Human somatic mutation assays as biomarkers of carcinogenesis.

This paper describes four assays that detect somatic gene mutations in humans: the hypoxanthine-guanine phosphoribosyl transferase assay, the glycophorin A assay, the HLA-A assay, and the sickle cell hemoglobin assay. Somatic gene mutation can be considered a biomarker of carcinogenesis, and assays for somatic mutation may assist epidemiologists in studies that attempt to identify factors associated with increased risks of cancer. Practical aspects of the use of these assays are discussed.


Introduction
Cancer development is a multistep process that may take years in humans. Critical steps in the process appear to be the production of stable changes in the genetic material, called mutations, and subsequent cell proliferation that produces daughter cells with mutated DNA.
A number ofmutational events are thought to be necessary to convert a normal cell into a malignant cancer cell. For example, sequential mutational changes have been identified in tissues of the colon that are related to clinical stages in the progression of colon cancer (1). The development of somatic mutation assays as biomarkers of the carcinogenic process may, therefore, help epidemiologists link carcinogenic exposures to cancer outcome in humans by looking at events early on in the cancer process. This paper provides an overview of current human somatic gene mutation assays and discusses some of the practical aspects involved in their application to epidemiological study populations.

Relationship between Mutation and Cancer
The lines of evidence that link somatic mutation with cancer can be broadly summarized as follows. a) Cancers often arise in proliferating tissues in contact with the environment, e.g., the mouth, gut, skin, and lungs (2). In a tissue in which there is rapid cell proliferation, damage to DNA may not be completely repaired before cell division occurs. The damage may then be passed on and made permanent in DNA strands of the daughter cells. b) Many carcinogens bind to or cause mutation in DNA. For example, benzo[a]pyrene DNA adducts have been detected in cells exposed to the carcinogen benzo[a]pyrene (3). c) DNA from tumor cells can "tansform" normal cells. When DNA from tumor cells was transferred to normal cells, the normal cells became transformed, i.e., exhibited characteristics of tumor cells (4). d) Stable, nonrandom changes in the genetic material are associated with certain cancers. For example, most cases of chronic myelogenous leukemia exhibit a characteristic "Philadelphia" chromosome, which is the product of a reciprocal translocation between chromosomes 22 and 9 (5). e) A tumor probably arises from the clonal expansion of a single cell, based upon observations of monoclonal X-inactivation patterns in tumors (6). J) Certain genetic conditions with a high risk of cancer have been shown to have defects in DNA repair processes, e.g., ataxia telangiectasia (7) and Bloom syndrome (8).
Mutation can have consequences other than cancer. Mutation can cause birth defects or mutations transmitted in the germ line (9), and it has been implicated in the causation of disease, e.g., atherosclerosis (JO). Also, aging may be due to the accumulation of mutations with the years (11). Biomarkers of mutation may, therefore, be useful as indicators ofeffects other than cancer. It should be remembered, however, that mutation also may have no effect, since much of DNA has no known function (12).

