P450 enzymes(in particular, P450IA2) converts PhIP into an intermediate moleculecontaining the hydroxy (OH–) group.This polar intermediate, N-hydroxyPhIP, can bind to DNA, but it does so with low affinity.

We have shown that N-hydroxy PhIPis transformed into still more biologicallyactive intermediates, which appear tobe necessary for the stronger bindingwith DNA in the living body. Theintermediate molecules include acetates(the acetate, N-acetoxy PhIP, is shownin Figure 3), sulfates, and other forms.

We have investigated manymetabolic pathways that lead to thegenetic toxicity of food mutagens likePhIP and MeIQx. In our investigations,we have used cells from animals andhumans, enzyme extracts, bacterial cellcultures, and radioactive labeledisotopes. We have studied whethercertain enzyme inhibitors could, ineffect, block the binding of suspectedintermediates to the DNA molecule,and whether other agents could increasethe levels of P450 enzymes and the rateof DNA binding. This type of research

8


Food Mutagens

exhaust, and the smoke from burningfat. Compared to benzo[a]pyrene,PhIP—a food mutagen we have studiedin considerable detail—is 10 times moremutagenic. The mutagen IQ is about100 times more potent than PhIP, and4-MeIQ is three times more potentthan IQ.

All the known heterocyclic aminesare very mutagenic in bacterial tests.Indeed, their mutagenic activity isestablished from these tests in the firstplace. The published numbers onmutagenic activity come from studyingspecific strains of the bacteriumSalmonella typhimurium used in theAmes mutation assay. (See the July 1995issue of Science and Technology Reviewfor a detailed description of this test.)Beyond the bacterial tests, at least 11heterocyclic amines have been shownto be carcinogenic in rodents, and atleast one, IQ, is a potent inducer of livertumors (carcinomas) in monkeys.1*

It is important to remember that thestudies establishing mutagenic activityare done in bacteria, and other studiesprompting further questions about diet-related carcinogenic compounds are donein animals, usually rodents. The studieson rodents involve very high doses ofmutagens, in part because such researchcan be quite costly, and most studies arelimited to about 30 to 50 animals per

experiment. Therationale of muchresearch on mutagensassumes that rodents areessentially the same ashumans and thatexposures to high dosesare proportional toexposures to low doses.These two assumptionshave not always beenclearly validated in thepast. The problem has been that, untilrecently, the effects from exposure tovery low doses of mutagens wereimpossible to test empirically becauseour measuring instruments were notsensitive enough to detect them.

One of our recent studies is a goodexample of how we are addressing theproblem of low-dose exposure. Thiswork makes use of instruments thatwere not previously available inbiomedical research. It was inspiredby other researchers who raised thepossibility that mothers eating well-done meat could pass on heterocyclicamines to their babies through breastmilk. Concern about this route fortransmitting mutagens is based on anexperiment involving nursing pups whenthe maternal rats are given 10 mg of themutagen PhIP per kilogram of bodyweight. Humans eating typical amounts

of well-done meats consume 10,000 to100,000 times less of the mutagenicmaterial daily per kilogram of bodyweight than do the rats in suchexperiments. What is the plausibility,then, of extrapolating from high-doseexperiments to the low dosesexperienced in actual human exposures?

If we are to make realistic estimatesof risk, we need to understand the specificeffects of chemicals at the relatively lowlevels that are characteristic of humanexposures. To do so, we conducted astudy in which we gave PhIP to rodentsat doses spanning many orders ofmagnitude. We found that even atextremely low doses—down to thelevel of mutagens found in a singlehamburger—the effects of PhIP were

Tumor

Bioactivation

Ingestion

PhIP

N-hydroxy PhIP

N-acetoxy PhIPAdduction

Mutagenadduction

DNA

AT

CG

Normal

AT

Mutation

C

Proliferation

Cell growth

Mutation

Figure 2. Some of the principal steps by which exposureto heterocyclic amines in the diet may lead to certaintypes of cancer in humans.

*All references are on p. 23.

11


possibly more) of the bases of the DNAsequence. The heterocyclic amines fromcooked food are relatively bulkymolecules. When reactive intermediates(such as N-hydroxy PhIP) chemicallybond to DNA, they distort the normalDNA helix. The adduction can causeerrors (mutations) to occur when the DNAreplicates, or it may even block the abilityto replicate at all. In either case, the normalfunction of the affected stretch of DNAwill be impaired.

One of the problems for researchers isthat it is difficult to detect adducts atthe low exposure levels that are relevantto humans. To address this key problem,researchers at Livermore have assessedDNA damage at very low doses using twotechniques: traditional 32P-postlabelingand accelerator mass spectrometry (AMS).

Detecting DNA Adducts

The technique of 32P-postlabelinginvolves tagging a chemical adduct ofinterest with the a radioactive isotope ofphosphorus, 32P. Researchers use 32Pbecause it is relatively easy to detect andhas low natural abundance in biologicalmaterial. As an assay for detecting DNAadducts, 32P-postlabeling does not requireprior knowledge about the type of exposureor the adduct structure, and it does notrequire prior radioactive labeling of thechemical of interest (thus, the namepostlabeling). First, DNA is broken downinto its component units (the nucleotidebases). Next, 32P is added to the bases.The specific radioactively labeled adductsare separated from the nonadducted,nonlabeled DNA using inexpensive,thin-layer chromatography.

