Chapter 9: Genetic Toxicology

Genetic toxicology is a branch of the field of toxicology that assesses the effects of chemical and physical agents on the hereditary material (DNA) and on the genetic processes of living cells. Such effects can be assessed directly by measuring the interaction of agents with DNA or more indirectly through the assessment of DNA repair or the production of gene mutations or chromosome alterations. Given the risk assessment framework of this chapter, it is important at the outset to distinguish between genotoxicity and mutagenicity. Genotoxicity covers a broader spectrum of endpoints than mutagenicity. For example, unscheduled DNA synthesis (UDS), sister chromatid exchanges (SCEs), and DNA strand breaks are measures of genotoxicity, not mutagenicity because they are not themselves transmissible from cell to cell or generation to generation. Mutagenicity on the other hand refers to the production of transmissible genetic alterations. In the last few years, there has been an increased emphasis on the role of epigenetic changes in the production of altered phenotypes. Such changes can be transmitted and so it is appropriate to include epigenetic changes such as alterations in DNA methylation or in histones involved in the control of gene expression as genotoxic endpoints. However, they are not mutations by definition because they do not involve changes in DNA sequence (Hamilton, 2011).

This chapter discusses the history of the development of the field of genetic toxicology, the use of genetic toxicology data in cancer and genetic risk assessments, the mechanisms underlying genetic toxicology assays, the assays that can be used for detecting genotoxic endpoints, the use of the same assays for better understanding mechanisms of mutagenesis, and new methods for the assessment of genetic alterations. The field is evolving rapidly, and a review of its past and present state will set the stage to allow for a consideration of what are likely next major landmarks.

The field of genetic toxicology can be considered to have its roots in the pioneering work of H.J. Muller (1927), who showed that x-rays could induce mutations in the fruit fly, Drosophila. In his studies he showed not only that radiation exposure could increase the overall frequencies of mutations but also that the types of mutations induced were exactly the same in effect, or phenotype, as those observed in the absence of radiation exposure. Thus, the induced mutagenic responses should be assessed in relation to background mutations. As a conclusion to this study of radiation-induced mutations, Muller predicted the utility of mutagenesis studies not only for the study of mutations themselves but also for gene mapping approaches.

Karl Sax (1938) built upon Muller's original studies of radiation-induced mutations by showing that X-rays could also induce structural alterations to chromosomes in Tradescantia pollen grains. Sax and his colleagues, notably in the absence of knowledge of DNA structure and chromosomal organization, showed that at least two critical lesions in a nuclear target are required for the production of an exchange within (intrachromosome) or between (interchromosome) chromosomes. We know now that the lesions identified by Sax are DNA double-strand breaks, base damages, or multiply damaged sites (reviewed by Ward, 1988). In addition, Sax and colleagues (Sax, 1939; Sax and Luippold, 1952) showed that the yield of chromosome aberrations was reduced if the total dose of x-rays was delivered over extended periods of time or split into two fractions separated by several hours. These observations led to the concept of restitution of radiation-induced damage, which was later recognized as involving specific DNA repair processes (see below).

Consideration of the genetic effects of exogenous agents on cells was expanded to include chemicals in 1946, when Charlotte Auerbach and colleagues reported that mustard gas could induce mutations in Drosophila and that these mutations were phenotypically similar to those induced by x-rays (Auerbach and Robson, 1946). Thus, the field of chemical mutagenesis was initiated to run in parallel with studies of radiation mutagenesis. These original studies of Auerbach (actually conducted in 1941) are placed in a historical and biological perspective by the delightful review of Geoffrey Beale (1993).

Although the scientific value of the analysis of mutations in Drosophila was clear, there was an impression that the extrapolation to predict similar effects in human populations was too wide a step. Thus, a research effort of great magnitude was initiated to attempt to assess radiation-induced mutations in mice. This effort resulted in the publication by William Russell (1951) of data on x-ray–induced mutations using a mouse-specific-locus mutation assay. These data clearly showed that the type of results obtained with Drosophila could be replicated in a mammalian system. The mouse tester strain developed for the specific-locus assay has recessive mutations at seven loci coding for visible mutations, such as coat color, eye color, and ear shape. This homozygous recessive tester strain can be used for identifying recessive mutations induced in wild type genes at the same loci in mice treated with radiation or chemical mutagens. It was noteworthy that the mutation rate for x-ray–induced mutations in germ cells was similar in mouse and Drosophila. Subsequent studies by Liane Russell and colleagues showed that chemicals could induce mutations at the same seven loci (Russell et al., 1981).

Over the next 20 years, genetic toxicologists investigated the induction of mutations and chromosomal alterations in somatic and germ cells, largely following exposures to radiation, but increasingly using chemical mutagens as well. The ability to grow cells in vitro, either as primary cultures or as transformed cell lines, enhanced these quantitative studies. The in vitro culture of human lymphocytes, stimulated to reenter the cell cycle by phytohemagglutinin, greatly expanded the information on the assessment of chromosomal alterations in human cells (an excellent review by Hsu [1979] is recommended). It also became feasible to use cytogenetic alterations in human lymphocytes as a biodosimeter for assessing human exposures to ionizing radiations (Bender and Gooch, 1962).

Two events during the 1970s served to expand the utility of mutagenicity data into the realm of risk assessment. The Millers and their colleagues (Miller and Miller, 1977) showed that chemical carcinogens could react to form stable, covalent derivatives with DNA, RNA, and proteins both in vitro and in vivo. In addition, they reported that these derivatives could require the metabolism of the parent chemical to form reactive metabolites. This metabolism is required for some chemicals to become mutagens and carcinogens. Metabolic capability is endogenous in vivo, but most cell lines in vitro do not express this capacity. To overcome this for in vitro mutagenicity studies, Heinrich Malling and colleagues developed an exogenous metabolizing system based on a rodent liver homogenate (S9) (Malling and Frantz, 1973; Malling, 2004). Although this exogenous metabolism system has had utility, it does have drawbacks related to species and tissue specificity and loss of cellular compartmentalization. The development of transgenic cell lines containing P450 genes has overcome this drawback to some extent (Sawada and Kamataki, 1998; Crespi and Miller, 1999).

The second development in the 1970s that changed the field of genetic toxicology was the development by Bruce Ames et al. (1975) of a simple, inexpensive mutation assay with the bacterium Salmonella typhimurium. This assay can be used to detect chemically induced reverse mutations in several histidine genes and can include the exogenous metabolizing S9 system described above. The Ames assay, as it is generally called, has been expanded and modified to enhance its performance as discussed below (under section “Gene Mutations in Prokaryotes”). The assay has been used extensively, especially for hazard identification, as part of the cancer risk assessment process. This use was based on the assumption that carcinogens were mutagens and that cancer required mutation induction. This latter dogma proved to be somewhat inhibitory to the field of genetic toxicology because it provided a framework that was too rigid. Nonetheless, over the decade of the mid-1970s to mid-1980s somewhere on the order of 200 short-term genotoxicity and mutagenicity assays were developed for screening potentially carcinogenic chemicals. The screens included mutation induction, DNA damage, DNA repair, and cell killing or other genotoxic activities.

Several international collaborative studies were organized to establish the sensitivity and specificity of a select group of assays as well as to assess interlaboratory variation (International Program on Chemical Safety [IPCS], 1988). Most assays were able to detect carcinogens or noncarcinogens with an efficiency of about 70% as compared with the outcome of two-year cancer bioassays. There are a number of possible reasons for the imperfect correspondence, the most likely being that there is a group of chemical carcinogens that do not induce cancer by a direct mutagenic action. The latter point was addressed to some extent by Tennant et al. (1987), who compared the effectiveness of a small standard battery of well-characterized short-term assays to identify carcinogens. Again, this battery predicted about 70% of known carcinogens. Subsequently, the lack of a tight correlation between carcinogenicity and mutagenicity (and the converse, noncarcinogenicity and nonmutagenicity) was found to be due to the fact that some chemicals were not directly mutagenic but instead induced the damage necessary for tumor development indirectly by, for example, clonally expanding preexisting mutant cells (ie, tumor promotion) or through the production of reactive oxygen species. Such chemicals were given the rather unfortunate name of nongenotoxic to contrast them with genotoxic ones; the classification as not directly mutagenic is more appropriate. In the context of the mechanism of their mutagenicity, it is preferable to distinguish between DNA-reactivity and its correlate non-DNA-reactivity. Emphasis has recently been placed on identifying mechanisms whereby nondirectly mutagenic chemicals can be involved in tumor production. Those identified include cytotoxicity with regenerative cell proliferation, mitogenicity, receptor-mediated processes, changes in methylation status, and alterations in cell–cell communication.

In the last 10 years or so, the field of genetic toxicology has moved away from the short-term assay approach for assessing carcinogenicity to a much more mechanistic approach, fueled to quite an extent by the advances in molecular biology. The ability to manipulate and characterize DNA, RNA, and proteins and to understand basic cellular processes and how they can be perturbed has advanced enormously over this period. Knowing how to take advantage of these technical developments is paramount. This chapter addresses these changes in approach to genetic toxicology: the assays for qualitative and quantitative assessment of cellular changes induced by chemical and physical agents, the underlying molecular mechanisms for these changes, and how such information can be incorporated into cancer and genetic risk assessments. In addition, the way forward for the field is addressed in the form of an epilogue. Thus, the preceding historical overview sets the stage for the rest of the chapter.

The importance of mutations and chromosomal alterations for human health is evident from their roles in genetic disorders, including birth defects and cancer. Therefore, mutations in both germ cells and somatic cells need to be considered when an overall risk resulting from mutations is concerned.

Somatic Cells

An association between mutation and cancer has long been evident, such as through the correlation between the mutagenicity and carcinogenicity of chemicals, especially in biological systems that have the requisite metabolic activation capabilities. Moreover, human chromosome instability syndromes and DNA repair deficiencies are associated with increased cancer risk (Friedberg, 1985). Cancer cytogenetics has greatly strengthened the association in that specific chromosomal alterations, including deletions, translocations, inversions, and amplifications, have been implicated in many human leukemias and lymphomas as well as in some solid tumors (Rabbitts, 1994; Zhang et al., 2010).

Critical evidence that mutation plays a central role in cancer has come from molecular studies of oncogenes and tumor-suppressor genes. Oncogenes are genes that stimulate the transformation of normal cells into cancer cells (Bishop, 1991). They originate when genes called proto-oncogenes, involved in normal cellular growth and development, are genetically altered. Normal regulation of cellular proliferation requires a balance between factors that promote growth and those that restrict it. Mutational alteration of proto-oncogenes can lead to overexpression of their growth-stimulating activity, whereas mutations that inactivate tumor-suppressor genes, which normally restrain cellular proliferation, free cells from their inhibitory influence (Hanahan and Weinberg, 2000, 2011).

The action of oncogenes is genetically dominant in that a single active oncogene is expressed, even though its normal allele is present in the same cell. Proto-oncogenes can be converted into active oncogenes by point mutations or chromosomal alterations. Base pair substitutions in ras proto-oncogenes are found in many human tumors (Bishop, 1991; Barrett, 1993; Croce, 2008). Among chromosomal alterations that activate proto-oncogenes, translocations are especially prevalent (Rabbitts, 1994, Croce, 2008; Zhang et al., 2010). For example, Burkitt lymphoma involves a translocation between the long arm of chromosome eight, which is the site of the c-MYC oncogene, and chromosome 14 (about 90% of cases), 22, or 2. A translocation can activate a proto-oncogene by moving it to a new chromosomal location, typically the site of a T-cell receptor or immunoglobulin gene, where its expression is enhanced. A similar translocation-based mechanism also applies to various other hematopoietic cancers. Alternatively, the translocation may join two genes, resulting in a protein fusion that contributes to cancer development. Fusions have been implicated in other hematopoietic cancers and some solid tumors (Rabbitts, 1994; Croce, 2008; Zhang et al., 2010). Like translocations, other chromosomal alterations can activate proto-oncogenes, and genetic amplification of oncogenes can magnify their expression (Bishop, 1991; Croce, 2008).

Mutational inactivation or deletion of tumor-suppressor genes has been implicated in many cancers. Unlike oncogenes, the cancer-causing alleles that arise from tumor-suppressor genes are typically recessive in that they are not expressed when they are heterozygous (Evans and Prosser, 1992). However, several genetic mechanisms, including mutation, deletion, chromosome loss, and mitotic recombination, can cause loss of heterozygosity (LOH), in which the normal dominant allele is inactivated or eliminated. LOH leads to the expression of the recessive cancer gene in a formerly heterozygous cell (Cavenee et al., 1983; Turner et al., 2003; Reliene et al., 2007). The inactivation of tumor-suppressor genes has been associated with various cancers, including those of the eye, kidney, colon, brain, breast, lung, and bladder (Fearon and Vogelstein, 1990; Marshall, 1991). Gene mutations in a tumor-suppressor gene called P53, located on chromosome 17, occur in many different human cancers, and molecular characterization of P53 mutations has linked specific human cancers to mutagen exposures (Harris, 1993; Aguilar et al., 1994; Royds and Iacopetta, 2006).

In the simplest model for the action of tumor-suppressor genes, two events are considered to be required for the development of retinoblastoma, a tumor of the eye, because both normal alleles must be inactivated or lost (Knudson, 1997). In sporadic forms of the cancer (ie, no family history), the two genetic events occur independently, but in familial forms (eg, familial retinoblastoma), the first mutation is inherited, leaving the need for only a single additional event for expression. The strong predisposition to cancer in the inherited disease stems from the high likelihood that a LOH will occur by mutation, recombination, or aneuploidy in at least one or a few cells in the development of the affected organ. The simple model involving two events and a single pair of alleles cannot explain all observations concerning tumor-suppressor genes because many cancers involve more than one tumor-suppressor gene. For example, the childhood kidney tumor called Wilms tumor can be caused by damage in at least three different genes (Marshall, 1991), and colorectal carcinomas are often found to have lost not only the wild-type P53 tumor-suppressor gene but also other tumor-suppressor genes (Fearon and Vogelstein, 1990; Stoler et al., 1999). Moreover, a single mutation in a tumor-suppressor gene, even though not fully expressed, may contribute to carcinogenesis. For example, a single P53 mutation in a developing colorectal tumor may confer a growth advantage that contributes to the development of the disease (Venkatachalam et al., 1998). Subsequent LOH will increase the growth advantage as the tumor progresses from benign to malignant (Fearon and Vogelstein, 1990). In this regard (mutation and selection), carcinogenesis has been likened to an evolutionary process, with genomic instability providing the substrate and with growth advantage as the selection pressure (Gatenby and Vincent, 2003; Fischer et al., 2004).