Human Mutation Assays
Assays that measure genetic damage in humans can be divided into two broad categories: assays ofchange at the level ofgross chromosomal structure and assays of change at the level of the gene. The DNA in the nucleus is packaged into chromosomes, which are linear DNA molecules with secondary and tertiary structure. Genes are short pieces ofDNA that code for RNA and proteins, and a gene is very small compared to a chromosome: genes make up approximately 3+ % ofa chromosome, and many genes reside on each chromosome (13).
Assays of change in chromosome structure include the chromosome aberration assay, which scores the number of abnormal, broken, or missing chromosomes in metaphase cells (14); the SCE assay, which counts sister chromatid exchanges (exchanges of identical pieces of chromosomes in duplicated sister chromatids during cell replication) in metaphase cells (15); and the micronucleus assay, which measures the fiequency of micronucleus formation (chromosomes or fragments of chromosomes lost to the cytoplasm during cell division) (16). These assays are usually performed in white blood cells (lymphocytes). Such assays may reflect damage as well as mutation, since not all of the events scored reflect a heritable change that is transmitted from a cell to a daughter cell, the definition of mutation. This article focuses on human somatic mutation assays at the level of the gene.
There are practical constraints to measuring gene mutations in humans. First, the approach must use an accessible tissue. Second, the method must be able to identify mutant cells and, since mutation is a rare event, to find and enumerate these cells against a high background number of normal cells.
Only a small number of human mutation assays have been developed to date because assays on people must use normal human genetic material. This is in contrast to the large number ofassays that have been developed in nonhuman species, where the application oftechniques ofgenetic engineenng or selection can be used to make detection ofmutants relatively simple. All ofthe current human mutation assays use blood cells, and all use a change in or absence ofa normally functioning protein to detect mutations in the gene coding for the protein.
In using these assays as biomarkers of the carcinogenic process, it is preferable to measure changes in genes known to be important in cancer. Many genes have been identified to date, both oncogenes (onco-for tumor or mass) which have been shown to be activated by specific mutation in certain cancers, and antioncogenes or tumor-suppressor genes, which have been shown to be deactivated by mutation (17). Even in the cases where it is possible to detect a change in these "cancer" genes, however, there is currently no method for identifying the mutant cells against the background of normal cells. Identifying cells with mutations in cancer genes using present technology would require screening thousands or millions ofcolonies grown from individual cells, and this is not practical. Current human gene mutation assays instead screen for mutations that can be easily identified and selected for by a change or absence of a normal protein produced by specific genes. The mutation fiequency (and type) produced in these genes is then used as a surrogate for the amount and types ofmutations potentially found in cancer genes.

The HPRT Assay
The hypoxanthine-guanine phosphoribosyl transferase (HRPT) assay was the first human somatic gene mutation assay to be developed (18)(19)(20). This assay identifies and selects (finds against the background ofnormal cells) mutant cells in one step by taking advantage ofthe biochemical pathways by which a cell synthesizes DNA.
In cells, DNA is synthesized in two ways, either from nucleotide bases (adenine, dymine, guanine, and cytosine, which make up the genetic code) made de novo, or from bases recycled from degraded DNA by the so-called "salvage" pathways. HPRT is one ofthe enzymes that recycles nucleotide bases. White blood cells that have mutations at the hprt gene that lead to a nonfunctioning HRPT protein can be detected by adding a toxic analog (6-thioguanine [6I(]) ofthe nucleotide bases to the cells. Normal white blood cells incorporate the toxic analog into newly synthesized DNA, leading to cell death (Fig. La). Mutant cells that have a nonfunctioning HPRT enzyme do not incorporate the toxic analog and survive (Fig. Ib). For the HPRT mutation assay, white blood cells isolated from a human blood sample are cultured in vitro with the 6TG, and the number of surviving mutant cells is determined after a period ofcell growth (1-3 weeks) by counting the number ofcell colonies. Counting is accomplished either by an autoradiographic method or a cell cloning technique.
Mutants can be detected in this assay because only one functional copy of the hprt gene is present per cell. If two copies of the gene were present, the loss of one copy would be hard to detect, as the other copy would probably supply the missing function. Only one copy ofthe hprt gene is present because it is located on the X chromosome, and there is only one active X chromosome per cell. Humans have 22 pairs ofautosomal (i.e., nonsex) chromosomes and one pair of sex chromosomes: females have two X chromosomes and males have one X and one Y chromosome. Females randomly inactivate one ofthe two X chromosomes in each cell during development so that females will have the same amount ofgene product from genes residing on the X chromosome as do males.
Because only one chromosome carries the functional hprt gene, the HPRT assay probably does not detect mutations requiring interaction between both chromosomes of a pair. Recent evidence from cancer biology indicates that chromosomechromosome interactions could be an important subclass of mutations in cancer (17).