In practice, the method of postlabelingis sensitive enough to qualitatively detect

10


Food Mutagens

PhIP

Carbon

Nitrogen

Hydrogen

Oxygen

N-hydroxy PhIP

N-acetoxy PhIP

Figure 3. How PhIP becomes mutagenic. Cytochrome P450 convertsthe mutagen PhIP into a more biologically reactive intermediatemolecule containing the hydroxy (OH–) group. N-hydroxy PhIP istransformed into an acetate, called N-acetoxy PhIP, which is highlyreactive with DNA.

NA is the spiral, double-strandedmacromolecule that contains the

genetic blueprint. As shown in Figure 4,DNA encodes the genetic informationin the sequence of four differentnucleotide bases: adenine (A), thymine(T), cytosine (C), and guanine (G). InDNA, the nucleotide base A on onestrand of DNA always pairs with T onthe other strand, and C always pairswith G. A specific string or sequenceof the base pairs, which can typicallyrange from about one thousand to twomillion pairs long, makes up a gene.

Adduction is the covalent bindingof chemicals with large molecules, suchas DNA or protein. Most often, adductionoccurs after the chemicals are metabolizedinto reactive intermediates through theprocess of bioactivation. Adduction isof great interest to researchers for severalreasons. It can serve as an integratedindicator of exposure to carcinogens andthe bioactivation of promutagens, thusrevealing individual susceptibility tocancer. Furthermore, DNA adductioncould be an important indicator(biologists use the term “marker”) aswe look for ways to intervene and reduceindividual cancer risk.

Figure 4 shows an artist’sinterpretation of DNA adduction. Theadduct is the large molecule—in our case,a mutagen derived from cooked food—that can chemically bond to one (or

DD

helps us to better understand themechanisms of toxicity and howcompounds like PhIP and MeIQx can pose human health risks.

Our research suggests that the rulesthat apply to the metabolism of onefood mutagen may not apply toanother, tissue differences areimportant, and so are speciesdifferences. However, we now believethat the primary bioactivation of foodmutagens reacting with P450 enzymestakes place mostly in the liver in bothrodents and monkeys. Such activationoccurs after the administration ofeither high experimental doses or verylow doses of the sort typically foundin human diets. Furthermore,bioactivation probably occurs in othertissues that are the targets of tumors,such as the breast and colon.Following the transport of the firstintermediate (such as N-hydroxyPhIP) in the blood from the liver,bioactivation in these target tissuesprovides the acetates and sulfates thatare close to the unknown, veryreactive molecular species (possiblythe nitrinium ion) that binds stronglywith DNA.

DNA MechanismsFood Mutagens:

DNA Mechanisms

13


Food Mutagens

present in cooked meat products typicalof the American diet. Animals were feddaily an amount of radioactively labeledMeIQx equivalent to what a human wouldreceive eating about two hamburgers aday (200 g).

We found that concentrations ofMeIQx stabilize in the tissues in about7 days. We could detect DNA adducts24 hours after ingestion, but it took about40 days for the number of adducts toreach maximum concentration in theliver and kidney.

In related studies, we gave rodents14C-labeled MeIQx for 7 days. Thedoses ranged from levels below those thatmay be typical of human exposure to veryhigh levels used in cancer assays. We then

12


Food Mutagens

roughly one adduct in about 10 cells(about one adduct per 10 billionnucleotides). However, it cansemiquantitatively measure one adduct inone million to one billion nucleotides.The assay’s principal use is in analyzinga group of structurally different adducts.Postlabeling is highly useful in studyingpotential human exposures to mutagensas well as carcinogens and themechanisms and levels of DNA binding.

AMS is a nuclear physics techniquefor measuring radioisotopes (see Figure 5). Its use as an extremely precise,sensitive, and versatile tool in the field ofbiomedical research is relatively recent.2LLNL researchers Ken Turteltaub andJohn Vogel are responsible for much ofthe AMS work described in this article.The instrument consists of severalmass spectrometers, separated by anelectrostatic accelerator, and a detectorfor counting rare isotopes. (The box onp. 13 describes the set-up andapplications of AMS in more detail.)

Rather than measuring atomic decay,as in liquid scintillation counting, AMSisolates and counts specific nuclei,particle by particle. Depending on whichisotope is used to tag the molecules ofinterest, AMS can be up to a milliontimes more efficient than decay countingin detecting specific, tagged molecules.When we tag mutagens with one ormore radiocarbon (14C) atoms, we canquantitatively measure as little as oneDNA adduct in about a thousand cells(roughly one adduct per trillionnucleotides). This sensitivity andaccurate quantification make it possiblefor us to study DNA adduction at dosesas low as 5 nanograms per kilogram ofbody weight—close to actual humanexposures from the environment.