Many cancers involve both activation of oncogenes and inactivation of tumor-suppressor genes (Fearon and Vogelstein, 1990; Bishop, 1991; Croce, 2008). The observation of multiple genetic changes supports the view that cancer results from an accumulation of genetic alterations and that carcinogenesis is a multistep process (Kinzler and Vogelstein, 1996; Hahn et al., 1999; Stoler et al., 1999; Croce, 2008). At least three stages have been defined in carcinogenesis: initiation, promotion, and progression (Barrett, 1993). Initiation involves the induction of a genetic alteration, such as the mutational activation of a ras proto-oncogene by a mutagen. It is an irreversible step that starts the process toward cancer. Promotion involves cellular proliferation in an initiated cell population. Promotion can lead to the development of benign tumors such as papillomas. Agents called promoters stimulate this process. Promoters may be mutagenic but are not necessarily so. Progression involves the continuation of cell proliferation and the accumulation of additional irreversible genetic changes; it is marked by increasing genetic instability and malignancy. More recent studies are beginning to change this view, leading to the concept of acquired capabilities (Hanahan and Weinberg, 2000). In their Hallmarks of Cancer, Hanahan and Weinberg (2000) describe a set of six acquired characteristics that are essential for the formation of all tumors irrespective of tumor type and species. These characteristics are broadly described as follows: self-sufficiency in growth signals, insensitivity to antigrowth signals, evading apoptosis, limitless replicative potential, sustained angiogenesis, and tissue invasion and metastasis. It seems probable that there is no specific order for obtaining these characteristics. Hanahan and Weinberg have revisited their hallmarks concept, and, based on progress in the field over the past decade, they have added two emerging hallmarks of potential generality to their existing six (Hanahan and Weinberg, 2011). These are reprogramming of energy metabolism and evading immune destruction. In addition, they describe genome instability and inflammation as general features that underlie their now eight hallmarks (Hanahan and Weinberg, 2011).

Genomic instability is a feature of all cancers, with the great majority having a large number of chromosomal and gene mutations and aneuploidies. In fact, one of the difficulties in trying to unravel the mechanisms of formation of tumors is establishing which genetic alterations are informative and which are merely incidental and a product of the cancer process itself. The advent of ultrahigh throughput sequencing has allowed for the complete sequencing and subsequent characterization of genomic changes in a number of tumor types. The outcome has been an ability to classify the genomic changes as drivers and passengers, with the former being essential key events in tumor formation, and the latter basically “coming along for the ride” (Stratton et al., 2009; Pleasance et al., 2010).

Gene mutations, chromosome aberrations, and aneuploidy are all implicated in the development of cancer. Mutagens and clastogens (chromosome breaking agents) contribute to carcinogenesis as initiators. Their role does not have to be restricted to initiation, however, in that mutagens, clastogens and aneugens (agents that induce aneuploidy) may contribute to the multiple genetic alterations that characterize progression or the development of acquired capabilities. Other agents that contribute to carcinogenesis, such as promoters, need not be mutagens. However, the role of mutations is critical, and analyzing mutations and mutagenic effects is essential for understanding and predicting chemical carcinogenesis.

Germ Cells

The relevance of gene mutations to health is evident from the many disorders that are inherited as simple Mendelian characteristics (Mohrenweiser, 1991). About 1.3% of newborns suffer from autosomal dominant (1%), autosomal recessive (0.25%), or sex-linked (0.05%) genetic diseases (National Research Council [NRC], 2007a,b; Sankaranarayanan, 1998; Elespuru and Sankaranarayanan, 2007). Molecular analysis of the mutations responsible for Mendelian diseases has revealed that almost half of these mutations are base pair substitutions; of the remainder, most are small deletions (Sankaranarayanan, 1998; Elespuru and Sankaranarayanan, 2007).

Many genetic disorders (eg, cystic fibrosis, phenylketonuria, Tay-Sachs disease) are caused by the expression of recessive mutations. These mutations are mainly inherited from previous generations and are expressed when an individual inherits the mutant gene from both parents. New mutations make a larger contribution to the incidence of dominant diseases than to that of recessive diseases because only a single dominant mutation is required for expression. Thus, new dominant mutations are expressed in the first generation. If a dominant disorder is severe, its transmission between generations is unlikely because of reduced fitness. For dominants with a mild effect, reduced penetrance, or a late age of onset, the contribution from previous generations is undoubtedly larger than it is for mutations with severe early expression. Estimating the proportion of all Mendelian genetic diseases that can be ascribed to new mutations is not straightforward; a rough estimate is 20% (Shelby, 1994).

Besides causing diseases that exhibit Mendelian inheritance, gene mutations undoubtedly contribute to human disease through the genetic component of disorders with a complex etiology (Sankaranarayanan et al., 1999; Sankaranarayanan, 2006). Some 3% (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2001) or 5% to 6% (Sankaranarayanan, 1998) of infants are affected by congenital abnormalities; if one includes multifactorial disorders that often have a late onset, such as heart disease, hypertension, and diabetes, the proportion of the population affected increases to more than 60% (Sankaranarayanan, 1998; UNSCEAR, 2001; Sankaranarayanan, 2006). Such frequencies are necessarily approximate because of differences among surveys in the reporting and classification of disorders. A higher prevalence would be found if less severe disorders were included in the tabulation. Nevertheless, such estimates provide a sense of the large impact of genetic disease.

Sensitive cytogenetic methods have led to the discovery of minor variations in chromosome structure that have no apparent effect. On the contrary, other relatively minor structural chromosome aberrations cause fetal death or serious abnormalities. Aneuploidy (gain or loss of one or more chromosomes) also contributes to fetal deaths and causes disorders such as Down syndrome. About four infants per 1000 live births have syndromes associated with chromosomal abnormalities, including translocations and aneuploidy. The majority of these syndromes (about 85%) result from trisomies (NRC, 2007a,b; Griffin, 1996; Nagaishi et al., 2004). Most of the adverse effects of chromosomal abnormalities occur prenatally. It has been estimated that 5% of all recognized pregnancies involve chromosomal abnormalities, as do about 6% of infant deaths and 30% of all spontaneous embryonic and fetal deaths (Mohrenweiser, 1991; Nagaishi et al., 2004). Among the abnormalities that have been observed, aneuploidy is the most common, followed by polyploidy. Structural aberrations constitute about 5% of the total. Unlike gene mutations, many of which are inherited from the previous generation, about 85% of the chromosomal anomalies observed in newborns arise de novo in the germ cells of the parents (Mohrenweiser, 1991). The frequency of aneuploidy assessed directly in human sperm, initially by standard karyotyping and more recently by fluorescence in situ hybridization (FISH), is 3% to 4%; about 0.4% are sex chromosome aneuploidies (Martin et al., 1991, 1996). The frequency of aneuploidy in human oocytes is about 18% (Martin et al., 1991).

Cancer Risk Assessment

The formalized process for conducting a cancer risk assessment has many variations based on national requirements and regulations. A summary of some of the different approaches can be found in Moolenaar (1994). A recent approach developed by the European Commission under the acronym REACH (Registration, Evaluation, Authorisation and Restriction of Chemical Substances) aims to ensure that the necessary information is acquired to assess hazards and risks for human health for the very large number of chemicals for which little information is currently available (ec.europa.eu/environment/chemicals/reach/reach_intro.htm). There are also ongoing attempts, for example, by IPCS and the International Life Sciences Institute (ILSI) to develop a harmonized approach to cancer risk assessments. Although no totally unified approach is currently available, there is a framework that has been developed around a mode of action/human relevance concept that is discussed later in this section (Boobis et al., 2006; Meek et al., 2003). This framework, together with the formalized approach developed by the US Environmental Protection Agency (EPA) based on the paradigm presented by the National Research Council (NRC, 1983), is discussed here to depict the use of genetic toxicology in the risk assessment process.

Genetic toxicology data have been used until recently almost exclusively for hazard identification. Namely, if a chemical is DNA-reactive, then tumors are considered to be produced by this chemical via direct mutagenicity. This has led, in turn, to the use of the default linear extrapolation from the rodent bioassay tumor data to exposure levels consistent with human environmental or occupational exposures (EPA, 1986). The assessment of risk requires the application of a series of default options, for example, from laboratory animals to humans, from high to low exposures, from intermittent to chronic lifetime exposures, and from route to route of exposure. Default options are “generic approaches, based on general scientific knowledge and policy judgment that are applied to various elements of the risk assessment process when specific scientific information is not available” (NRC, 1994). The default options have been, in some ways, the Achilles heel of the cancer risk-assessment process because they have a very significant impact on low exposure risk but are based on an uncertain database. This concern led the EPA (1996) to develop a very different approach, initially described in the Proposed Guidelines for Carcinogen Risk Assessment, now released as the Guidelines for Carcinogen Risk Assessment (EPA, 2005). In these guidelines, the emphasis is on using mechanistic data, when available, to inform the risk assessment process, particularly for dose–response assessment and risk characterization. The goal is to develop biologically based or other forms of dose–response models for estimating cancer risk at low environmental exposures. This does, in general, bring the EPA approach into some harmony with those in other countries (Moolenaar, 1994), where a more narrative approach to risk assessment is preferred to a strictly quantitative one, and with the approaches described by IPCS (Boobis et al., 2006) and ILSI (Meek et al., 2003). The outcome of a more mechanistically based cancer risk assessment process is that there is a greater impetus to developing databases for key events in adverse outcome pathways and mechanisms of disease production in addition to the yes/no output from genotoxicity assays in support of hazard identification. The same genotoxicity assays can be used for the collection of all these types of information, and the application of molecular biology techniques has certainly aided in the pursuit of mechanisms of mutagenicity and carcinogenicity. It is anticipated that the cancer risk assessment process will evolve as these new types of data are obtained.

Some of the issues that remain to be more firmly elucidated are (1) the relative sensitivities of different species (particularly rodent and human) to the induction of organ-specific mutations and tumors by chemicals and radiation; (2) the shape of the dose response for key events (eg, genetic alterations) in the formation of tumors and for the tumors themselves at low (environmental) exposure levels, especially for genotoxic chemicals; and (3) the relative sensitivity of susceptible subpopulations of all types. A better understanding of these major issues will greatly reduce the uncertainty in cancer risk assessments by, in part, replacing default options with biological data.

The most recent EPA guidelines for cancer risk assessment (EPA, 2005) provide a framework for cancer risk assessment that utilizes a mode-of-action (MoA) as the means of describing the “necessary but not sufficient” steps required for a chemical to produce a tumor. A particular MoA can further be defined by a set of key events that are required for tumor development (Preston and Williams, 2005). In addition, the key events can be used to establish whether or not a particular MoA described for a rodent model is plausible in humans (the so-called Human Relevance Framework, Meek et al., 2003; Boobis et al., 2006). This more defined approach based on the use of the best available science can possibly be extended to include noncancer health effects (Seed et al., 2005; Boobis et al., 2008).

Genetic Risk Assessment

The approach for conducting a genetic risk assessment is less well defined than that for cancer risk. In fact, only a handful of genetic risk assessments have been conducted. An in-depth discussion of the topic can be found in the book Methods for Genetic Risk Assessment (Brusick, 1994). The reader is also referred to the genetic risk for ethylene oxide developed by the EPA (Rhomberg et al., 1990) and the discussion of this and a recalculation presented by Preston et al. (1995). These two articles serve to highlight the difficulties with and uncertainties in genetic risk assessments.

The general approach is to use rodent germ cell and somatic cell data for induced genetic alterations and human data for induced genetic alterations in somatic cells (when available) to estimate the frequency of genetic alterations in human germ cells. This is the “parallelogram approach” (Fig. 9-1) first used by Brewen and Preston (1974) for X-irradiation and subsequently more fully developed for chemical exposures by Sobels (1982). The aim of this approach is to develop two sensitivity factors: (1) somatic to germ cell in the rodent and (2) rodent to human using somatic cells. These factors can then be used to estimate genetic alterations in human germ cells. Of course, for a complete estimate of genetic risk, it is necessary to obtain an estimate of the frequency of genetic alterations transmitted to the offspring (UNSCEAR, 2001). In addition, separate genetic risk assessments need to be conducted for males and females, given the considerable difference in germ cell development and observed and predicted sensitivity differences. Of particular note with regard to genetic risk assessment, to date there has been no unequivocal demonstration of an effect that can be detected in the children following parental exposure for chemicals or ionizing radiation.

DNA Damage

The types of DNA damage produced by ionizing radiations, nonionizing radiations, and chemicals are many and varied, including single- and double-strand breaks in the DNA backbone, cross-links between DNA bases or between DNA bases and proteins, and chemical addition to the DNA bases (adducts) (Fig. 9-2). The aim of this section is to introduce the topic of DNA damage because such damage is the substrate for the formation of genetic alterations and genotoxicity in general. However, much greater detail can be found in recent reviews that are referenced at the appropriate places within each section. It should be noted that endogenous processes and exogenous agents can produce DNA damage, but mutations themselves are produced by errors in DNA repair or replication that are a consequence of the induced DNA damage.

Ionizing Radiations

Ionizing radiations such as x-rays, γ-rays, and α particles produce DNA single- and double-strand breaks and a broad range of base damages from oxidative processes (Goodhead, 1994; Wallace, 1994; Ward, 1994; Cadet et al., 2010). In addition, recent evidence indicates that multiply damaged sites or clustered lesions can be formed that appear to be more difficult to repair (Eccles et al., 2011). Such lesions consist of multi single lesions, including oxidized purine or pyrimidine bases, sites of base loss, and single-strand breaks. These multiple lesions can be formed in DNA from the same radiation energy deposition event (Blaisdell et al., 2001). The relative proportions of these different classes of DNA damage vary with type of radiation. For example, single-strand breaks and base damages predominate with x-rays, for which ionization density is sparse, whereas the frequencies of single- and double-strand breaks are more similar with α particles, for which ionization is dense. The frequencies of individual base damages have been assessed using monoclonal antibodies, for example (Le et al., 1998), but only a very few of the total spectrum of lesions have so far been studied. More recently, it has been demonstrated that the modified histone γ-H2AX can be used as a sensitive marker of DNA double-strand breaks (Nakamura et al., 2006; Mah et al., 2010).

Ultraviolet Light

Ultraviolet light (a nonionizing radiation) induces two predominant lesions, cyclobutane pyrimidine dimers and 6,4-photoproducts. These lesions have been studied extensively because they can be quantitated by both chemical and immunological methods (Friedberg et al., 1995). In part because of this feature, the repair of cyclobutane dimers and 6,4-photoproducts has been well characterized, as discussed below.

Chemicals

Chemicals can produce DNA alterations either directly (DNA-reactive) by forming adducts or indirectly by intercalation of the chemical between the base pairs (see Heflich, 1991, for a review). Intercalation of acridine compounds (eg, 9-aminoacridine) and other planar molecules in repetitive DNA sequences has long been associated with the induction of frameshift mutations in which one or two base pairs have been gained or lost (Ferguson and Denny, 1990; Hoffmann et al., 2003). Many electrophilic chemicals react with DNA, forming covalent addition products (adducts). The DNA base involved and the positions on DNA bases can be specific for a given chemical. Such specificity of DNA damage can result in a spectrum of mutations that is chemical specific, that is, a fingerprint of sorts (Dogliotti et al., 1998; Jarabek et al., 2009). Some alkylated bases can mispair, causing mutations when DNA is replicated. Alkylated bases can also lead to secondary alterations in DNA. For example, the alkyl group of an N⁷-alkylguanine adduct, which is a major adduct formed by many alkylating agents, labilizes the bond that connects the base to deoxyribose, thereby stimulating base loss. Base loss from DNA leaves an apurinic or apyrimidinic site, commonly called an AP site. The insertion of incorrect bases into AP sites causes mutations (Laval et al., 1990).