Chromosomal-Chromosomal Mutational Mechanisms in Cancer
Studies ofgenetic change found in several cancers suggest that chromosome-chromosome mutational mechanisms are of importance in cancer. The first clear evidence came from studies ofthe disease retinoblastoma, a childhood cancer of the eye. A prerequisite for formation of the tumor appears to be that both copies ofa gene (called rb for retinoblastoma, one copy on each chromosome 13) have been altered and have lost the ability to produce a protein that suppresses cancer. The rb gene is a member of the class of recently discovered tumor-suppressor genes. The children with the hereditary (from a parent or a de novo germ line mutation) form ofretinoblastoma have one constitutive defective copy ofthis gene, and tumors apparently arise from the loss of the remaining functional copy of rb during development (21,22).
In order to determine the molecular mechanisms by which the functional copy ofrb was lost, the tumor DNA was analyzed by new molecular techniques. It was found that two general classes ofmutational events led to the loss ofthe functional rb gene, called here gene-loss/inactivating and gene-duplicating mutations (Fig. 2) "Gene-loss/inactivating" mutations refer to changes to the functional rb gene that eliminate gene function, such as deletion ofpart or all ofthe gene, or point mutations that change the HUAlAN  The HPRT mutation assay selects mutants from T-lymphocytes by exploiting a biochemical pathway of DNA synthesis. DNA is made from new or recycled nucleotide bases. HPRT is one of the enzymes in the salvage, or recycling, pathway. In nonmutant cells with functional HPRT enzyme, a toxic analog ofnucleotide bases (6-thioguanine) is incorporated into DNA, leading to cell death. In mutant cells that lack a functional HPRT enzyme, the toxic analog is not incorporated, and the cells survive.
base sequence ofDNA. Mutations ofthis type are detected by the HPRT assay since they require that only one copy ofthe gene be present. "Gene-duplicating" mutations are those in which the inherited dysfunctional copy of the gene on the other chromosome "replaces" the good copy ofthe gene. Mitotic recombination or reduplication of the chromosome carrying the dysfunctional gene are gene-duplicating mechanisms, and they have been demonstrated in many ofthe rb cases (23). Such gene-duplicating events obviously require that both chromosomes of a pair be present.
Two recently developed human mutation assays, the glycophorin A assay (GPA) and the HLA-A assays, are able to detect chromosome-chromosome interactions because the genes studied are located on autosomal (nonsex) chromosomes. (This means there are two copies of each gene, one on each of the paired chromosomes.) Both methods assay for the loss of a surface protein on cells heterozygous for the gene making the protein. The cells must be heterozygous because a mutation at one allele (form of the gene) in a heterozygous cell will lead to complete loss ofthat form ofthe normal protein product from the cell. In a homozygous individual, a mutation in one allele would probably not affect the protein produced by the other allele. Since mutant cells are detected by the loss of the protein, these cells would not be detected as mutants.

GPA Assay
The GPA assay measures cells that have lost one form of the GPA protein present on the surface of red blood cells (23,24). The gene has two alleles called M and N and is located on chromosome 4, so a heterozygous individual will have one chromosome 4 with an M allele ofthe GPA gene and one with an N allele ofthe GPA gene (Fig. 3). Normal red blood cells from this individual will have both M and N glycoproteins on their cell surface. About 50% ofthe population is MN heterozygous, i.e., of MN blood type.
For the GPA mutation assay, red blood cells from a person with blood type MN are reacted with anti-glycoprotein antibodies that can discriminate between the M and N forms of glycoprotein. The blood cells are first fixed (surface proteins are cross-linked) to stabilize the cells and prevent agglutination when reacted with antibodies. The antibodies are labeled with fluorescent molecules-anti-N with green and anti-M with red. Thus, a normal red blood cell from a person heterozygous for the MN allele will fluoresce red and green. Blood cells that have suffered a mutation in the GPA M gene that prevents proper expression ofGPA M on the cell surface will fail to bind antibody to M and will fluoresce green only (Fig. 3). The antibody-bound cells are analyzed by a flow cytometer, which measures fluorescence from all cells and counts the number of green-only mutant cells.
Two classes ofvariant red blood cells can be detected, No and _ MN GPA N NN. N45 cells have the normal amount ofN-GPA on the cell surface and no detectable anti-M antibody binding and presumably arise from gene-loss/inactivating mutations. NN cells have twice the normal amount of N on the cell surface and no M-binding, and presumably arise from gene-duplicating mutations (Fig. 3). Finding mutant cells by screening all cells (as opposed to the HPRT assay where all normal cells are eliminated) is made possible because flow cytometers can analyze cells very rapidly. In the GPA assay, fluorescence from each of five million cells can be measured in 20 min and the mutant cells enumerated.
The GPA assay has the advantage that it can detect chromosome-chromosome interactions such as mitotic recombination. It is limited in that the red blood cell lacks a nucleus, and this precludes any investigation of mutational mechanisms at the molecular level. In addition, it cannot be proven that variant cells are mutant cells, although evidence suggests that this is the case (24).