The Effect of Low Doses

In one series of studies, we followedthe DNA–food mutagen interactions thatoccur when rodents are given low dosesof MeIQx. This mutagen is sometimes

Figure 5. LLNLresearcher John Vogelloads samples foranalysis by acceleratormass spectrometry(AMS). Originallydesigned for measuringradioisotopes in nuclearphysics experiments,AMS has proven to beprecise and versatilein tracking very lowdoses of radioisotope-tagged food mutagensin laboratory animals.

Chromosome

(a) Normal DNA

Coiled DNA

DNA double helix

Sugar–phosphatebackbone of DNA

The fournucleotide bases

Adenine

Adenine

Cytosine

Thymine

Guanine

Guanine

Cytosine

Thymine

(b) Adduction of a cooked food mutagen

PhIP

Figure 4. DNA adduction. (a)Normally, a fixed sequence of fourmolecular units (the nucleotide bases A, T,C, and G) make up the genes. (b) When bulkymolecules (adducts) chemically bond to DNA, theydistort the helix and can block the ability of DNA toreplicate or can cause errors in replication.

Accelerator Mass Spectrometry as a Biological Tool

Accelerator mass spectrometry, or AMS, is a nuclear physicstechnique that has been developed over the past 15 years primarilyfor detecting long-lived isotopes for the earth and space sciences.Originally, it was a tool for carbon dating geological events, but itscapabilities and applications are now far-ranging. In biomedicalresearch, it is a new technology that is still developing. AMS wasfirst applied to LLNL research on DNA adducts in 1990. The mainadvantage is its precision and sensitivity for quantifyingradionuclides, especally 14C. Because it requires highly specializedequipment and expertise, its use today is limited to a few laboratories.

As shown in the illustration, AMS uses two mass spectrometersseparated by an electrostatic (Van de Graaff) accelerator.Negative ions from samples are initially sorted by alow-energy mass spectrometer (top right). The ionsare then accelerated to much higher energy in theVan de Graaff accelerator, where they pass througha thin foil that dissociates (essentially destroys)any remaining molecules. Rare ions (14C)—thosetagging the molecules of interest—are separatedfrom the more abundant, naturally occurringisotope (13C) in a high-energy mass spectrometer(top left) and are subjected to other selectiontechniques. Finally, the rare ions are counted in agas-ionization detector and reported relative tothe abundant ions measured in a Faraday cup.

The important points are that this instrumentallows for specific counting of radionuclei particle-by-particle, that all interference from molecules is

destroyed by the foil, and that the acceleration of ions to high energyallows the rare ions to be transmitted to the detector with very highefficiency. For most of our current biomedical applications, we tagmolecules of interest with the isotope 14C because of its low naturalabundance in biological material, thus giving a reduced background.We are developing animals depleted to 1% in 14C, which willincrease our sensitivity a hundredfold or more. We can achieve low levels of 14C in mouse tissue by feeding thempetroleum-based diets. We are also developingprotocols for the AMS-based measurement oftritium in biomedical research.

Low-energymass spectrometer

(selects mass 13 or 14)

Gas-ionization detector identifies particles by specific energy lossElectrostatic (Wein) filterMagnetic filter

High-energymass spectrometer

Tandem electrostatic Van de Graaff accelerator is also a molecular dissociator

Stable isotope detector (Faraday cup)

Negative-ionsputter source

Samples

Thin foil strips electronsfrom molecules

2–10 megavolts

15 m

C1+... C6+ C—

13C4+

14C

15


Food Mutagens

tissues, influence the persistence ofadducts and the development of tumors.

Picture of an Adduct

Our studies with bacterial and animalmodels indicate that only three or fourprincipal kinds of DNA adducts areformed by most food mutagenintermediates. In bacterial studies, wedeveloped a DNA cloning method toanalyze the mutated DNA sequences inspecial strains of bacteria. Bacterialstrains exposed to food mutagens showedthat these compounds have a high affinityfor DNA segments containing thenucleotide bases cytosine and guaninein an alternating sequence—a DNA“hotspot.”

We used 32P-postlabeling to look forindividual “fingerprints” that show upwhen a food mutagen intermediate bindswith DNA. Most of the adducts of IQ,MeIQx, and PhIP prefer not only theguanine in DNA, but also one particularcarbon atom position (C-8) on that base(Figure 8). Our animal studies with theother three nucleotide bases—adenine,thymine, and cytosine—show noadducts.

We currently are using a highlydetailed, quantum-chemistry approachto study the mechanism by which PhIPinteracts with DNA.3 This research is acollaborative effort among scientists inLLNL’s Biology and BiotechnologyResearch and Chemistry and MaterialsScience programs and computationalengineers at Sandia NationalLaboratories, Livermore. As part of theanalysis, we take into consideration thefact that, unlike most of the otherheterocyclic amines, the PhIP moleculeis not a flat structure. Rather, PhIP has a phenyl ring structure that is 40 to 45 degrees out of plane with the rest ofthe molecule. To study the possiblemodes of binding of a bulky moleculewith complex twists and angles (seeFigure 3), the rates of reaction change,and the specificity of binding to various

locations in DNA, we are usingultraviolet, fluorescence, and nuclearmagnetic resonance spectroscopiestogether with quantum mechanicalcalculations.