Bulky DNA adducts formed, for example, by metabolites of benzo(a)pyrene or N-2-acetylaminofluorene are recognized by the cell in a similar way to UV damages and are repaired similarly (see below). Such adducts can also hinder polymerases and cause mutation as a consequence of errors that they trigger in replication.

Endogenous Agents

Endogenous agents are responsible for several hundred DNA damages per cell per day (Lindahl, 2000). The majority of these damages are altered DNA bases (eg, 8-oxoguanine and thymine glycol) and AP sites. The cellular processes that can lead to DNA damage are oxygen consumption that results in the formation of reactive oxygen species (eg, superoxide Display Formula$O2•¯$, hydroxyl free radicals ^•OH, and hydrogen peroxide) and deamination of cytosine and 5-methylcytosine leading to uracil and thymine, respectively. The process of DNA replication itself is somewhat error-prone, and an incorrect base can be added by replication polymerases (one in about 10⁶ bases replicated). However, the frequency of misinserted bases in replication is not the sole determinant of the spontaneous mutation frequency. Living cells reduce the rate of polymerase error appreciably through error recognition and repair processes (Johnson, 2010). A high-fidelity polymerase may make one error in a million bases while copying roughly 300 base pairs per second, but polymerase proofreading and mismatch correction then lower the error rate as much as 1000-fold (Johnson, 2010). The result is a low spontaneous mutation rate. A recent study based on the sequencing of 19 Escherichia coli genomes in a 40,000-generation experiment estimated the rate of spontaneous base pair substitutions to be 8.9 × 10⁻¹¹ per base pair per generation (Wielgoss et al., 2011). Despite the fidelity of these biological processes, the frequencies of endogenously produced DNA damages can be increased by exogenous (genotoxic) agents (Swenberg et al., 2011).

DNA Repair

The cell is faced with the problem of how to cope with the quite extensive DNA damage that it sustains. In a general sense, two processes are present to achieve this. If the damage is extensive, the cell can undergo apoptosis (programmed cell death), effectively avoiding its becoming a mutant cell (Evan and Littlewood, 1998). If the damage is less severe, it can be repaired by a range of processes that are part of a generalized cellular DNA damage response network that returns the DNA to its undamaged state (error-free repair) or to an improved but still altered state (error-prone repair). As a feature of this error-prone repair, it has been demonstrated that a family of polymerases, the eukaryotic translesion synthesis polymerases (eg, human Y-family polymerases eta, iota, kappa, and Rev1), can bypass lesions that otherwise would block replication by the normal processive polymerases (Rattray and Strathern, 2003; Prakash et al., 2005). These polymerases have the ability to bypass specific DNA lesions or groups of lesions. The result of the bypass can be an incorrect DNA sequence or a correct one depending on the induced lesion and the particular bypass polymerase. The basic principles underlying most repair processes (but not translesion synthesis) are damage recognition, followed by either direct reversal of the damage (eg, sealing of strand breaks or cleavage of pyrimidine dimers) or removal of the damage, repair DNA synthesis, and ligation. In order to achieve this for different types of DNA lesions, cells have modified the protein complexes used for other housekeeping processes (eg, transcription, replication, and recombination). This chapter presents a brief outline of the major classes of DNA repair; much greater detail can be found in the reviews provided for each section and general reviews by Van Houten and Albertini (1995), Friedberg (2000), Wood et al. (2005), and Bansbach and Cortez (2011).

Base Excision

Repair The major pathways by which DNA base damages are repaired involve a glycosylase that removes the damaged base, causing the production of an apurinic or apyrimidinic site that can be filled by the appropriate base or processed further (Demple and Harrison, 1994; Seeberg et al., 1995; Wood, 1996; McCullough et al., 1999; Sung and Demple, 2006). The resulting gap from this further processing can be filled by a DNA polymerase, followed by ligation to the parental DNA. The size of the gap is dependent on the particular polymerase involved in the repair (ie, polymerase β for short patches; polymerase δ or ∊ for longer patches). Oxidative damage, either background or induced, are important substrates for base excision repair (Lindahl, 2000). The role of translesion bypass polymerases in the repair of DNA base alterations is discussed above.

Nucleotide Excision Repair

The nucleotide excision repair (NER) system provides the cell's ability to remove bulky lesions from DNA. In the past decade the NER process has been studied extensively, and a complete characterization of the genes and proteins involved has been obtained (Reardon and Sancar, 2006; Reed, 2011). NER uses about 30 proteins to remove a damage-containing oligonucleotide from DNA. The basic steps are damage recognition, incision, excision, repair synthesis, and ligation. The characterization of these steps has been enhanced by the use of rodent mutant cell lines and cells from individuals with the UV-sensitivity, skin cancer-prone syndrome xeroderma pigmentosum (XP, for which there are at least seven distinct genetic complementation groups). Of particular interest is the link between NER and transcription, for which the DNA damage in actively transcribing genes, and specifically the transcribed strand, is preferentially and thus more rapidly repaired than the DNA damage in the rest of the genome (Lommel et al., 1995; Jiang and Sancar, 2006). Thus, the cell protects the integrity of the transcription process. This link between transcription and repair appears to be provided by two factors: (1) when a bulky lesion is located on the transcribed strand of an active gene, RNA polymerase II is blocked, thus providing a signal for recruiting the NER complex, and (2) a major component of the NER complex is the TFII H basal transcription factor. The involvement of TFII H in repair also provides some specificity to the incisions in the DNA required to remove the damaged nucleotide. An incision on the 3′ side of the damage is made first by the XPG protein followed by 1 on the 5′ side by the XPF–ERRC1 complex. The lesion is removed in the 27- to 30-nucleotide segment formed by the two incisions. The gap is filled by polymerase δ or ∊ in the presence of replication factor C and proliferating cell nuclear antigen (PCNA). Ligation by DNA ligase I completes the process. This NER process has been reconstituted in vitro, allowing for complete characterization, kinetic studies, and estimates of fidelity (Aboussekhra et al., 1995).

Double-Strand Break Repair

Cell survival is seriously compromised by the presence in the cell of broken chromosomes. Unrepaired double-strand breaks trigger one or more DNA damage response systems to either check cell-cycle progression or induce apoptosis. In order to reduce the probability of persistent DNA double-strand breaks, cells have developed an array of specific repair pathways. These pathways are largely similar across a broad range of species from yeast to humans, although the most frequently used one is different among species. There are two general pathways for repair of DNA double-strand breaks: homologous recombination and nonhomologous end-joining. These two can be considered as being in competition for the double-strand break substrate (Haber, 2000; Sonoda et al., 2006; Mladenov and Iliakis, 2011).

Homologous Recombination

Eukaryotes undergo homologous recombination as part of their normal activities both in germ cells (meiotic recombination) and somatic cells (mitotic recombination) Zheng et al., 2011. The repair of double-strand breaks (and single-strand gaps) basically uses the same process and complex of proteins, although some different protein–protein interactions are involved (Shinohara and Ogawa, 1995). In eukaryotes, the process has been characterized most extensively for yeast, but evidence is accumulating that a very similar process occurs in mammalian cells, including human (Johnson et al., 1999; Cahill et al., 2006). The basic steps in double-strand break repair are as follows. The initial step is the production of a 3′-ended single-stranded tail by exonucleases or helicase activity. Through a process of strand invasion, whereby the single-stranded tail invades an undamaged homologous DNA molecule, together with DNA synthesis, a so-called Holliday junction DNA complex is formed. By cleavage of this junction, two DNA molecules are produced (with or without a structural crossover), neither of which now contain a strand break. Additional models have been proposed but probably play a minor role in mammalian cells (Haber, 2000). A detailed description of the specific enzymes known to be involved can be found in Shinohara and Ogawa (1995), Cahill et al. (2006), and Hiom (2010).

Nonhomologous End-Joining

The characterization of nonhomologous end-joining (NHEJ) in mammalian cells was greatly enhanced by the observation that mammalian cell lines that are hypersensitive to ionizing radiation are also defective in the V(D)J recombination process, which is the means by which the huge range of an antibody's antigen-binding sites and T-cell receptor proteins are generated during mammalian lymphoid cell development. V(D)J recombination requires the production of double-strand breaks, recombination of DNA pieces, and subsequent religation. A major component of the NHEJ repair complex is a DNA-dependent protein kinase (DNA-PK). This protein, a serine/threonine kinase, consists of a catalytic subunit (DNA-PK_cs) and a DNA-end-binding protein consisting of KU70 and KU80 subunits. The specific role of DNA-PK in the repair of double-strand breaks is unclear in mammalian cells; a detailed discussion of what is known and some possible models of NHEJ are presented in the reviews by Critchlow and Jackson (1998) and Burma et al. (2006). Perhaps the most viable role of DNA-PK is to align the broken DNA ends to facilitate their ligation. It also appears that DNA-PK helps in the selection of the specific repair pathway that is ultimately used for repair (Neal and Meek, 2011). In addition, DNA-PK might serve as a signal molecule for recruiting other repair proteins known to be involved in yeast and to some extent in mammalian cells. The final ligation step is performed by DNA ligase IV in human cells.

Mismatch Repair

The study of DNA mismatch repair systems has received considerable attention over the past few years, in part, because an association has been demonstrated between genetic defects in mismatch repair genes and the genomic instability associated with cancer susceptibility syndromes and sporadic cancers. In general, DNA mismatch repair systems operate to repair mismatched bases formed during DNA replication, genetic recombination, and as a result of DNA damage induced by chemical and physical agents. Detailed reviews can be found in Kolodner (1995), Jiricny (1998), Modrich and Lahue (1996), and Jun et al. (2006).

The principal steps in all cells from prokaryotes to human are damage recognition by a specific protein that binds to the mismatch, stabilizing of the binding by the addition of one or more proteins, cutting the DNA at a distance from the mismatch, excision past the mismatch, resynthesis, and ligation. In some prokaryotes, the cutting of the DNA (for DNA replication mismatches) is directed to the strand that contains the incorrect base by using the fact that recently replicated DNA is unmethylated at N⁶-methyladenine at a GATC sequence. The question of whether or not strand-specific mismatch repair occurs in mammalian cells has not been resolved, although evidence does point to its occurrence (Modrich, 1997; Mastrocola and Heinen, 2010). Strand specificity for DNA mismatches resulting from induced DNA damage has not been identified.

O6-Methylguanine-DNA Methyltransferase Repair

The main role for O⁶-methylguanine-DNA methyltransferase (MGMT) is to protect cells against the toxic effects of simple alkylating agents. The methyl group is transferred from O⁶-methylguanine in DNA to a cysteine residue in MGMT. The adducted base is reverted to a normal one by the enzyme, which is itself inactivated by the reaction. Details of the MGMT enzyme properties and the gene isolation and characterization can be found in Tano et al. (1990), Grombacher et al. (1996), and Margison et al. (2003).

The probability that induced DNA damage can be converted into a genetic alteration is influenced by the particular repair pathways recruited, the rate of repair of the damage, and the fidelity and completeness of the repair. The mechanisms of induction of gene mutations and chromosome alterations discussed in the following sections build upon the assessment of the probability of repair versus misrepair versus nonrepair that can be derived from a knowledge of the mechanism of action of the different DNA repair mechanisms. The preceding sections, together with the references provided, should assist in this assessment.

Formation of Gene Mutations

Somatic Cells

Gene mutations are considered to be small DNA sequence changes confined to a single gene; larger genomic changes are considered below, under the section “Formation of Chromosomal Alterations.” The general classes of gene mutations are base substitutions and small additions or deletions. More detailed classifications can be found in the review by Ripley (1991). Base substitutions are the replacement of the correct nucleotide by an incorrect one; they can be further subdivided as transitions where the change is purine for purine or pyrimidine for pyrimidine, and transversions where the change is purine for pyrimidine and vice versa. Frameshift mutations are strictly the addition or deletion of one or a few base pairs (not in multiples of three) in protein-coding regions. The definition is more generally expanded to include such additions and deletions in any DNA region. For the discussion of the mechanism of induction of gene mutations and chromosomal alterations, it is necessary to distinguish chemicals by their general mode of action. Chemicals that can produce genetic alterations with similar effectiveness in all stages of the cell cycle are called radiomimetic because they act like radiation in this regard. Chemicals that produce genetic alterations far more effectively in the S phase of the cell cycle are described as nonradiomimetic. The great majority of chemicals are nonradiomimetic; the radiomimetic group includes bleomycin, streptonigrin, neocarzinostatin, and 8-ethoxycaffeine.

Gene mutations can arise in the absence of specific exogenous exposures to radiation and chemicals. The great majority of so-called spontaneous (background) mutations arise from replication of an altered template. The DNA alterations that arise are either the result of oxidative damage or produced from the deamination of 5-methylcytosine to thymine at CpG sites resulting in G:C → A:T transitions. Mutations induced by ionizing radiations tend to be deletions ranging in size from a few bases to multilocus events (Thacker, 1992). The rapid rate of repair of the majority of radiation-induced DNA damages greatly reduces the probability of DNA lesions being present at the time of DNA replication. Thus, mutations induced by ionizing radiations are generally the result of errors of DNA repair (Preston, 1992). The low frequencies of gene mutations are produced from any unrepaired DNA base damage present during DNA replication.

Gene mutations produced by a majority of chemicals and nonionizing radiations are base substitutions, frameshifts, and small deletions. Of these mutations, a very high proportion is produced by errors of DNA replication on a damaged template. Thus, the probability of a DNA adduct, for example, being converted into a mutation is determined, to a significant extent by the number of induced DNA adducts that remain in the DNA at the time that it is replicated and by the nature of the adduct itself (Jarabek et al., 2009). Thus, relative mutation frequency will be the outcome of the race between repair and replication, that is, in general terms, the more repair that takes place prior to replication, the lower the mutation frequency for a given amount of induced DNA damage. Significant regulators of the race are cell cycle checkpoint genes (eg, P53) because if the cell is checked from entering the S phase at a G₁/S checkpoint, more repair can take place prior to the cell starting to replicate its DNA (Mercer, 1998; Giglia-Mari et al., 2011).

The proportion of chemically induced gene mutations that result from DNA repair errors is low, given that the DNA repair processes, unlike translesion synthesis, are typically error-free. Moreover, repair of chemically induced DNA damage is generally slower than for ionizing radiation damage, for which the balance tips toward replication prior to repair, especially for cells in the S phase at the time of exposure. In the case of translesion bypass, discussed above, gene mutations can be produced at relatively high frequencies.

Germ Cells

The mechanism of production of gene mutations in germ cells is basically the same as in somatic cells. Ionizing radiations produce mainly deletions via errors of DNA repair; the majority of chemicals induce base substitutions, frameshifts, and small deletions by errors of DNA replication (Favor, 1999).