HLA-A Assay
The recently developed HLA-A assay (25) is performed on white blood cells, which contain DNA, and therefore the assay can be used to provide detailed information on mutational mechanism. HLA-A is a surface protein found on most nucleated cells in the body and is involved in the immune response and self versus nonself discrimination (26). The HLA-A gene is on  _*"~NN chromosome 6 and is one ofthe most polymorphic loci known, which means that most people are heterozygous for the HLA-A protein, i.e., have more than one form on their cells. The HLA-A human somatic mutation assay is similar to the GPA assay in that mutants are detected by lack of antibody binding to a cell surface protein. In order to detect mutants, however, normal cells are eliminated as in the HPRT assay, rather than counted as in the GPA assay.
To determine HLA-A mutant frequencies for an individual, white blood cells are isolated from a blood sample and typed with HLA-A antibodies. The assay currently works on individuals heterozygous for A2 or A3 forms (alleles) of HLA-A. More than 60% ofan average population is heterozygous for one or the other of these alleles. Blood from an A2Ax or A3Ax blood type individual (x = other allele) is incubated with the A2 or A3 antibody, and the cells are treated with complement. Complement is a complex of serum proteins that is part ofthe body's natural immune defenses and acts to kill antibody-covered cells. The addition of complement, therefore, kills the normal A2 or A3 antibody-covered cells, and mutant cells with altered or missing HLA-A protein fail to bind the antibody and survive (Fig. 4). The cells are plated, and the surviving cells are grown into clones, which are counted to obtain a mutant frequency after correcting for growth on nonselected plates. The clones can then be analyzed to characterize the nature ofthe mutation in the cell. In a sample ofnormal individuals, mitotic recombination was responsible for an average of one-third of the mutations observed (27).  . The HLA-A assay requires blood samples from subjects heterozygous atthe HLA-A gene, a gene with many alleles. In the current assay, subjects must be heterozygous for A2 or A3, i.e., A2Ax or A3Ax, where Ax = other HLA-A allele. The assay measures mutant T-lymphocytes that lack a normal A2 or A3 protein due to mutation at the HLA-A gene. To identify mutants, T-lymphocytes are reacted with anti-HLA-A-A2 or -A3 antibody. Normal cells bind antibody and are killed by the addition ofcomplement (C'), which targets antibody-bound cells. Mutant cells do not bind antibody and survive. The assay detects gene-loss/inactivating mutants and gene-duplicating mutants, shown at the lower left.

Hb-S and Other Assays
A fourth human somatic mutation assay, Hb-S, detects the production of a mutant form of hemoglobin (Hb-S) caused by a specific point mutation in one ofthe Hb genes (28)(29)(30). Hb-S is the hemoglobin responsible for sickle cell anemia. Fluorescently labeled antibodies for the mutated Hb are added to blood preparations fixed on a slide. The slide is then scanned by microscopy (automated image analysis) to detect anti-Hb-S antibody binding (28). This assay detects only one specific mutation, a change ofthe base adenine to thymine. The assay therefore does not detect a wide range of mutational mechanisms and so has a much lower mutant frequency than is observed in either the HPRT, GPA, or HLA-A assays. However, modifications are being introduced that will increase the number of mutations detected (28).
Other assays that are currently under development look at changes in the DNA directly rather than using a cellular protein selection process (31). Such assays hold promise for the future because they could be performed on many more tissue types, as they do not require cell growth and do not require the gene product to be expressed. This could allow the detection ofmutations in tissues thought to be directly targeted by a certain exposure. Current somatic mutation assays are all performed on one tissue, the blood, and there are uncertainties inherent in extrapolating from mutations in a nontarget tissue to mutations in a target tissue. DNA assays also have the potential to be used on stored samples. In addition, the bias inherent in the phenotype selection approach (which detects only mutations that produce an altered protein) would be eliminated, since these assays would be able to detect mutations even ifthey did not lead to an altered protein.