We know that PhIP bonds covalentlywith guanine, and we now believe thatsuch bonding follows the physical(noncovalent) association of thatcarcinogen with a groove of DNA(specifically, the minor groove). Suchphysical association, or groove binding,may be the initial event in the adductionof DNA by PhIP.

DNA Undergoes Repair

New data suggest that humans candiffer substantially in their susceptibilityto chemically induced cancer. Peoplecan inherit genetically based traits thatmay make them more or less prone todeveloping tumors. The inheriteddifferences may be governed by howmuch bioactivation or deactivation takesplace when foreign substances are takeninto the body, the likelihood that DNAwill undergo repair when molecules areadducted to the genetic material, and therates at which tissue cells replicate.

The fact that DNA can encode itsown repair was first demonstrated in

bacteria. For more than a decade, LLNLhas had a program investigating howDNA repair works and what genes andrepair proteins are involved. The box onp. 17 gives background information onthis important field of investigation. Anarticle in the April 1993 issue of Energyand Technology Review contains moredetail on this subject.

Cells from the ovary of the Chinesehamster offer a unique research tool forstudying DNA repair processes. Onemajor attribute is that they are fast-growing in culture. For several years,LLNL biologists have been using speciallines of these ovary cells that either do,or do not, show a biochemical deficiencyin DNA repair because of mutations. Thecells that are “repair-deficient mutants”(repair incompetent) are highly useful inevaluating what happens when foodmutagens are administered and the DNArepair process is essentially turned off.

In essence, we have found that if theDNA repair gene ERCC2 is present andfunctioning in a repair-competent cell,then DNA repair takes place afteradministering a food mutagen, and weobserve less DNA damage. On the otherhand, if that gene is absent or notfunctioning in a repair-incompetent cell,

14


Food Mutagens

used AMS to analyze adduct levels inDNA from the liver and other tissues.

Our results in Figure 6 are expressedin a standard type of graph used inbiomedical research, which plots doseon one axis and response on the other.This dose–response graph shows alinear relation over many orders ofmagnitude between the administereddose of MeIQx and the response,namely, adduct levels in the liver.Seven days after the beginning of thestudy, the levels of DNA adducts inrodents fed a low dose were proportional

to those for rodents fed one milliontimes more MeIQx for the same lengthof time. These data tell us that adductscan form at human exposure levels andthat DNA adducts can indicate theamount of exposure for this carcinogen.In addition, the data tell us that thebioactivation processes and the DNArepair mechanisms function at the samerelative rates at high and low doses.Next, we need to study what happensafter continuous exposure to thisheterocyclic amine.

In other studies, we assessed DNAadducts in a variety of tissues after givingrodents varying doses of the foodmutagens PhIP, MeIQx, or IQ. For thisresearch, we analyzed the DNA adductsusing the technique of 32P-postlabeling.By varying both dose and type ofmutagen, we can see whether differentamounts of mutagens are handleddifferently in different types of tissues.We found that the response to varyingdoses depends on the tissue type and themutagen type. As shown in Figure 7, thepancreas of mice had the greatest level ofadducts with PhIP. In contrast, wefound high levels of adducts in the liverwith MeIQx and IQ. Other studies haveshown that PhIP does not cause livertumors, whereas MeIQx and IQ do. Suchresults clearly show a correlation betweenDNA adducts and tumors in specifictissues. However, even though PhIPgenerates high levels of adducts in thepancreas, it does not appear to causepancreatic tumors in rodents.

In general, data like these suggest thatlevels of DNA adducts correlate withexposure, but not necessarily with thedevelopment of tumors in specific tissues.The fact that metabolism of a particularfood mutagen may be high in the liver,for example, but adduct levels can be lowand liver tumors infrequent, suggeststhat food mutagen metabolites mightcirculate throughout an organism. Webelieve that PhIP is an example of thistype of mutagen. It is also likely thatother factors, such as DNA repair in the

Dose, milligrams per kilogram of PhIP

10–1

5 mol

e of

add

uct p

er m

icro

gram

of D

NA

10100

20 30

Pancreas

0.4

0.8

1.2

1.6

Thymus

Heart

Liver

Kidney

Figure 7. Theresponse to varyingdoses of the mutagenPhIP depends onthe tissue type andspecies. For mice,the pancreas had the greatest level ofadducts with PhIP,followed by thethymus, heart, liver,and kidney. Ratsshow a differentprofile.

DN

A a

dduc

ts p

er tr

illio

n (1

012) n

ucle

otid

es

Dose, micrograms of MeIQx per kilograms of body weight

107

106

105

104

103

102

10

1

10–1

10–2

10–3

10–4 10–3 10–2 10–1 1

Postlabeling data(very high doses used

in cancer research)

Limit ofimmunoassay

sensitivity

Limit of 32Ppostlabelingsensitivity

Limit of AMS sensitivity

10 102 103 104 105

Figure 6. Using accelerator mass spectrometry (AMS), we analyzed the levels of DNAadducts in the livers of rodents that had been exposed to varying doses of the food mutagenMeIQx for seven days. Doses ranged from the amount in about one bite of a hamburger tothe very large amounts used in cancer studies. On the y axis, one adduct per trillionnucleotides is roughly one adduct per 1000 cells. The level of sensitivity was not possiblebefore the advent of AMS. The level of adduct formation is a linear function of dose. Data likethese tell us that DNA adducts can form at the doses equivalent to human dietary exposure.