An important consideration for assessing gene mutations induced by chemicals in germ cells is the relationship between exposure and the timing of DNA replication. Fig. 9-3 depicts the stages in oogenesis and spermatogenesis. A few features are worthy of note. The spermatogonial stem cell in humans and rodents has a long cell cycle time, eight days or longer, with only a small fraction being occupied by the S phase. Thus, the probability of DNA repair taking place prior to DNA replication is high, for both acute and chronic treatments. However, for considerations of genetic risk, it is the spermatogonial stem cell that is the major contributor because it is present, in general, throughout the reproductive lifetime of an individual. Each time a spermatogonial stem cell divides it produces a differentiating spermatogonium and a stem cell. Thus, the stem cell can accumulate genetic damage from chronic exposures. Differentiating spermatogonia, as far as the induction of gene mutation is concerned, are the same as mitotically dividing somatic cells.

The first S phase after gametogenesis occurs in the zygote, formed by fertilization. This fact needs to be balanced by the lack of DNA repair in late spermatids and sperm. Thus, DNA damage induced in these stages will remain until the zygote. Postmeiotic germ cells are particularly sensitive to mutation induction by nonradiomimetic chemicals, especially following acute exposures (Russell, 2004). The fairly short duration of this stage (approximately 21 days in the mouse) means that their contribution to genetic risk following chronic exposures is quite small.

For oogenesis (Fig. 9-3) similar observations on gene mutation induction and timing of S phase can be made. In this case the primary oocyte arrests prior to birth, and there is no further S phase until the zygote. For this reason, the oocyte is resistant to the induction of gene mutations by nonradiomimetic chemicals but not to radiation, for which DNA repair is the mode of formation of mutations, and DNA repair occurs in oocytes (Brewen and Preston, 1982).

These mechanistic aspects of the production of gene mutations (and chromosome alterations described in the following two sections) by chemicals and radiation in somatic and germ cells are most important for considerations of the design of genetic toxicology assays, the interpretation of the data generated, and the incorporation of the data into cancer and genetic risk assessments.

Formation of Chromosomal Alterations

Somatic Cells

Structural Chromosome Aberrations

There are components of the formation of chromosome aberrations, SCEs (the apparently reciprocal exchange between the sister chromatids of a single chromosome), and gene mutations that are similar. In particular, damaged DNA serves as the substrate leading to all these events. However, chromosome aberrations induced by ionizing radiations are generally formed by errors of DNA repair, whereas those produced by nonradiomimetic chemicals are generally formed by errors of DNA replication on a damaged DNA template.

The DNA repair errors that lead to the formation of chromosome aberrations following ionizing radiation (and radiomimetic chemical) exposure arise from misligation of double-strand breaks or interaction of coincidentally repairing regions during NER of damaged bases. The details of the DNA damage types and their repair are described in the previous section. Thus, the overall kinetics and fidelity of DNA repair influence the sensitivity of cells to the induction of chromosomal aberrations produced by misrepair. The broad outcomes of misrepair are that incorrect rejoining of chromosomal pieces during repair leads to chromosomal exchanges within (eg, inversions and interstitial deletions) and between (eg, dicentrics and reciprocal translocations) chromosomes. In fact, using FISH, it can be shown that very complex rearrangements take place (Anderson et al., 2000). Failure to rejoin double-strand breaks or to complete repair of other types of DNA damage leads to terminal deletions.

Acentric fragments arise from interstitial deletions, terminal deletions, and the formation of dicentric chromosomes and rings. The failure to incorporate an acentric fragment into a daughter nucleus at anaphase/telophase, or the failure of a whole chromosome to segregate at anaphase to the cellular poles, can result in the formation of a membrane-bound micronucleus that resides in the cytoplasm.

Errors of DNA replication on a damaged template can lead to a variety of chromosomal alterations. The majority of these involve deletion or exchange of individual chromatids (chromatid-type aberrations). Thus, nonradiomimetic chemicals induce only chromatid-type aberrations, whereas radiations and radiomimetic chemicals induce chromatid-type aberrations in the S and G₂ phases of the cell cycle, but chromosome-type aberrations affecting both chromatids in G₁. The reason for this latter observation is that the G₁ (or G₀) chromosome behaves as a single DNA molecule and aberrations formed in it will be replicated in the S phase and will involve both chromatids. This distinction is important for considerations of outcome of the aberrations and the probability of an effect on cells because for chromatid-type aberrations, one chromatid remains intact and genetically unaltered, in contrast to chromosome-type aberrations in which both chromatids are damaged (Preston et al., 1995).

Numerical Chromosome Changes

Numerical changes (eg, monosomies, trisomies, and ploidy changes) can arise from errors in chromosomal segregation. The complexity of the control and the mechanics of the mitotic process means that alteration of various cellular components can result in failure to segregate the sister chromatids to separate daughter cells or in failure to segregate a chromosome to either pole (Bickel and Orr-Weaver, 1996; Preston, 1996; Hunt, 2006). The mechanisms underlying chromosomal loss are pertinent to those involved in the formation of micronuclei.

A limited set of chemicals has been demonstrated to cause aneuploidy through interaction with components of the structures that facilitate chromosome movement (Preston, 1996; Aardema et al., 1998). These include benomyl, griseofulvin, nocodazole, colchicine, colecemid, vinblastine, and paclitaxel. These chemicals affect tubulin polymerization or spindle microtubule stability. To date, other mechanisms of aneuploidy induction by chemicals have not been firmly identified.

Sister Chromatid Exchanges

SCEs are produced during the S phase and are presumed to be a consequence of errors in the replication process, perhaps at the sites of stalled replication complexes (Painter, 1980; Heartlein et al., 1983; Preston, 1991; Wilson and Thompson, 2007). Because SCEs are apparently reciprocal exchanges, it is quite possible that they result from a recombination process occurring at the site of the stalled replication fork. It is, in fact, this mode of action that makes assays for SCE less than ideal for detecting effects due directly to a chemical exposure. For example, the creation of intracellular conditions that slow the progress of DNA replication could lead to the formation of SCE.

Germ Cells

The formation of chromosomal alterations in germ cells is basically the same as that for somatic cells, namely, via misrepair for ionizing radiations and radiomimetic chemicals for treatments in G₁ and G₂, and by errors of replication for all radiations and chemicals for DNA damage present during the S phase. Also, the restrictions on the timing of formation of chromosomal alterations induced by nonradiomimetic chemicals in germ cells is as described above for gene mutations, namely at the specific stages where DNA synthesis occurs, as depicted in Fig. 9-3.

The types of aberrations formed in germ cells are the same as those formed in somatic cells (eg, deletions, inversions, translocations), although their appearance in diplotene/diakinesis of meiosis I, where analysis is frequently conducted, is rather different because of the homologous chromosome pairing that takes place in meiotic cells (see the review by Léonard, 1973). The specific segregation of chromosomes during meiosis influences the probability of recovery of an aberration, particularly a reciprocal translocation, in the offspring of a treated parent. This is discussed in detail in Preston et al. (1995).

Introduction to Assay Design

Genetic toxicology assays are used to identify germ cell mutagens, somatic cell mutagens, and potential carcinogens. These assays can detect diverse kinds of genetic alterations (eg, gene mutations, chromosome aberrations, and aneuploidy) that are relevant to the production of adverse human health outcomes. Over the last three decades, hundreds of chemicals and complex mixtures have been evaluated for genotoxic effects. Genetic toxicology assays serve two interrelated but distinct purposes in the toxicologic evaluation of chemicals: (1) identifying mutagens for purposes of hazard identification and (2) characterizing dose–response relationships and mutagenic mechanisms, both of which contribute to an understanding of genetic and carcinogenic risks.

A common experience when surveying the genetic toxicology literature is encountering a bewildering array of assays in viruses, bacteria, fungi, cultured mammalian cells, plants, insects, and mammals. More than 200 assays for mutagens have been proposed, and useful information has been obtained from many of them. Although most genetic toxicology testing and evaluation relies on relatively few assays, data from relatively obscure assays can sometimes contribute to a judgment about the genetic activity of a compound.

Table 9-1 lists key assays that have a prominent place in genetic toxicology. Table 9-2 is a more comprehensive list that provides literature citations to many of the assays that one might encounter in the genetic toxicology literature. Even this extensive table is not exhaustive, in that it emphasizes methods in applied genetic toxicology and not those assays whose use has been largely restricted to studies of mutational mechanisms. The commonly used assays rely on phenotypic effects as indicators of gene mutations or small deletions and on cytological methods for observing gross chromosomal damage. Detailed information on assay design, testing data, controls, sample sizes, and other factors in effective testing is found in the references cited.

Some assays for gene mutations detect forward mutations whereas others detect reversion. Forward mutations, such as those detected in the thymidine kinase gene (tk) in the widely used assay in mouse lymphoma cells (Clements, 2000; Wang et al., 2009), are genetic alterations in a wild type gene that are detected by a change in phenotype caused by the alteration or loss of gene function. In contrast, back mutations are mutations that restore gene function in a mutant, such as the histidine revertants detected in the Ames assay in Salmonella (Ames et al., 1975; Mortelmans and Zeiger, 2000; Claxton et al., 2010). Thus, a back mutation or reversion that restores gene function in a mutant brings about a return to the wild type phenotype. In principle, forward-mutation assays should respond to a broad spectrum of mutagens because any mutation that interferes with gene expression should confer the detectable phenotype. A reversion assay might be expected to have a more restricted mutational response because only mutations that correct or compensate for the specific mutation in a particular mutant will be detected. In fact, some reversion assays respond to a broader spectrum of mutational changes than one might expect because mutations at a site other than that of the original mutation, either within the test gene or in a different gene (ie, a suppressor mutation), can sometimes confer the selected phenotype. Both forward mutation assays and reversion assays are used extensively in genetic toxicology.

The simplest gene mutation assays rely on selection techniques to detect mutations. A selection technique is a means of imposing experimental conditions under which only cells or organisms that have undergone mutation can grow. For example, only cells that have a mutation in the tk gene can grow in medium containing the inhibitory chemical trifluorothymidine (Seifried et al., 2006; Wang et al., 2009). Selection techniques greatly facilitate the identification of rare cells that have experienced mutation among the many cells that have not. Forward mutations (Clements, 2000; Vlasakova et al., 2005) and reversions (Josephy, 2000; Mortelmans and Riccio, 2000; Mortelmans and Zeiger, 2000; Kamber et al., 2009) can both be detected by selection techniques in microorganisms and cultured mammalian cells. Because of their speed, low cost, and ease of detecting events that occur at low frequency (ie, mutation), assays in microorganisms and cell cultures have figured prominently in genetic toxicology.

Studying mutagenesis in intact animals requires assays of more complex design than the simple selection methods used in microorganisms and cultured cells. Genetic toxicology assays therefore range from inexpensive short-term tests (Zeiger, 2010) that can be performed in a few days to complicated assays for mutations in mammalian germ cells (Favor, 1999; Russell, 2004; Singer et al., 2006). Even in multicellular organisms, there has been an emphasis on designing assays that detect mutations with great efficiency (Vogel et al., 1999; Casciano et al., 1999; Lambert et al., 2005; Dobrovolsky et al., 2010). Nevertheless, there remains a gradation in which an increase in relevance for human risk entails more elaborate and costly tests (Table 9-2). The most expensive mammalian tests are typically reserved for agents of special importance in basic research or risk assessment, whereas the simpler assays can be applied more broadly.

Cytogenetic assays differ in design from typical gene mutation assays because of their reliance on cytological rather than genetic methods. The goal in cytogenetic methods is the unequivocal visual recognition of cells that have experienced genetic damage. The alterations measured include chromosome aberrations (Preston et al., 1981; Ishidate et al., 1988; Corvi et al., 2008), micronuclei (Hayashi et al., 2007; Corvi et al., 2008; Fenech et al., 2011a,b), SCEs (Tucker et al., 1993a; Wilson and Thompson, 2007), and changes in chromosome number (Aardema et al., 1998; Paccierotti and Sgura, 2008). The latter include ploidy changes (eg, polyploid cells) and aneuploidy, in which one or a few chromosomes have been gained (ie, hyperploidy) or lost (ie, hypoploidy) relative to the normal chromosome number. Aneuploidy is of great interest because of its implications for human health, but assays for aneuploidy are not yet as refined or systematically applied as those for other classes of chromosomal alterations.

In all mutagenicity testing, one must be aware of possible sources of error. Factors to consider in the application of mutagenicity assays are the choice of suitable organisms and growth conditions, appropriate monitoring of genotypes and phenotypes, effective experimental design and treatment conditions, inclusion of proper positive and negative controls, and sound methods of data analysis. The articles cited in Table 9-2 discuss these aspects of genetic toxicology testing.

Many compounds that are not themselves mutagenic or carcinogenic can be activated into mutagens and carcinogens by metabolism (Guenngerich, 2001; Malling, 2004). Such compounds are called promutagens and procarcinogens. Because microorganisms and mammalian cell cultures lack many of the metabolic capabilities of intact mammals, provision must be made for metabolic activation in order to detect promutagens in many genetic assays. Incorporating an in vitro metabolic activation system derived from a mammalian tissue homogenate is the most common means of adding metabolic activation to microbial or cell culture assays (Malling and Frantz, 1973; Ames et al., 1975; Clements, 2000; Malling, 2004). For example, the promutagens dimethylnitrosamine and benzo[a]pyrene are not themselves mutagenic in bacteria, but they are mutagenic in bacterial assays if the bacteria are treated with the promutagen in the presence of a homogenate from mammalian liver.

The most widely used metabolic activation system in microbial and cell culture assays is a postmitochondrial supernatant from a rat liver homogenate, along with appropriate buffers and cofactors (Maron and Ames, 1983; Kirkland et al., 1990). The standard liver metabolic activation system is called an S9 mixture, designating a supernatant from centrifugation at 9000g (Malling and Frantz, 1973; Maron and Ames, 1983). Most of the short-term assays in Table 9-2 require exogenous metabolic activation to detect promutagens. Exceptions are assays in intact mammals and a few simpler assays that have a high level of endogenous cytochrome P450 metabolism, such as the detection of UDS or DNA strand breakage in cultured hepatocytes (Madle et al., 1994; Gealy et al., 2007).

Rat liver S9 provides a broad assemblage of metabolic reactions, but they are not necessarily the same as those of hepatic metabolism in an intact rat. Metabolic activation systems based on homogenates from mice, guinea pigs, hamsters, or monkeys and preparations from organs other than liver have found some use in mutagenicity testing (Mortelmans and Zeiger, 2000). In some cases, these systems detect mutagenicity more efficiently than rat liver S9. However, like a homogenate from rat liver, these systems may differ from the species or organs of their origin. Therefore, alternative metabolic activation systems tend to be more useful if chosen for mechanistic reasons rather than simply testing another species or organ. Such systems include the use of intact hepatocytes (Madle et al., 1994; Storer et al., 1996; Gealy et al., 2007) to preserve the cellular compartmentalization of reactions; an in vitro system that can carry out the reduction reactions needed to detect the activity of substances whose mutagenicity requires reductive metabolism, such as some azo compounds (Mortelmans and Zeiger, 2000; Seifried et al., 2006); and the use of mammalian cells or bacteria engineered to express foreign genes that encode enzymes of metabolic activation (Sawada and Kamataki, 1998; Crespi and Miller, 1999; Josephy, 2000, 2002; Oda et al., 2009).