Practical Considerations in Using These Assays in Epidemiological Studies
Two biologic characteristics of cells used in any assay, their lifespan and location in the body, affect the sensitivity ofthe assay to the mutagenic effects ofa specific exposure. The HLA-A and HPRT assays use white blood cells, specifically T-lymphocytes, which have a lifetime ofseveral years (usually estimated as 1-4+ years) and so can accumulate HLA-A mutations over this time period. The GPA and Hb-S assays detect red blood cell mutants whose mutagenic events occurred in progenitor cells in the bone marrow, either in the differentiating cells or in the stem cells that give rise to all blood cells. For mutagenic events occurring in the differentiating cell pool, a red blood cell assay can detect effects ofexposures occurring no more than to 4 to 5 months previously, as the lifespan ofthe mame erythrocyte is 120 days. Stem cells are long-lived, and stem cell mutants will persist for the lifetime of an individual, as demonstrated by the high GPA mutant frequencies remaining in atomic bomb survivors > 40 years after radiation exposure (32).
Location ofthe cells in the body also affects assay sensitivity. The GPA and Hb assays measure mutations in the bone marrow compartnent only, and mutagens must penetrate to this compartment to be detected. The HLA-A and HPRT assays record mutants in T-cells in the circulating blood, so these assays may be sensitive to a wider variety of mutagens. Practical matters ofassay protocol may limit sensitivity. White blood cell (T-lymphocyte) assays require 10 to 20 mL of blood, andthecellsstdertestmustbeisolatedfromfreshlydrawn, sterile blood. Theseassays requirecellgrowth, and lOto20days elapse before a mutant frequency canbe obtained. In contrast, the GPA assay is rapid(2 days) andrequires < 1 mLofblood, whichdoes not have to be sterile and can be analyzed for up to 2 weeks if refrigerated. IntheHb-S assay, slidescanbestoredat -20C for atleast5 months (28). TheGPAandHLA-Aassayscanbeapplied only topersonsofcertainbloodtypes, whereastheHPRTandHb-S assays can be used on all subjects, which may be an issue ifthe size ofa study populations is limited.
The HPRTassay has beenthemostwidely applied to date, and intrmation has been developed on age as a risk factor in a sample population, on radiation and chemotherapy-exposed patients, on smokers versus nonsmokers, and on persons with cancerprone syndromes (17)(18)(19). The GPA assay has been applied to radiation and chemotherapy-exposed individuals, persons with occupational exposures to styrene, persons with cancerprone syndromes, and a limited general population (23,24,32,(57)(58)(59)(60)(61)(62)(63). The HLA-A assay has been applied to the study of the effects of age and to chemotherapy-exposed individuals (25,27,64,65). The Hb-S assay has been applied to a small number of subjects with different mutagenic exposures (28) and with cancer-prone syndromes.
Further studies are required to supply epidemiologists with more information on the following subjects. a) Major risk factors for these assays, which can be confounders in a study, need to be determined. Age has been identified as a risk factor in the GPA and HLA-A assays, and both age and smoking are risk factors in the HPRT assay. b) The range of values for these assays among individuals in the general population (interindividual variability) needs to be characterized, as does the variation in assay values for given individuals over time (intra-individual variation). Some of this information is available for the HPRT assay (40). Subpopulations with extremely high or low assay values could help to identify important risk factors. c) The relation between the rate and tpes ofsomatic mutation measured in these and future assays needs to be determined relative to prior mutagenic exposures and relative to prospective risk ofclinical disease. Issues such as the relation between mutations in the tissue measured and mutations in target tissues and the validity ofusing noncancer genes as surrogates for cancer genes need to be examined in such studies.
If further information can be collected, current and future assays ofsomatic mutation in humans show promise for a range ofepidemiological studies: investigating cancer clusters; identifying occupational cohorts at risk; providing dose-response data for risk assessment; monitoring potentially harmful effects ofclinical tatments (e.g., chemotherapy); and, tirough the use of intervention studies, identfying exposures that are mutgenic, and possibly carcinogenic, to humans.