17


Food Mutagens

then we observe more DNA damage.Moreover, we are learning that the repairthat is done seems to favor certain typesof DNA sequences—in other words,ERCC2 removes specific types ofdamage.

We have also manipulated other genesin Chinese hamster ovary cells. Forexample, in very recent work, we havetaken the genes responsible for

manufacturing enzymes that bioactivate(acetylate) food mutagens and placed themin cells to mimic human metabolism.The result, as expected, is a dramaticincrease in the amount of mutagenscaused by heterocyclic amines.

The lesson from this type of researchis that the ability to repair DNA damagehas a major impact on how harmfulfood mutagens will be after they are

eaten. If humans vary in their ability torepair DNA, as we suspect, then theywill also vary in susceptibility toheterocyclic amines and in the amountof genetic damage that occurs. To date,no one has quantified this type ofhuman variability. Therefore, the issueof differences in human susceptibility tofood mutagens represents a major newdirection for our research.

16 Food Mutagens

CH3

N H

N

N

N

O

CH3

CH3

N

N

N

N

N

N

N

N

O

N

N

N

N

N

H

N

N

NO

N

N

Thymine

Cytosine

Adenine

Guanine

Thymine

Cytosine

Adenine

Guanine

N-PhIP

Strand 1

Strand 2

Strand 1

Strand 2

The four nitrogenous bases

(a) Normal double strand of DNA

(b) The food mutagen PhIP adducted to guanine

Straininghydrogen bonds

Hydrogen bonds

H

H

N

H

H

H

N

O

H

H

H

N

O

N

H

H

N

O

N

NO

HNN

H

H

H

H

ON

H

Figure 8. (a) A normaldouble strand of DNAcompared to (b) oneshowing the adduction ofthe amino nitrogen ofPhIP to guanine. Note thedistortion of the DNA helixand the strain on thehydrogen bonds between theguanine-cytosine base pairthat results from the presenceof the adduct.

DNA Repair

DNA makes up chromosomes and genes, and it isthe genetic blueprint for life. The idea that DNAincludes genes that could encode its own repair wasfirst demonstrated in bacteria about three decades ago.We now know that virtually all living organisms, frombacteria and yeast to humans, have acquired someremarkable biochemical machinery for preserving thegenetic material.

After being subjected to certain types of “insult,”the usual double-helix structure of DNA can bechanged. In the case of many environmental pollutantsand food mutagens, bulky molecules (adducts) becomeattached to DNA and distort the phosphate backbone ofthis macromolecule. However, DNA has the ability toessentially fix itself by the action of DNA repairproteins. LLNL researchers have been studying theDNA repair process for more than a decade.

Several types of DNA damage-containment systemsare known today. We have focused on the details of themost important repair system in human cells, one thatcorrects DNA damage produced by nearly all bulkychemicals, such as food mutagens, and thephotoproducts of ultraviolet radiation. This precise,multistep process is called nucleotide excision repair(see the illustration). In essence, a certain fixed length ofthe DNA backbone on both sides of the damaged site iscut, the segment containing the damage is removed,and new material is synthesized in a predetermineddirection to produce a repair patch. Finally, the repairpatch is tied or rejoined to the parental DNA strand (aprocess called ligation). Our recent evidence suggeststhat 30 to 50 different proteins play a role in thedamage-recognition, cutting, and mending process.

If DNA repair does not take place or if it is doneincorrectly, mutations can occur. The consequences forthe affected individual can be profound, including thetriggering of tumors and cancer.

Helix-distorting defect

Damage recognition and incision

Excision of ~29 nucleotides

Polymerization (synthesis)

Ligation (rejoining) repair patch º 29 nucleotides

Ligase

DNA backbone

DNA polymerase

Nucleotide excision repair process

CH3

NH2N

N

N

CH3

NH2N

N

CH3

NH2N

N

N

N


from most to least potent: MeIQ, Trp-P-1, IQ, MeIQx, Glu-P-1, Glu-P-2,Trp-P-2, MeAαC, PhIP, and AαC.The potency values we obtained wereslightly higher than those suggested inprevious studies. We estimate thatDiMeIQx could be among the mostcarcinogenic of all the heterocyclicamines identified to date on the basis of its mutagenic activity, although notumor data are available yet. Thesenew estimates of potency serve as animportant piece of information indetermining the ultimate cancer risk to humans from eating food mutagens.

Risk and Dose

Before we can say anything about aconnection between food mutagens andcancer risk in humans, we must firstestablish what doses or intake levels arerealistic. The difficulty of this taskarises from the many sources ofpossible error. People eat a variety offoods, and they prepare foods indifferent ways. Cooking descriptionsare often not well defined, and cookingtimes and temperatures have large effectson the formation of promutagens.

Although the relative amounts of theheterocyclic amines found in foods aregenerally consistent among differentstudies and laboratories, the preciseamount of one particular mutagen pergram of a given cooked food can spana tenfold range. A recent study ofcommercially cooked meat showedthat products cooked differently had a 400-fold range of mutagenic activity.There are several explanations for the lack of concordance, the mostimportant having to do with the time and temperature of cooking.