Besides metabolic activation, some chemicals are subject to photochemical activation. The genotoxicity of such chemicals depends on the chemical being irradiated with ultraviolet or visible light. Many of the assays listed in Table 9-2, including gene-mutation assays in bacteria and cultured mammalian cells, cytogenetic assays, and the comet assay, have been adapted so that they can measure photogenotoxic effects (Brendler-Schwaab et al., 2004; Lynch et al., 2011).

Genes encoding enzymes of xenobiotic metabolism have been incorporated by recombinant DNA technology into microorganisms or cell cultures to expand their capacity for metabolic activation. The genes incorporated into bacteria may be derived from other bacteria or from humans (Josephy, 2002; Oda et al., 2009). For example, bacterial genes that cause overexpression of N-acetyltranserase enhance the sensitivity of the bacteria to the mutagenicity of aromatic amines or nitroarenes (Josephy, 2002). The expression of human cytochrome P450 enzymes in Salmonella tester strains from the Ames assay (Josephy 2002; Yamazaki et al., 2004), a Salmonella SOS-induction assay (Oda et al., 2009), or E coli strains of the lacZ reversion assay (Josephy, 2000) permits the activation of such promutagens as 2-aminoanthracene and 2-aminofluorene without an S9 mixture. Mammalian cell lines have also been genetically engineered to express human Phase-I and Phase-II enzymes, including those catalyzing reactions of metabolic activation (Sawada and Kamataki, 1998). Many cell lines stably expressing a single form of P450 have been established. Mutagenesis can be measured through such endpoints as HPRT mutations and cytogenetic alterations, and the cells are well suited to analyzing the contribution of different enzymes to the activation of promutagens (Crespi and Miller, 1999).

Metabolic activation is so central to genetic toxicology that all mutagenicity testing programs must provide for it in the choice of assays and procedures. Despite their usefulness, in vitro metabolic activation systems, however well refined, cannot mimic mammalian metabolism perfectly. There are differences among tissues in reactions that activate or inactivate foreign compounds, and organisms of the normal flora of the gut can contribute to metabolism in intact mammals. Agents that induce enzyme systems or otherwise alter the physiological state can also modify the metabolism of toxicants, and the balance between activation and detoxication reactions in vitro may differ from that in vivo.

Structural Alerts and In Silico Assays

The first indication that a chemical is a mutagen often lies in chemical structure. Potential electrophilic sites in a molecule serve as an alert to possible mutagenicity and carcinogenicity because such sites confer reactivity with nucleophilic sites in DNA (Tennant and Ashby, 1991). Structural alerts in combination with critical interpretation are a valuable adjunct to mutagenicity testing (Tennant and Ashby, 1991; Ashby and Paton, 1993). Attempts to formalize the structural prediction through automated computer programs have not yet led to an ability to predict mutagenicity and carcinogenicity of new chemicals with great accuracy (Snyder et al., 2004), but promising developmental work on such systems continues (Votano et al., 2004; Snyder and Smith, 2005).

Computer-based systems for predicting genotoxicity based on chemical properties are sometimes called in silico assays. These assays include computational and structural programs (Knudsen et al., 2011; Snyder and Smith, 2005; Snyder, 2009; Mahadevan et al., 2011) and the modeling of quantitative structure–activity relationships (Votano et al., 2004; Snyder and Smith, 2005; Mahadevan et al., 2011). Although there is much skepticism that such approaches can replace biological testing, they hold promise of improving the efficiency of testing strategies and reducing current levels of animal use (Guha, 2011).

DNA Damage and Repair Assays

Some assays measure DNA damage itself, rather than mutational consequences of DNA damage. They may do so directly, through such indicators as chemical adducts or strand breaks in DNA, or indirectly, through the measurement of biological repair processes. Adducts in DNA are detected by ³²P-postlabeling, high-performance liquid chromatography (HPLC), fluorescence-based methods, mass spectrometry, immunological methods using antibodies against specific adducts, isotope-labeled DNA binding, and electrochemical detection (Phillips et al., 2000; Farmer and Singh, 2008; Himmelstein et al., 2009). The ³²P-postlabeling method is highly versatile, in that it is sensitive and can be applied to diverse mutagens, but it may fall short of other methods for quantitative accuracy (Farmer and Singh, 2008). Through a combination of methods, many classes of adducts, including those of such environmentally widespread compounds as polynuclear aromatic hydrocarbons, can be detected. The measurement of adducts after human chemical exposures has proven useful in human monitoring, molecular dosimetry, and risk assessment Phillips et al., 2000; Farmer and Singh, 2008; Himmelstein et al., 2009). Adducts in somatic cells are relevant to carcinogenesis (Himmelstein et al., 2009), whereas those in reproductive cells permit molecular dosimetry for germ cell mutagenesis (Olsen et al., 2010; 2011; Verhofsrad et al., 2011).

DNA strand breakage can be measured by alkaline elution (Storer et al., 1996; Gealy et al., 2007) and electrophoretic methods. Single-cell gel electrophoresis, also called the comet assay, is a widely used, rapid method of measuring DNA damage (Fairbairn et al., 1995; Singh, 2000; Tice et al., 2000; Collins, 2004; Olive, 2009). In this assay cells are incorporated into agarose on slides, lysed so as to liberate their DNA, and subjected to electrophoresis. The DNA is stained with a fluorescent dye for observation and image analysis. Because broken DNA fragments migrate more quickly than larger pieces of DNA, a blur of fragments (a “comet”) is observed when the DNA is extensively damaged. The extent of DNA damage can be estimated from the length and other attributes of the comet tail (Collins, 2004). Variations in the procedure permit the general detection of DNA strand breakage under alkaline conditions (Fairbairn et al., 1995; Singh, 2000; Tice et al., 2000) or the preferential detection of double-strand breaks under neutral conditions (Fairbairn et al., 1995; Olive, 2009). A recent development is the combination of the comet assay with FISH to detect damage in specific regions of the genome (Glei et al., 2009; Shaposhnikov et al., 2009). Although the comet assay is relatively new, it has proven to be a sensitive indicator of DNA damage with broad applicability. It has been used most commonly with human lymphocytes (Fairbairn et al., 1995; Singh, 2000; Collins, 2004) and other mammalian cells (Tice et al., 2000), but it can be adapted to diverse species, including plants, worms, mollusks, fish, and amphibians (Cotelle and Férard, 1999; Lee and Steinert, 2003; Jha, 2004). This adaptability makes it well suited to use in environmental genetic toxicology. The applicability of the comet assay and other DNA damage assays to rodent testes (Bentley et al., 1994; Speit et al., 2009) makes these methods helpful in interpreting risks to germ cells.

The occurrence of DNA repair can serve as an easily measured indicator of DNA damage. Repair assays have been developed in microorganisms, cultured mammalian cells, and intact mammals (Table 9-2). Greater toxicity of a chemical in DNA-repair-deficient strains than in their repair-proficient counterparts (eg, rec⁺ and rec⁻ in Bacillus subtilis) can serve as an indicator of DNA damage in bacteria (Takigami et al., 2002). Bacterial repair assays find occasional application but are used less commonly today than historically. The measurement of UDS, which is a measure of excision repair, is a mammalian DNA repair assay. The occurrence of UDS indicates that DNA has been damaged (Madle et al., 1994). The absence of UDS, however, does not provide evidence that DNA has not been damaged because some classes of damage are not readily excised, and some excisable damage may not be detected as a consequence of assay insensitivity (Kirkland and Speit, 2008). Despite these limitations, UDS assays continue to find some use because of their applicability to cultured hepatocytes with endogenous cytochrome P450 enzyme activities and to tissues of intact animals, including hepatocytes (Madle et al., 1994) and germinal tissue (Bentley et al., 1994; Sotomayor and Sega, 2000).

Besides specific DNA repair processes, the induction of general responses to genotoxic stress has been used as an indicator of genetic damage. The induction of SOS functions, indicated by phage induction or by colorimetry, can serve as an indicator of genetic damage in bacteria (Quillardet and Hofnung, 1993; Yasunaga et al., 2004; Oda et al., 2009). The GADD45a-GFP assay, also called “Green Screen,” detects genotoxic stress in human lymphoblastoid TK6 cells (Hastwell et al., 2009). The stress response is detected by a green-fluorescent protein reporter in the genetic construct, and the induction of fluorescence has been observed with mutagens, clastogens, and aneugens (Hastwell et al., 2009). The assay can be conducted with S9 metabolic activation and lends itself to automated detection by flow cytometry (Jagger et al., 2009).

Gene Mutations in Prokaryotes

The most common means of detecting mutations in microorganisms is selecting for reversion in strains that have a specific nutritional requirement differing from wild type members of the species; such strains are called auxotrophs. For example, the widely used assay developed by Bruce Ames and his colleagues is based on measuring reversion in several histidine auxotrophs in Salmonella enterica serovar Typhimurium, commonly called S typhimurium (Ames et al., 1975).

In the Ames assay one measures the frequency of histidine-independent bacteria that arise in a histidine-requiring strain in the presence or absence of the chemical being tested. Auxotrophic bacteria are treated with the chemical of interest by one of several procedures (eg, the standard plate-incorporation assay) and plated on medium that is deficient in histidine (Ames et al., 1975; Maron and Ames, 1983; Mortelmans and Zeiger, 2000). The assay is conducted using genetically different strains so that reversion by base pair substitutions and frameshift mutations in several DNA sequence contexts can be detected and distinguished. Because Salmonella does not metabolize promutagens in the same way as mammalian tissues, the assay is generally performed in the presence and absence of a rat liver S9 metabolic activation system. Hence, the Ames assay is also called the Salmonella/microsome assay.

The principal strains of the Ames test and their characteristics are summarized in Table 9-3. In addition to the histidine alleles that provide the target for measuring mutagenesis, the Ames tester strains contain other genes and plasmids that enhance the assay. Part I of the table gives genotypes, and Part II explains the rationale for including specific genetic characteristics in the strains. Part III summarizes the principal DNA target in each strain and the mechanisms of reversion. Taken together, the Ames strains detect a broad array of mutations, and they complement one another. For example, strains TA102 and TA104, which are sensitive to agents that cause oxidative damage in DNA, detect the A:T → G:C base pair substitutions that are not detected by hisG46 strains (Mortelmans and Zeiger, 2000). TA102 also detects agents that cause DNA cross-links because it has an intact excision repair system whereas the other common tester strains do not.

The most common version of the Ames assay is the plate-incorporation test (Ames et al., 1975; Maron and Ames, 1983; Mortelmans and Zeiger, 2000). In this procedure, the bacterial tester strain, the test compound (or solvent control), and the S9 metabolic activation system (or buffer for samples without S9) are added to 2 mL of molten agar containing biotin and a trace of histidine to allow a few cell divisions, mixed, and immediately poured onto the surface of a petri dish selective for histidine-independent revertants. For general testing it is recommended that at least three plates per dose and five doses be used with and without S9, along with appropriate concurrent positive and negative controls (Mortelmans and Zeiger, 2000). Variations on the standard plate-incorporation assay confer advantages for some applications. These include a preincubation assay that facilitates the detection of unstable compounds and short-lived metabolites, a desiccator assay for testing volatile chemicals and gases, a microsuspension assay for working with small quantities of test agent, assays incorporating reductive metabolism rather than the conventional S9 system, and assays under hypoxic conditions (Mortelmans and Zeiger, 2000).

Although simplicity is a great merit of microbial assays, it can also be deceptive. Even assays that are simple in design and application can be performed incorrectly. For example, in the Ames assay one may see very small colonies in the petri dishes at highly toxic doses (Maron and Ames, 1983; Kirkland et al., 1990; Mortelmans and Zeiger, 2000). Counting such colonies as revertants would be an error because they may actually be nonrevertant survivors that grew on the low concentration of histidine in the plates. Were there millions of survivors, the amount of histidine would have been insufficient to allow any of them (except real revertants) to form colonies. This artifact is easily avoided by checking that there is a faint lawn of bacterial growth in the plates; one can also confirm that colonies are revertants by streaking them on medium without histidine to be sure that they grow in its absence. Such pitfalls exist in all mutagenicity tests. Therefore, anyone performing mutagenicity tests must have detailed familiarity with the laboratory application and literature of the assay and be observant about the responsiveness of the assay.

Although information from the Ames assay has become a standard in genetic toxicology testing, equivalent information can be obtained from other bacterial assays. Like the Ames assay, the WP2 tryptophan reversion assay in E coli (Kirkland et al., 1990; Mortelmans and Riccio, 2000) incorporates genetic features that enhance assay sensitivity, can accommodate S9 metabolic activation, and performs well in many laboratories. Mutations are detected by selecting for reversion of a trpE allele from Trp⁻ to Trp⁺. Its responsiveness to mutagens most closely resembles TA102 among the Ames strains (Mortelmans and Riccio, 2000).

Bacterial reversion assays are commonly used for testing purposes, but they also provide information on molecular mechanisms of mutagenesis. The broader understanding of mutational mechanisms that comes from refined genetic assays and molecular analysis of mutations can contribute to the interpretation of mutational hazards. The primary reversion mechanisms of the Ames strains, summarized in Table 9-3, were initially determined by genetic and biochemical means (Maron and Ames, 1983). An ingenious method called allele-specific colony hybridization greatly facilitated the molecular analysis of revertants in the Ames assay (Koch et al., 1994), and many spontaneous and induced revertants have been cloned or amplified by the polymerase chain reaction (PCR) and sequenced (Levine et al., 1994; DeMarini, 2000).

Part III of Table 9-3 is by necessity a simplification with respect both to targets and mechanisms of reversion of the Ames strains. Some mutations that bring about reversion to histidine independence fall outside the primary target, and the full target has been found to be as much as 76 base pairs in hisD3052 (DeMarini et al., 1998). Other revertants can arise by suppressor mutations in other genes. It has been shown that hisG46, hisG428, hisC3076, hisD6610, and hisD3052 all revert by multiple mechanisms and that the spectrum of classes of revertants may vary depending on the mutagen, experimental conditions, and other elements of the genotype (Cebula and Koch, 1990; Prival and Cebula, 1992; DeMarini et al., 1998; Mortelmans and Zeiger, 2000).

The development of Salmonella strains that are highly specific with respect to mechanisms of reversion has made the identification of particular base pair substitutions more straightforward. These strains (TA7001–TA7006) each revert from his⁻ to his⁺ by a single kind of mutation (eg, G:C to T:A), and collectively they permit the specific detection of all six possible base pair substitutions (Gee et al., 1994; Mortelmans and Zeiger, 2000; Kamber et al., 2009). The assay is usually conducted using a fluctuation test in 24-well plates, rather than the plate-incorporation or preincubation method. The ability to discern mutagens and nonmutagens is comparable to the standard Ames assay (Kamber et al., 2009).