Thus, estimating the risk fromexposure to carcinogens depends onmaking several generalizations andassumptions. Some of the important

variables for estimating the cancer riskassociated with food mutagens include:• What type of risk to estimate; forexample, the maximum (upper-bound)credible risk or some more conservativeestimate.• Which population to use; for example,worldwide, the U.S., or a high-exposuresubgroup in the U.S.• Which of the more than dozen knownfood mutagens are most prevalent in thediet of the chosen population.• The level of chronic dietary exposure.• An estimated human lifetime dose forthe chosen mutagens.• Data on the cancer potency ofheterocyclic amines prevalent in the diet.

Assumptions must also be made, each adding uncertainty to the estimates.The most unavoidable are that:• The relation between dose andresponse (for example, tumorsin animal studies) isessentially linear.

• In extrapolating from one species toanother, a given dose has equal toxicity.As with other assumptions, this one maybe reasonable, but it also entails someuncertainty because repair processes andmetabolism do differ from rodents toman.

How Big Is the Risk?

We recently made an improvedestimate of the cancer risk to humansposed by the presence of heterocyclicamines in the U.S. diet.4 This work wasdone in collaboration with Ken Bogenand Dave Layton in LLNL’s Health andEnvironmental Assessment Division.More refined estimates are now possiblebecause research over the past few yearshas given us better values for the levelsof mutagens in foods, the genetic

toxicity of mutagens, and theircarcinogenic potency.

19


Food Mutagens


EVERAL of the most importantfood mutagens in the diet become

highly reactive with DNA and formbulky, helix-distorting lesions calledadducts. What, then, is the likelihoodthat such adducts will lead to cancer?

Most of the initial research on cancerarising from food mutagens was done inthe 1980s by researchers in Japan usingmice and rats. Feeding animals largedoses of the mutagens IQ, 4-MeIQ, and8-MeIQx induced tumors in the liver.For example, 27 of 36 female micedeveloped liver tumors after being fedIQ for 96 weeks. Lung tumors wereincreased with IQ and 8-MeIQx,stomach tumors with IQ and 4-MeIQ,and intestinal tumors with 8-MeIQx.Other studies have found an increasedincidence of skin, colon, small intestine,clitoral, mammary gland, and ear ducttumors in animals fed IQ.

Recent studies show that most of theheterocyclic amines induce tumors atmultiple sites at least in some species orsexes of rodents tested experimentally.For example, PhIP, which is highlymutagenic in mammalian cells, inducesboth breast and colon cancers in ratsand lymphomas in mice. Tumors of theliver have also been demonstrated after

IQ was fed to monkeys. Thus, we havegood reason to think that the foodmutagens might be potent carcinogensin humans. Researchers believe that thegastrointestinal tract and breast ofhumans may be targets for tumorsinduced by PhIP.

Standard methods previously used byother researchers to estimate the cancerpotency of food mutagens were basedon their potency in inducing specifictypes of tumors, not on the total tumor-inducing potency. But, in fact, we nowknow that most of the heterocyclicamines are multipotent. This meansthat they can induce tumors in manydifferent and distinct types of tissues.We suspected that the earlier studieson cancer potency might haveunderestimated the actual potentialhuman risk of cancer because theaggregate potency was not taken intoaccount. Thus, within the last year, wepublished new estimates of the potentialhuman cancer potency for tenheterocyclic amines.1

Our new estimates are derived fromexperimental data on 36 differentspecies, strains, or sexes of animals—mostly mice and rats—that were fedheterocyclic amines and subsequentlyshown to develop one or moremalignant or potentially malignanttumors. We considered 82 differenttypes of cancer potency associated withspecific tumors plus 24 additionalestimates of aggregate potency.

We found that the carcinogenicpotencies of the ten heterocyclic amineswe investigated can be rankedapproximately in the following order

Daily intake,

nanograms of

mutagen per

kilogram of

body weight

Beef

Fish

Lamb

Food group Bacon

Pork

Chicken

8

7

6

5

4

3

2

1

0 PhIPMelQxDiMelQxIQ

AåC

Heterocyclic amine

Figure 9. Averagedaily intake in theAmerican diet offive heterocyclicamines by type offood. The primarymutagen-bearingfood is fish,followed by beefand chicken.

SS

18

Cancer RiskFood Mutagens:

Cancer Risk

Fish(40%)

Beef(41%)

Bacon(7%)Pork

(3%)

Lamb(1%)

Chicken(8%)

literature values, is fish, followed bybeef and chicken. However, we mustremember that cooking methods andcancer potencies of the heterocyclicamines are extremely importantvariables in estimating risk.

In contrast to daily intake, we findthat the carcinogenic potencies of thefive heterocyclic amines have almostthe reverse order of their prevalence inthe diet: IQ, DiMeIQx, MeIQx, PhIP,AαC. Thus, whereas averageAmericans consume less IQ by weightthan, say, PhIP, IQ is much more potent.Cancer potency depends on severalfactors, including the effectivebiological dose within target tissues(such as the liver, colon, or pancreas)and the likelihood of tumors actuallybeing induced in those tissues.

We estimated the risk of cancer bymultiplying the intake of the five majorheterocyclic amines by their cancerpotencies. Figure 10 shows theincremental risk of cancer from each ofthe five principal heterocyclic amines.PhIP accounts for nearly half (46%) ofthe total risk, followed by MeIQx (27%)and DiMeIQx (15%). The mutagens IQ(7%) and AaC (5%) contribute the leastrisk of the five substances.

Another way to view the results is bylooking at the risks arising from each ofthe food groups we studied. Figure 11shows that the highest cancer risks, byfar, result from eating beef (steaks andground beef) and fish.

We estimate that the overall cancerrisk to the U.S population is one in tenthousand. Put another way, our latestprediction is that about 28,000 peopleliving in the U.S. today will developcancer during their lifetime (over 70 years) from dietary exposure to thefive heterocyclic amines included in ourcalculations. For comparison, the

American Cancer Society estimated that149,000 cases of colorectal cancersoccurred in the U.S. in the single year1994. From a public health standpointthen, the magnitude of the cancer riskwe currently predict from eatingheterocyclic amines is not alarming,but it is certainly not negligible.

Our estimates are highlyconservative, representing averageexposures and risks for the U.S.population based on available data inthe literature. Depending on individualdietary habits, some people could have10 to 50 times higher doses of foodmutagens in their diet. Such peoplewould be at much higher risk ofdeveloping cancer (up to one chance in200). Other considerations that couldlead to even higher risk include thepossibility that some individuals have

an increased susceptibility to cancer andthat humans might be even moresensitive to food mutagens than rodents.

Table 1 provides an additionalperspective on our new estimates ofcancer risk associated with foodmutagens. In cancer etiology, a riskgreater than one in a million is deemedsignificant by the EnvironmentalProtection Agency (EPA). A cancerrisk of one in ten thousand would behigh enough to trigger regulatory actionif the substance in question were anenvironmental contaminant, such as a pesticide.

21


Food Mutagens

We based our new estimates on:• Actual measurements of theheterocyclic amines in a variety offoods common in the U.S. diet.• The average intake of these commonfoods in the U.S. diet. This estimatewas based on a random dietary surveyconducted under the auspices of theU.S. Department of Agriculture. Wecomputed the average daily intake ofmeat and fish consumed by 3563

individuals who completed the mostrecent survey.

• Cancer potencies derivedfrom the results of animal

bioassays, or establishedrelations between themutagenic activity ofheterocyclic aminesand theircarcinogenic

potency (for example, in the case of4,8-DiMeIQx).

For many years, we have beenadding information on food mutagensto a database that now contains some261 records classified by food item,cooking method, and type of mutagen.Of the 13 different food mutagens inour database, we found that only fiveare commonly consumed at significantlevels in the U.S. diet. Therefore, inestimating average dietary intake forthe U.S. population, we considered onlyfive heterocyclic amines. They are, indescending order of exposure: PhIP,AαC, MeIQx, DiMeIQx, and IQ.

Figure 9 is a convenient way to showthe average daily intake of these fiveheterocyclic amines by type of food.The graph shows that the primarymutagen-bearing food, based on the

20


Food Mutagens

Table 1. Estimates of some lifetime cancer risks and cancer-related regulatoryguidelines in the U.S.

Risk to U.S. females of developing breast cancer 1 in 9

Risk in the U.S. of developing colorectal cancer 1 in 15

Estimated average cancer risk in the U.S. from eatingheterocyclic amines in cooked foods 1 in 10,000

Risk of cancer after consuming average amounts offish containing the pesticide dieldrin* 1 in 10,000

Risk of cancer after consuming average amounts offish containing the pesticide DDT* 1 in 100,000

Risk level deemed significant by the EPA in cancer etiology 1 in 1,000,000

*These pesticides have been banned for about 20 years. However, declining amounts of residues are still sometimes found in foods likefish at levels that exceed the EPA’s negligible risk standards.

Figure 11. Principal foodtypes contributing to thetotal calculated cancerrisk from ingesting thefive heterocyclic aminesshown in Figure 10. Thehighest cancer riskresults from eating beefproducts and fish.

Figure 10. Total calculated cancer risk to humans from the intake of the fiveprincipal heterocyclic amines in the U.S.diet is represented as a circle. Theentire circle represents nearly28,000 cases of cancer for theU.S. population, or a one in tenthousand chance of developingcancer over a 70-yearlifetime.

AåC(5%)IQ

(7%)

DiMelQx(15%)

MelQx(27%)

PhlP(46%)

meats contributes most to the total risk.Specific subgroups of the populationeating large amounts of muscle foodcooked well-done may be at muchhigher risk.

Key Words: accelerator mass spectrometry(AMS); adduct dosimetry; amino-imidazoazaarene (AIA); bioactivation;cancer risk assessment; carcinogen;carcinogenicity; DNA adducts; foodmutagen; mutagens—2-amino-9Hpyrido[2,3-b]indole (AαC); 2-amino-3,4,8-trimethylimidazo[4,5- f]quinoxaline(DiMeIQx); 2-amino-3-methylimidazo[4,5-f]quinoline (IQ), 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline(MeIQx), 2-amino-1-methyl-6-phenyl-imidazo[4,5-b]pyridine (PhIP); 32P-postlabeling.