Specific reversion assays are also available in E coli. A versatile system based on reversion of lacZ mutations in E coli permits the specific detection of all six possible base pair substitutions (Cupples and Miller, 1989; Josephy, 2000) and frameshift mutations for which one or two bases have been added or deleted in various sequence contexts (Cupples et al., 1990; Josephy, 2000; Hoffmann et al., 2003). The versatility of the lacZ assay has been expanded through the introduction of useful characteristics into the strains parallel to those incorporated into the Ames strains. Among the features added to the lacZ assay are DNA repair deficiencies, permeability alterations, plasmid-enhanced mutagenesis, and enzymes of mutagen metabolism (Josephy, 2000).

Bacterial forward mutation assays, such as selections for resistance to arabinose or to purine or pyrimidine analogs in Salmonella (Jurado et al., 1994; Vlasakova et al., 2005), are also used in research and testing, although less extensively than reversion assays. A versatile forward mutation assay that has contributed greatly to an understanding of mechanisms of mutagenesis is the lacI system in E coli (Calos and Miller, 1981; Halliday and Glickman, 1991). Mutations in the lacI gene, which encodes the repressor of the lactose operon, are easily identified by phenotype, cloned or amplified by PCR, and sequenced. The lacI gene is widely used as a target for mutagenesis both in E coli and in transgenic mice, and thousands of lacI mutants have been sequenced.

Genetic Alterations in Nonmammalian Eukaryotes

Gene Mutations and Chromosome Aberrations

Many early studies of mutagenesis used yeasts, mycelial fungi, plants, and insects as experimental organisms. Even though well-characterized genetic systems permit the analysis of a diverse array of genetic alterations in these organisms (Table 9-2), they have been largely supplanted in genetic toxicology by bacterial and mammalian systems. Exceptions are to be found where the assays in nonmammalian eukaryotes permit the study of genetic endpoints that are not readily analyzed in mammals or where the organism has special attributes that fit a particular application.

The fruit fly, Drosophila, has long occupied a prominent place in genetic research. In fact, the first unequivocal evidence of chemical mutagenesis was obtained in Scotland in 1941 when Charlotte Auerbach and J.M. Robson demonstrated that mustard gas is mutagenic in Drosophila. Drosophila continues to be used in modern mutation research (Potter et al., 2000) but its role in genetic toxicology is now more limited. The Drosophila assay of greatest historical importance is the sex-linked recessive lethal (SLRL) test. A strength of the SLRL test is that it permits the detection of recessive lethal mutations at 600 to 800 different loci on the X chromosome by screening for the presence or absence of wild type males in the offspring of specifically designed crosses (Mason et al., 1987; Vogel et al., 1999). The genetic alterations include gene mutations and small deletions. The spontaneous frequency of SLRL is about 0.2%, and a significant increase over this frequency in the lineages derived from treated males indicates mutagenesis. Although it requires screening large numbers of fruit fly vials, the SLRL test yields information about mutagenesis in germ cells, which is lacking in all microbial and cell culture systems. However, means of exposure, measurement of doses, metabolism, and gametogenesis in insects differ from those in mammalian toxicology, thereby introducing doubt about the relevance of Drosophila assays to human genetic risk. Drosophila assays are also available for studying the induction of chromosome abnormalities in germ cells, specifically heritable translocations (Mason et al., 1987; Vogel et al., 1999).

Genetic and cytogenetic assays in plants (Ma et al., 2005; Grant and Owens, 2006; Misík et al., 2011) also occupy a more restricted niche in modern genetic toxicology than they did years ago. However, plant assays continue to find use in special applications, such as in situ monitoring for mutagens and exploration of the metabolism of promutagens by agricultural plants. In situ monitoring entails looking for evidence of mutagenesis in organisms that are grown in the environment of interest. Natural plant populations can also be examined for evidence of genetic damage, but doing so requires utmost precaution when characterizing the environments and defining appropriate control populations.

Mitotic Recombination

Assays in nonmammalian eukaryotes continue to be important in the study of induced recombination. Recombinogenic effects in yeast have long been used as a general indicator of genetic damage (Zimmermann et al., 1984), and interest in the induction of recombination has increased as recombinational events have been implicated in the etiology of cancer (Sengstag, 1994; Reliene et al., 2007). LOH is central to the expression of the mutant alleles of tumor-suppressor genes, and mitotic recombination is a major mechanism of LOH. Widely used assays for recombinogens detect mitotic crossing over and mitotic gene conversion in the yeast Saccharomyces cerevisiae (Zimmermann, 1992), and hundreds of chemicals have been tested for recombinogenic effects in straightforward yeast assays (Zimmermann et al., 1984). In yeast strain D7, for example, mitotic crossing over involving the ade2 locus is detected on the basis of pink and red colony color, mitotic gene conversion at the trp5 locus by selection for growth without tryptophan, and gene mutations by reversion of the ilv1-92 allele (Zimmermann, 1992; Freeman and Hoffmann, 2007). Newer yeast assays have been constructed to discern whether LOH has occurred by mitotic recombination or chromosome loss (Daigaku et al., 2004; Nunoshiba et al., 2007). Strategies have also been devised to detect recombinogenic effects in human lymphocytes (Turner et al., 2003), other mammalian cells, mice, plants, and mycelial fungi (Hoffmann, 1994; Reliene et al., 2007). At least 350 chemicals have been evaluated in Drosophila somatic cell assays in which recombinogenic effects are detected by examining wings or eyes for regions in which recessive alleles are expressed in heterozygotes (Vogel et al., 1999; Vogel and Nivard, 2000).

Gene Mutations in Mammals

Gene Mutations In Vitro

Mutagenicity assays in cultured mammalian cells have some of the same advantages as microbial assays with respect to speed and cost, and they use similar approaches. The most widely used assays for gene mutations in mammalian cells detect forward mutations that confer resistance to a toxic chemical. For example, mutations in the gene encoding hypoxanthine-guanine phosphoribosyltransferase (HPRT enzyme; HPRT gene) confer resistance to the purine analogue 6-thioguanine (Li et al., 1988; Parry et al., 2005), and thymidine kinase mutations (TK enzyme; TK gene) confer resistance to the pyrimidine analogue trifluorothymidine (Clements, 2000; Wang et al., 2009). HPRT and TK mutations may therefore be detected by attempting to grow cells in the presence of purine analogues and pyrimidine analogues, respectively. For historical reasons, HPRT assays have most commonly been conducted in Chinese hamster cells or human cells, whereas TK assays have used mouse lymphoma cells or human cells. The mouse lymphoma assay, long used for detecting gene mutations, is now also used to detect other endpoints, including recombination, deletion, and aneuploidy (Wang et al., 2009). Forward-mutation assays typically respond to diverse mechanisms of mutagenesis, but there are exceptions such as resistance to ouabain, which only occurs by a specific alteration (DeMarini et al., 1989). Assays that do not detect various kinds of mutations are not useful for general mutagenicity testing. Genetic or molecular evidence that an assay is responsive to diverse mechanisms of mutagenesis is essential.

Instead of using selective media, flow cytometry can be used to detect gene mutations by immunological methods in mammalian cell cultures and intact animals. An in vitro assay for CD59 mutations is performed in CHO-human hybrid A_L cells (Zhou et al., 2006). A_L cells contain a single human chromosome 11 along with the Chinese hamster chromosome complement of the Chinese hamster ovary (CHO) cells. CD59 is a cell-surface protein encoded by a gene on the human chromosome. Fluorescent anti-CD59 antibody is used to quantify CD59 cells by flow cytometry. Besides CD59 itself, mutations in other CD genes on chromosome 11 and in their glycosylphosphatidylinositol (GPI) anchor can be detected in the assay (Ross et al., 2007). An assay for mutations in Pig-a, the GPI anchor gene, is discussed with in vivo assays.

Gene Mutations In Vivo

In vivo assays involve treating intact animals and analyzing appropriate tissues for genetic effects. The choice of suitable doses, treatment procedures, controls, and sample sizes is critical in the conduct of in vivo tests. Mutations may be detected either in somatic cells or in germ cells. The latter are of special interest with respect to risk for future generations.

The mouse spot test is a traditional genetic assay for gene mutations in somatic cells (Styles and Penman, 1985; Lambert et al., 2005). Visible spots of altered phenotype in mice heterozygous for coat-colored genes indicate mutations in the progenitor cells of the altered regions. Although straightforward in design, the spot test is less used today than other somatic cell assays or than its germ cell counterpart, the mouse specific-locus test.

Cells that are amenable to positive selection for mutants when collected from intact animals form the basis for efficient in vivo mutation-detection assays analogous to those in mammalian cell cultures. Lymphocytes with mutations in the HPRT gene are readily detected by selection for resistance to 6-thioguanine. The hprt assay in mice, rats, and monkeys (Casciano et al., 1999) is of special interest because it permits comparisons to the measurement of HPRT mutations in humans in mutational monitoring (Cole and Skopek, 1994; Albertini and Hayes, 1997; Albertini, 2001).

The Pig-a assay is a newer assay that has versions suitable for human monitoring and laboratory studies. The assay detects mutations that block GPI synthesis (Dobrovolsky et al., 2010). Pig-a is a sex-linked gene whose gene product functions as an anchor for cell-surface proteins. Mutations in Pig-a can be detected in red blood cells from rats, mice, monkeys, and humans by means of fluorescent antibodies against GPI-anchored cell-surface proteins, such as CD59. Using antibodies to more than one GPI-linked marker has been suggested as a means of making the assay specific to Pig-a rather than also detecting mutants for a particular cell-surface protein. Frequencies can be measured by clonal growth of Pig-a cells using proaerolysin (ProAER) selection or by flow cytometry (Dobrovolsky et al., 2010; Miura et al., 2011; Dobo et al., 2011). The fact that the same assay can be performed in several species makes this a promising assay for comparisons of human monitoring and controlled exposures in laboratory animals.

Besides determining whether agents are mutagenic, mutation assays provide information on mechanisms of mutagenesis that contributes to an understanding of mutational hazards. Base pair substitutions and large deletions, which may be indistinguishable on the basis of phenotype, can be differentiated through the use of probes for the target gene and Southern blotting, in that base substitutions are too subtle to be detectable on the blots, whereas gross structural alterations are visible (Cole and Skopek, 1994; Albertini and Hayes, 1997). Molecular analysis has been used to determine proportions of mutations ascribable to deletions and other structural alterations in several assays, including the specific-locus test for germ cell mutations in mice (Favor, 1999) and the human HPRT assay (Cole and Skopek, 1994). Gene mutations have been characterized at the molecular level by DNA sequence analysis both in transgenic rodents (Lambert et al., 2005; Singer et al., 2006) and in endogenous mammalian genes (Cariello and Skopek, 1993). Many HPRT mutations from human cells in vitro and in vivo have been analyzed at the molecular level and classified with respect to base pair substitutions, frameshifts, small deletions, large deletions, and other alterations (Cole and Skopek, 1994).

Transgenic Assays

Transgenic animals are products of DNA technology in which the animal contains foreign DNA sequences that have been added to the genome. The foreign DNA is represented in all the somatic cells of the animal and is transmitted in the germ line to progeny. Mutagenicity assays in transgenic animals combine in vivo metabolic activation and pharmacodynamics with simple microbial detection systems, and they permit the analysis of mutations induced in diverse mammalian tissues (Lambert et al., 2005; Sykes et al., 2006; Valentine et al., 2010).

The transgenic animals that have figured most heavily in genetic toxicology are rodents that carry lac genes from E coli. The bacterial genes were introduced into mice or rats by injecting a vector carrying the genes into fertilized oocytes (Lambert et al., 2005). The strains are commonly referred to by their commercial names—the “Big Blue mouse” and “Big Blue rat,” which use lacI as a target for mutagenesis, and the “Muta Mouse,” which uses lacZ (Lambert et al., 2005). After mutagenic treatment of the transgenic animals, the lac genes are recovered from the animal, packaged in phageλ, and transferred to E coli for mutational analysis. Mutant plaques are identified on the basis of phenotype, and mutant frequencies can be calculated for different tissues of the treated animals. The cII locus may be used as a second target gene in both the lacZ and lacI transgenic assays (Swiger, 2001; Lambert et al., 2005). Its use offers technical advantages as a small, easily sequenced target in which mutations are detected by positive selection, and it permits interesting comparisons both within and between assays.

A lacZ transgenic mouse that uses a plasmid-based system rather than a phage vector has the advantage that deletion mutants are more readily recovered than in the phage-based lac systems (Lambert et al., 2005). Deletions may also be detected in the gpt delta mouse and rat using a phage vector system. These transgenic animals detect two kinds of genetic events in two targets—point mutations in gpt detected by resistance to 6-thioguanine and spi deletions that permit growth on P2 lysogens (Okada et al., 1999; Lambert et al., 2005). Other transgenic assays are under development and offer the prospect of expanding the versatility of such assays (Lambert et al., 2005). These include pKZ1 mice in which inversions and deletions arising in various tissues by intrachromosomal recombination have been used to study effects of low doses of ionizing radiation (Sykes et al., 2006) and an assay that detects both forward mutations and reversion in mice that carry the genome of phage ΦX174 (Valentine et al., 2010).

Various mutagens, including alkylating agents, nitrosamines, procarbazine, cyclophosphamide, and polycyclic aromatic hydrocarbons have been studied in transgenic mouse assays, and mutant frequencies have been analyzed in such diverse tissues as liver, skin, spleen, kidney, bladder, small intestine, bone marrow, and testis (Lambert et al., 2005). Mutation frequencies in transgenes in testes have been compared to results in standard germ cell mutagenesis assays (Singer et al., 2006). Female germ cells are less amenable to study in transgenic assays because of the difficulty of collecting sufficient numbers of oocytes, but it has been suggested that granulosa cells from ovarian follicles may serve as a surrogate for the exposure of female germ cells to mutagens (Singer et al., 2006). Mutant frequencies have been compared to the formation of adducts in various tissues and to the site specificity of carcinogenesis and DNA repair capacity (Lambert et al., 2005). An important issue that remains to be resolved is the extent to which transgenes resemble endogenous genes. Although their mutational responses tend to be comparable (Lambert et al., 2005), some differences have been noted (Burkhart and Malling, 1993; Lambert et al., 2005), and questions have been raised about the relevance of mutations that might be recovered from dying or dead animal tissues (Burkhart and Malling, 1994). Therefore, transgenic animals offer promising models for the study of chemical mutagenesis, but they must be further characterized before their ultimate place in hazard assessment is clear.

Mammalian Cytogenetic Assays

Chromosome Aberrations

Cytogenetic assays rely on the use of microscopy for the direct observation of the effects of interest. This approach differs sharply from the indirectness of traditional genetic assays in which one observes a phenotype and reaches conclusions about genes. It is only through the addition of DNA sequencing that genetic assays can approach the directness of cytogenetic assays.