Notes and References1. K.T. Bogen, “Cancer Potencies of

Heterocyclic Amines Found in CookedFoods,” Food Chemical Toxicology 32(6), 505–515 (1994). (UCRL-JC-114333)

2. “AMS in Biomedical Dosimetry:Determining Molecular Effects of Low-Level Exposure to Toxic Chemicals,”Energy and Technology Review (May/June 1991), UCRL-52000-91-5·6, pp. 48–56.

3. G. A. Marsch, R. L. Ward, M. Colvin,and K. W. Turteltaub, “Non-CovalentDNA Groove-Binding by 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine,”Nucleic Acids Research 22 (24),5408–5415 (1994). (UCRL-JC-116468)

4. D. W. Layton, K. T. Bogen, M. G.Knize, F. T. Hatch, V. M. Johnson, andJ. S. Felton, “Cancer Risk of HeterocyclicAmines in Cooked Foods: An Analysisand Implications for Research,”Carcinogenesis 16 (1), 39–52 (1995).(UCRL-JC-116654)

5. “Food Mutagens: The Role of CookedFood in Genetic Change,” Science andTechnology Review (July 1995),UCRL-52000-95-7, pp. 6–25.

The following scientists contributed tothis article: Kenneth Bogen (riskassessment), Mark Knize (chemicalanalysis), David Layton (riskassessment), James Tucker (cytogeneticanalysis), Kenneth Turteltaub (DNAadducts), Lawrence Thompson (DNArepair), John Vogel (accelerator massspectrometry), and Rebekah Wu(mutation analysis).

For further information, contactJames S. Felton (510) 422-5656([email protected]).

23


Food Mutagens

Future Research

To date, few studies have evenattempted to measure all of the knownheterocyclic amines present in cookedfoods. No systematic studies on themutagen content of cooked foods havebeen reported. Additional research inthese areas would give us improvedassessments of exposure and risk.

At present, DNA-adduct studies relyon measuring adduction averaged overthe entire complement of geneticmaterial, that is, over the entire genome.However, the sensitivity of AMS nowallows us to analyze very smallsamples. Thus, we are beginning to useAMS to investigate the formation andrepair of DNA adducts in specificgenes.

It should be possible to use AMS tomeasure DNA adducts in urine or

exfoliated cells from humans. Itis also possible to directly

measure the levels ofunmetabolized heterocyclic

amines in urine. Thesetypes of assays do

not require the

administration ofradioisotopes to humans andcould be used to estimate the variabilityof human susceptibility to carcinogens.

Our work on risk assessment carrieswith it some important implications thatwarrant follow-up. In particular, it maybe possible to identify strategies tomanage potential cancer risks associatedwith food mutagens. One key objective

would be to develop guidance oncooking methods that could reduce theconcentrations of PhIP and MeIQx, whichcontribute the most to the predictedcancer risks.

Summary

Based on the research reported in thefirst installment of this two-part serieson food mutagens,4 diets rich in well-done meat cooked (especially fried) attemperatures over 200°C will havesignificant levels of the heterocyclicamines. Meats cooked to rare or medium-rare (below 150°C, or hotter for shortperiods) have markedly less mutagencontent than well-done meats.Pretreatment of ground beef bymicrowave cooking, then discardingthe clear fluid before frying, lowers themutagen content of even well-done meat.

A comparison of fried meats showsthat beef and chicken are the mostmutagenic. When account is taken ofthe relative amounts of different meatsconsumed by Americans, and of thepotency of mutagens in them, ground beefis probably the most important sourceof food mutagens in the U.S. diet.

As reported in this installment, weare beginning to understand the processby which food mutagens becomeadducted to DNA, an important stepthat can lead to cancer. The binding ofmutagens depends on the formation ofintermediate, biologically reactivemolecules. The intermediate formsappear to link preferentially to the DNAbase guanine in many cases. The extentof DNA adduction, and the subsequentoccurrence of tumors, varies considerablyin different types of tissues and indifferent animal species.

The overall, average, upper-boundlifetime cancer risk in the U.S. fromeating heterocyclic amines in cookedfoods is estimated to be about one in tenthousand. The consumption of muscle

22


Food Mutagens

JAMES FELTON joined the Biomedical Sciences Division ofLawrence Livermore National Laboratory as a Senior BiomedicalScientist in 1976. He is currently the Group Leader of the MolecularToxicology Group of the Biology and Biotechnology Program at theLaboratory. He received his A.B. in Zoology from the University ofCalifornia, Berkeley, in 1967 and his Ph.D. in Molecular Biologyfrom the State University of New York at Buffalo in 1973. From

1973 until 1976, he was a Fellow at the National Institutes of Health in Maryland.In more than 147 professional publications, James Felton has explored the role of

diet in carcinogenesis and mutagenesis. He has been a part of the Laboratory’s researchon food mutagens since it began 17 years ago and has led it for the past 8 years.

About the Scientist

7 food mutagens - lawrence livermore national … · genetic toxicity of food mutagens like phip...

Documents