In conventional cytogenetics, metaphase analysis is used to detect chromosomal anomalies, especially unstable chromosome and chromatid aberrations. A key factor in the design of cytogenetic assays is obtaining appropriate cell populations for treatment and analysis (Preston et al., 1981; Ishidate et al., 1988; Kirkland et al., 1990; Galloway et al., 1994). Cells with a stable, well-defined karyotype, short generation time, low chromosome number, and large chromosomes are ideal for cytogenetic analysis. For this reason, Chinese hamster cells have been used widely in cytogenetic testing. Other cells are also suitable, and human cells, especially peripheral lymphocytes, have been used extensively. Cells should be treated during a sensitive period of the cell cycle (typically S), and aberrations should be analyzed at the first mitotic division after treatment so that the sensitivity of the assay is not reduced by unstable aberrations being lost during cell division. Examples of chromosome aberrations are shown in Fig. 9-4.

Cytogenetic assays require careful attention to growth conditions, controls, doses, treatment conditions, and time intervals between treatment and the sampling of cells for analysis (Preston et al., 1981; Ishidate et al., 1988; Kirkland et al., 1990; Galloway et al., 2011). Data collection is a critical part of cytogenetic analysis. It is essential that sufficient cells be analyzed because a negative result in a small sample is inconclusive. Results should be recorded for specific classes of aberrations, not just an overall index of aberrations per cell. The need for detailed data is all the more important because of nonuniformity in the classification of aberrations and disagreement on whether small achromatic (ie, unstained) gaps in chromosomes are true chromosomal aberrations. Gaps should be quantified but not pooled with other aberrations.

In interpreting results on the induction of chromosome aberrations in cell cultures, one must be alert to the possibility of artifacts associated with extreme assay conditions because aberrations induced under such circumstances may not be a reflection of a chemical-specific genotoxicity (Scott et al., 1991; Galloway, 2000; Galloway et al., 2011). Questionable positive results have been found at highly cytotoxic doses (Galloway et al., 2011), high osmolality, and pH extremes (Scott et al., 1991). The possibility that metabolic activation systems may be genotoxic also warrants scrutiny (Scott et al., 1991). Although excessively high doses may lead to artifactual positive responses, the failure to test to a sufficiently high dose also undermines the utility of a test. Therefore, testing should be extended to a dose at which there is some cytotoxicity, such as a reduction in a replicative index or the mitotic index (the proportion of cells in division). If the chemical is nontoxic, testing dosages should extend up to an arbitrary limit of dosage (Galloway et al., 2011). By a consensus of cytogeneticists and genetic toxicologists, a limit of 10 mM or 5 mg/mL, whichever is lower, has been recommended (Galloway et al., 2011). Some have argued that the limit should be lowered, perhaps to 1 mM, but no consensus could be reached on this point (Galloway et al., 2011).

In vivo assays for chromosome aberrations involve treating intact animals and later collecting cells for cytogenetic analysis (Preston et al., 1981; Kirkland et al., 1990; Tice et al., 1994). The main advantage of in vivo assays is that they include mammalian metabolism, DNA repair, and pharmacodynamics. The target is typically a tissue from which large numbers of dividing cells are easily prepared for analysis. Bone marrow from rats, mice, or Chinese hamsters is most commonly used. Peripheral lymphocytes are another suitable target when stimulated to divide with a mitogen such as phytohemagglutinin. Effective testing requires dosages and routes of administration that ensure adequate exposure of the target cells, proper intervals between treatment and collecting cells, and sufficient numbers of animals and cells analyzed (Preston et al., 1981; Kirkland et al., 1990; Tice et al., 1994).

An important development in cytogenetic analysis is FISH, in which a nucleic acid probe is hybridized to complementary sequences in chromosomal DNA (Tucker et al., 1993b, 2005; Paccierotti and Sgura, 2008). The probe is labeled with a fluorescent dye so that the chromosomal location to which it binds is visible by fluorescence microscopy. Composite probes have been developed from sequences unique to specific human chromosomes, giving a uniform fluorescent label over the entire chromosome. Slides prepared for standard metaphase analysis are suitable for FISH after they have undergone a simple denaturation procedure. The use of whole-chromosome probes is commonly called “chromosome painting” (Tucker et al., 1993b; Speicher and Carter, 2005; Paccierotti and Sgura, 2008). Another significant advantage of FISH methods is that the probes can be used with interphase cells/chromosomes making any tissue potentially available for analysis (Vorsanova et al., 2010). Examples of cells showing chromosome painting and reciprocal translocations are shown in Figs. 9-5 and 9-6.

Chromosome painting facilitates cytogenetic analysis because aberrations are easily detected by the number of fluorescent regions in a painted metaphase. For example, if chromosome 4 were painted with a probe while the other chromosomes were counter-stained in a different color, one would see only the two homologues of chromosome 4 in the color of the probe in a normal cell. However, if there were a translocation or a dicentric chromosome and fragment involving chromosome 4, one would see three areas of fluorescence—one normal chromosome 4 and the two pieces involved in the chromosome rearrangement. Aberrations are detected only in the painted portion of the genome, but this disadvantage can be offset by painting a few chromosomes simultaneously with probes of different colors (Tucker et al., 1993b). FISH reduces the time and technical skill required to detect chromosome aberrations, and it permits the scoring of stable aberrations, such as translocations and insertions, that are not readily detected in traditional metaphase analysis without special labeling techniques. Using FISH, some chromosomal analysis can even be conducted in interphase cells (Paccierotti and Sgura, 2008; Vorsanova et al., 2010). Although FISH is not routinely used in genotoxicity testing, it is a valuable research tool for studying clastogens and is having a substantial impact in monitoring human populations for chromosomal damage.

Micronuclei

Metaphase analysis is time consuming and requires considerable skill, so simpler cytogenetic assays have been developed, of which micronucleus assays have become especially important. Micronuclei are membrane-bound structures that contain chromosomal fragments, or sometimes whole chromosomes, that were not incorporated into a daughter nucleus at mitosis. Because micronuclei usually represent acentric chromosomal fragments, they are most commonly used as simple indicators of chromosomal damage. However, the ability to detect micronuclei containing whole chromosomes has led to their use for detecting aneuploidy as well. Micronucleus assays may be conducted in primary cultures of human lymphocytes (Fenech et al., 2003; Corvi et al., 2008; Fenech, 2008; Fenech et al., 2011b), mammalian cell lines (Kirsch-Volders et al., 2003; Corvi et al., 2008), or mammals in vivo (Krishna and Hayashi, 2000; Hayashi et al., 2007; Dertinger et al., 2011).

Micronucleus assays in lymphocytes have been greatly improved by the cytokinesis-block technique in which cell division is inhibited with cytochalasin B, resulting in binucleate and multinucleate cells (Fenech et al., 2003, 2011b; Kirsch-Volders et al., 2003; Fenech, 2008). In the cytokinesis-block assay in human lymphocytes, nondividing (G₀) cells are treated with ionizing radiation or a radiomimetic chemical and then stimulated to divide with the mitogen phytohemagglutinin. Alternatively, the lymphocytes may be exposed to the mitogen first, so that the subsequent mutagenic treatment with radiation or chemicals includes the S period of the cell cycle. In either case, cytochalasin B is added for the last part of the culture period, and micronuclei are counted only in binucleate cells so as to ensure that the cells have undergone a single nuclear division that is essential for micronucleus development. The assay thereby avoids confusion owing to differences in cellular proliferation kinetics.

Although first devised using primary lymphocytes (Fenech et al., 2003; Fenech, 2008; Fenech, 2011b), the cytokinesis-block micronucleus assay has since been adapted for use in continuous cell cultures, including the Chinese hamster and mouse lymphoma cells that are widely used in other genotoxicity assays (Kirsch-Volders et al., 2003; Corvi et al., 2008). Micronucleus assays should be conducted in such a way that cellular proliferation is monitored along with the micronucleus frequency, and this is facilitated by the cytokinesis block. Reliable data have been obtained in cultured cells both with and without cytokinesis block, but scoring results only in binucleate cells after blockage of cytokinesis with cytochalasin B confers advantages with respect to the measurement of proliferation, recognizing whether an agent is cytostatic, and obtaining clear dose–response relationships (Kirsch-Volders et al., 2003).

Although micronuclei resulting from chromosome breakage comprise the principal endpoint in the cytokinesis-block micronucleus assay, the method can also provide evidence of aneuploidy, chromosome rearrangements that form nucleoplasmic bridges, inhibition of cell division, necrosis, apoptosis, and excision-repairable lesions (Fenech et al., 2003; Fenech, 2008; Fenech et al., 2011b). Micronuclei in a binucleate human lymphocyte are shown in Fig. 9-7. A recent review of the International Human Micronucleus (HUMN) Project provides a comprehensive description of standardized protocols for micronucleus assays in human lymphocytes and buccal cells together with a review of associations between micronucleus data and disease outcomes (Fenech et al., 2011a).

The in vivo micronucleus assay is often performed by counting micronuclei in immature (polychromatic) erythrocytes in the bone marrow of treated mice, but it may also be based on peripheral blood (Krishna and Hayashi, 2000; Hayashi et al., 2007). Micronuclei remain in the cell when the nucleus is extruded in the maturation of erythroblasts. In vivo micronucleus assays are increasingly used in genotoxicity testing as a substitute for bone marrow metaphase chromosome analysis. Micronucleus assays have been developed for mammalian tissues other than bone marrow and blood, including skin, duodenum, colon, liver, lung, spleen, testes, bladder, buccal mucosal cells, stomach, vagina, and fetal tissues (Coffing et al., 2011; Morita et al., 2011). Although assays in bone marrow and blood are the mainstay of genotoxicity testing, the new assays are important for mechanistic studies and research on the site specificity of genetic damage and carcinogenesis.

Micronuclei are most commonly visualized through microscopy, but automated means of detecting micronuclei are being developed through the application of flow cytometry. Flow cytometric detection is effective in micronucleus assays in rodent bone marrow or blood (Dertinger et al., 2011). It can also be used to detect micronuclei in Chinese hamster CHO cells, where altered flow cytometric parameters can reveal whether the micronuclei arose primarily by chromosome breakage or by aneuploidy (Bryce et al., 2011).

Sister Chromatid Exchanges

SCE, in which there has been an apparently reciprocal exchange of segments between the two chromatids of a chromosome, are visible cytologically through differential staining of chromatids. Fig. 9-8 shows SCE in human cells. Many mutagens induce SCE in cultured cells and in mammals in vivo (Tucker et al., 1993a; Wilson and Thompson, 2007). Despite the convenience and responsiveness of SCE assays, data on SCE are less informative than data on chromosome aberrations. There is uncertainty about the underlying mechanisms by which SCEs are formed and how DNA damage or perturbations of DNA synthesis stimulate their formation (Preston, 1991). SCE assays are therefore best regarded as general indicators of mutagen exposure, analogous to DNA damage and repair assays, rather than measures of a mutagenic effect.

Aneuploidy

Although assays for the induction of aneuploidy are not yet as refined as those for gene mutations and chromosome aberrations, they are being developed (Aardema et al., 1998). The targets of aneugens are often components of the mitotic or meiotic apparatus, rather than DNA. Therefore, aneugens should not be expected to overlap strongly with mutagens and clastogens. For example, a chemical that interferes with the polymerization of tubulin and thereby disrupts the formation of a mitotic spindle is likely to show specificity as an aneugen. Assays for chemicals that induce aneuploidy should therefore encompass all the relevant cellular targets that are required for the proper functioning of the mitotic and meiotic process. Means of detecting aneuploidy include chromosome counting (Galloway and Ivett, 1986; Natarajan, 1993; Aardema et al., 1998; Paccierotti and Sgura, 2008), the detection of micronuclei that contain kinetochores (Lynch and Parry, 1993; Natarajan, 1993; Aardema et al., 1998; Fenech, 2008; Paccierotti and Sgura, 2008), and the observation of abnormal spindles or spindle–chromosome associations in cells in which spindles and chromosomes have been differentially stained (Parry, 1998). FISH-based assays have also been developed for the assessment of aneuploidy in interphase somatic cells (Rupa et al., 1997; Paccierotti and Sgura, 2008) and in sperm (Russo, 2000; Marchetti et al., 2006, 2008).

A complication in chromosome counting is that a metaphase may lack chromosomes because they were lost during cell preparation for analysis, rather than having been absent from the living cell. To avoid this artifact, cytogeneticists generally use extra chromosomes (ie, hyperploidy) rather than missing chromosomes (ie, hypoploidy) as an indicator of aneuploidy in chromosome preparations from mammalian cell cultures (Galloway and Ivett, 1986; Aardema et al., 1998) or mouse bone marrow (Adler, 1993). Techniques for counting chromosomes in intact cells may allow reliable measures of hypoploidy (Natarajan, 1993), but the detection of hyperploidy remains the norm in lieu of clear evidence that artifactual chromosome loss has been avoided. It has been suggested that counting polyploid cells, which is technically straightforward, may be an efficient way to detect aneugens (Aardema et al., 1998), but there is disagreement on the point (Parry, 1998).

Micronucleus assays can detect aneugens as well as clastogens. Micronuclei that contain whole chromosomes tend to be somewhat larger than those containing chromosome fragments, but the two categories are not readily distinguished in typically stained preparations (Natarajan, 1993). However, one can infer that a micronucleus contains a whole chromosome if it is shown to contain a kinetochore or centromeric DNA. Aneuploidy may therefore be detected by means of antikinetochore antibodies with a fluorescent label or FISH with a probe for centromere-specific DNA (Lynch and Parry, 1993; Natarajan, 1993; Krishna and Hayashi, 2000; Fenech, 2008; Paccierotti and Sgura, 2008). Micronuclei containing kinetochores or centromeric DNA may be detected in cultured cells (Lynch and Parry, 1993; Aardema et al., 1998; Fenech, 2008; Paccierotti and Sgura, 2008) and in mouse bone marrow in vivo (Adler, 1993; Krishna and Hayashi, 2000). Frequencies of micronuclei ascribable to aneuploidy and to clastogenic effects may therefore be determined concurrently by tabulating micronuclei with and without kinetochores.

Germ Cell Mutagenesis

Gene Mutations

Germ cell mutagenesis assays are of special interest as indicators of genetic damage that can enter the gene pool and be transmitted through generations. Mammalian germ cell assays provide the best basis for assessing risks to human germ cells and therefore hold a central place in genetic toxicology despite their relative complexity and expense. The design of the test must compensate for the fact that mutations occur at low frequency, and even the simplest animal systems face the difficulty of their having a sufficiently large sample size. One can easily screen millions of bacteria or cultured cells by selection techniques, but screening large numbers of mice poses practical limitations. Therefore, a germ cell assay must offer a straightforward, unequivocal identification of mutants with minimal labor (Singer and Yauk, 2010).

The mouse specific-locus test detects recessive mutations that produce easily analyzed, visible phenotypes (coat pigmentation and ear size) conferred by seven genes (Russell and Shelby, 1985; Ehling, 1991; Russell and Russell, 1992; Favor, 1999; Russell, 2004; Singer et al., 2006). Mutants may be classified as having point mutations or chromosomal alterations on the basis of genetic and molecular analysis (Favor, 1999). The assay has been important in assessing genetic risks of ionizing radiation and has been used to study various chemical mutagens. Although they have been used less extensively, there are other gene mutation assays in mouse germ cells based on dominant mutations that cause skeletal abnormalities (Selby et al., 2004) or cataracts (Ehling, 1991) and recessive mutations that cause electrophoretic changes in proteins (Lewis, 1991).

Mammalian assays permit the measurement of mutagenesis at different germ cell stages (Favor, 1999; Russell, 2004). Late stages of spermatogenesis are often found to be sensitive to mutagenesis, but effects in spermatocytes, spermatids, and spermatozoa are transitory. Mutagenesis in stem-cell spermatogonia and resting oocytes is of special interest in genetic risk assessment because of the persistence of these stages throughout reproductive life. Chemical mutagens show specificity with respect to germ cell stages. For example, ethylnitrosourea and chlorambucil are both potent mutagens in the mouse specific-locus test, but the former induces primarily point mutations in spermatogonia, whereas the latter mostly induces deletions in spermatids (Russell and Russell, 1992). The ratio of deletions to point mutations is not only a function of the nature of the mutagen but depends on germ cell stage, as some mutagens induce higher proportions of gross alterations in late stages of spermatogenesis than in spermatogonia (Lewis, 1991; Favor, 1999; Russell, 2004).

There is currently no unequivocal evidence of induced gene mutations in human germ cells, but studies in mice leave little doubt about the susceptibility of mammalian germ cells to mutagenesis by radiation and chemicals. New molecular methods, particularly those involving the assessment of changes in tandem repeat loci (Yauk, 2004; Dubrova, 2005; Singer et al., 2006; Somers, 2006), hold great promise for the development of systems that will permit the efficient detection of genetic alterations in human germ cells. The development of methods based on expanded simple tandem repeat (ESTR) loci in mice and other species is important, in that it permits in situ monitoring for environmental mutagens and the quantification of mutagenesis after controlled exposures of laboratory animals using systems parallel to those being developed for human monitoring (Yauk, 2004; Dubrova, 2005; Wu et al., 2006; Somers, 2006). Basic research on mechanisms underlying ESTR changes is essential, as it is still unclear how ESTR changes relate to the gene mutations that have been long studied by population geneticists and are detected in classical gene-mutation assays.

Chromosomal Alterations

Cytogenetic assays in germ cells are not routinely included in mutagenicity testing, but they are an important source of information for assessing risks to future generations posed by the induction of chromosome aberrations. Metaphase analysis of germ cells is feasible in rodent spermatogonia, spermatocytes, or oocytes (Kirkland et al., 1990; Tease, 1992; Russo, 2000; Marchetti et al., 2001). A micronucleus assay has also been developed in which chromosomal damage induced in meiosis is measured by the observation of rodent germ cells, principally spermatids (Russo, 2000; Hayashi et al., 2007).

Aneuploidy originating in mammalian germ cells may be detected cytologically through chromosome counting for hyperploidy (Allen et al., 1986; Adler, 1993; Aardema et al., 1998; Russo, 2000; Marchetti et al., 2001) or genetically in the mouse sex-chromosome loss test (Russell and Shelby, 1985), but these methods are not widely used in toxicological testing. A promising development is the detection of aneuploidy in the sperm of mice or rats by FISH with chromosome-specific probes (Baumgarthner et al., 1999; Russo, 2000; Marchetti et al., 2006, 2008). The presence of two fluorescent spots indicates the presence of an extra copy of the chromosome identified by the probe; probes for several chromosomes are used simultaneously so that aneuploid sperm are distinguishable from diploid sperm.

Besides cytological observation, indirect evidence for chromosome aberrations is obtained in the mouse heritable translocation assay, which measures reduced fertility in the offspring of treated males (Russell and Shelby, 1985; Singer et al., 2006). This presumptive evidence of chromosomal rearrangements can be confirmed through cytogenetic analysis. Data from the mouse heritable translocation test in postmeiotic male germ cells have been used in an attempt to quantify human germ cell risk for ethylene oxide, a mutagen used as a fumigant, sterilizing agent, and reactant in chemical syntheses (Rhomberg et al., 1990; Preston et al., 1995).

Dominant Lethal Mutations

The mouse or rat dominant lethal assay (Adler et al., 1994; Singer et al., 2006) offers an extensive database on the induction of genetic damage in mammalian germ cells. In the most commonly used version of the assay, males are treated on an acute or subchronic basis with the chemical of interest and then mated with virgin females at appropriate intervals. The females are killed and necropsied during pregnancy so that embryonic mortality may be characterized and quantified. Most dominant lethal mutations, manifested as intrauterine deaths, are thought to arise from chromosomal anomalies.

Development of Testing Strategies

Concern about adverse effects of mutation on human health, principally carcinogenesis and the induction of transmissible damage in germ cells, has provided the impetus to identify environmental mutagens. Priorities must be set for testing because it is not feasible to conduct multiple tests of all chemicals to which people are exposed. Such factors as production volumes, intended uses, the extent of human exposure, environmental distribution, and effects that may be anticipated on the basis of chemical structure or previous testing must be considered in order to ensure that compounds with the greatest potential for adverse effects receive the most comprehensive study. The most obvious use of genetic toxicology assays is screening chemicals to detect mutagens, but they are also used to obtain information on mutagenic mechanisms and dose–responses that contribute to an evaluation of hazards. Besides testing pure chemicals, environmental samples are tested because many mutagens exist in complex mixtures (DeMarini 1998; Ohe et al., 2003; White, 2004). The analysis of complex mixtures often requires a combination of mutagenicity assays and refined analytical methods (White, 2004; Hewitt and Marvin, 2005). Assessment of a chemical's genotoxicity requires data from well-characterized assays. Assays are said to be validated when they have been shown to perform reproducibly and reliably with many compounds from diverse chemical classes in several laboratories. An evaluation of test performance, however, sometimes extends beyond determining whether the assay effectively detects the specific endpoint that it actually measures to whether it is predictive of other endpoints of interest. For example, there is great interest in the ability of mutagenicity tests, which do not measure carcinogenicity per se, to predict whether chemicals are carcinogens.

Mutagenicity testing, combined with an evaluation of chemical structure, has been found to identify a large proportion of trans-species, multiple-site carcinogens (Tennant and Ashby, 1991; Gold et al., 1993). In contrast, some carcinogens are not detected as mutagens. Putatively nongenotoxic carcinogens often give responses that are more specific with respect to species, sites, and conditions (Ashby and Paton, 1993; Gold et al., 1993). In predicting carcinogenicity, one should consider both the sensitivity and the specificity of an assay. Sensitivity refers to the proportion of carcinogens that are positive in the assay, whereas specificity is the proportion of noncarcinogens that are negative (Tennant et al., 1987; McGregor et al., 1999). Sensitivity and specificity both contribute to the predictive reliability of an assay. The commonly held view that deficiencies in the sensitivity or specificity of individual assays may be circumvented by using assays in complementary combinations called tiers or batteries has fallen into disfavor because, rather than offsetting each other's strengths and weaknesses, genetic toxicology assays are often consistent with one another (Tennant et al., 1987; Ashby and Tennant, 1991; Kim and Margolin, 1999).

Strategies for testing have evolved over the last few decades, such that data from a few well-chosen assays are now considered sufficient (MacGregor et al., 2000). Rather than trying to assemble extensive batteries of complementary assays, it is prudent to emphasize mechanistic considerations in choosing assays. Such an approach makes a sensitive assay for gene mutations (eg, the Ames assay) and an assay for clastogenic effects in mammals pivotal in the evaluation of genotoxicity, and this is the basis for our highlighting these assays in Table 9-1. The Ames assay has performed reliably with hundreds of compounds in laboratories throughout the world. Other bacterial assays and mammalian cell assays also provide useful information on gene mutations. Beyond gene mutations, one should evaluate damage at the chromosomal level with a mammalian in vitro or in vivo cytogenetic assay. Cytogenetic assays in rodents are especially useful for this purpose because they combine a well-validated genetic assay with mammalian pharmacodynamics and metabolism. The other assays in Table 9-1 offer an extensive database on chemical mutagenesis (ie, Drosophila SLRL), a unique genetic endpoint (ie, aneuploidy; mitotic recombination), applicability to diverse organisms and tissues (ie, DNA damage assays, such as the comet assay), or special importance in the assessment of genetic risk (ie, germ cell assays). The more extensive listing of assays in Table 9-2 provides references that can be helpful in interpreting genetic toxicology data that can be found in the scientific literature.

For cancer risk assessment considerations, the human data utilized most frequently, in the absence of epidemiologic data, are those collected from genotoxicity/mutagenicity assessments in human populations. For this purpose, the studies conducted most frequently are for chromosome aberrations, micronuclei, mutations (for several loci), and SCEs in peripheral lymphocytes. Cytogenetic alterations have also been assessed in a small number of bone marrow samples. Mutations at the HPRT locus have been assessed in peripheral lymphocytes. Glycophorin A variants have been studied in red blood cells.

An important component of any population monitoring study is the selection of the study groups, namely those individuals who are potentially exposed and the matched unexposed controls. The size of each study group should be sufficiently large to avoid any confounder having undue influence. Certain characteristics should be matched among exposed and unexposed groups. These include age, sex, smoking status, and general dietary features. Certain characteristics are exclusionary, namely current or recent medication, radiation exposure, and certain illnesses. It is possible to develop a lengthy list of additional possible confounders of response that would make the selection of suitable study groups very difficult indeed. Study groups of 20 or more individuals can be used as a reasonable substitute for exact matching because confounders will be less influential on chromosome alteration or mutation frequency in larger groups (discussed in Au et al., 1998; Battershill et al., 2008)). In some instances, it might be informative to compare exposed groups with a historical control, as well as to a concurrent control.

The magnitude of different known confounders varies considerably among studies, based in part on the size of the study populations. Some general indication of the magnitude of the effects of age and smoking status on the frequencies of chromosome aberrations and SCE is presented to illustrate the importance of accounting for confounders in the design of a population monitoring study. The comparisons presented are for large studies only. For chromosome aberrations, the frequency of aberrations has been reported in one large study to be about 50% higher in smokers (l.5 aberrations per 100 cells in smokers vs 1.0 per 100 cells in the nonsmokers) (Galloway et al., 1986) and in another no difference between smokers and nonsmokers (Bender et al., 1988a,b). The complete data set has been reviewed by Au et al. (1998). In general, the frequency of SCE is increased by about one SCE per cell in smokers compared with nonsmokers (Bender et al., 1988a,b; Barale et al., 1998). The study by Barale et al. (1998) also reported a dose–response association between SCE frequency and smoking level. The differences among these studies might well be accounted for by differences among confounders in the respective groups, but it is virtually impossible to correct for these in this type of study.

The frequency of chromosome aberrations, particularly chromosome-type (reciprocal) exchanges, has been shown to increase with age of subject (Tucker and Moore, 1996). Galloway et al. (1986) reported an increase from 0.8 per 100 cells at about 25 years of age to about 1.5 at 60. Bender et al. (1988a,b) reported an increase with age only for chromosome-type dicentric aberrations, but the increase over a broad age range was small and just statistically significant. Ramsey et al. (1995), using chromosome painting techniques, reported that individuals 50 years and older had frequencies of stable aberrations, dicentrics, and acentric fragments that were 10.6-, 3.3-, and 2.9-fold, respectively, greater than the frequency in cord bloods. Bender et al. (1998a,b) did not find an increase in SCE frequency with the increasing age of the subject. The differences among the results from these large control studies emphasize the difficulty of adequately accounting for confounders (age and smoking presented here) when only a small control group is used, as is frequently the case.

Similar sources of variation have been identified for the monitoring of individuals for HPRT mutations. The data are reviewed in detail by Albertini and Hayes (1997). There is less information on sources of variation of glycophorin A (GPA) variants, although quite considerable interindividual variation exists (reviewed in Cole and Skopek, 1994; Kyoizumi et al., 2005).

For cytogenetic assays (chromosome aberrations, SCEs, and micronuclei) the alterations are produced as a consequence of errors of DNA replication, as discussed in previous sections. From the nature of the alterations, assessed in traditional cytogenetic assays in which nontransmissible alterations are analyzed, it can be established that these alterations were produced at the first in vitro S phase. Irrespective of the duration of exposure, the frequency of cytogenetic alterations will be proportional to that fraction of the DNA damage that remains at the time of in vitro DNA replication. All the DNA damage induced by potent clastogens that results in chromosome alterations is repaired within a relatively short time after exposure for G₀ human lymphocytes. Thus, for chronic exposures the lymphocyte cytogenetic assay as typically conducted is insensitive.

It is now possible to analyze reciprocal translocations using FISH methods (reviewed in Tucker et al., 1997; Kleinerman et al., 2006; Beskid et al., 2006), and because this aberration type is transmissible from cell generation to generation, its frequency can be representative of an accumulation over time of exposure. The importance of this is that stable chromosome aberrations observed in peripheral lymphocytes exposed in vivo, but assessed following in vitro culture, are produced in vivo in hematopoietic stem cells or other precursor cells of the peripheral lymphocyte pool. To date, population cytogenetic monitoring studies involving the analysis of reciprocal translocations in chemically exposed individuals or radiation-exposed individuals have been conducted quite rarely (Lucas et al., 1992; Smith et al., 1998; Kleinerman et al., 2006). The overall sensitivity of the FISH analysis of reciprocal translocations for assessing effects of chronic, low levels of exposure to chemical clastogens has not been established. However, a cautionary note is provided by the study of Director et al. (1998), who showed that there was no increase in reciprocal translocations assessed by FISH following exposure to cyclophosphamide (0, 32, 64, or 96 ppm) or urethane (0, 5000, 10,000, or 15,000 ppm) for up to 12 weeks. In contrast, recent data on ethylene oxide (Donner et al., 2010) have shown that exposure of male mice to ethylene oxide at concentrations of 0, 25, 50, 100, 200 ppm for 6, 12, 24, or 48 weeks resulted in a time and concentration-dependent increase in reciprocal translocations assessed by FISH.

Another factor that certainly affects the utility of population monitoring data with reciprocal translocations using FISH is that the frequency of reciprocal translocations increases significantly with increasing age (Ramsey et al., 1995), but to a lesser extent for nontransmissible aberrations (Bender et al., 1988a,b). Ramsey et al. (1995) provided data on the influence of other confounders on the frequency of reciprocal translocations in human groups. These confounders include smoking, consumption of diet drinks and/or diet sweeteners, exposure to asbestos or coal products, and having a previous major illness. This reemphasizes the point that the selection of study groups and accounting for confounders is essential for human population cytogenetic monitoring studies to be of utility.

Thus, very few of the published studies of cytogenetic population monitoring for individuals have analyzed the appropriate endpoint for detecting the genetic effects of long-term exposure to chemicals. It is quite surprising that positive responses have been reported for increases in unstable, chromatid aberrations because these are nontransmissible, and as noted above are induced at the first in vitro S phase. This anomaly is especially concerning when positive responses at very low levels of exposure are reported (reviewed for ethylene oxide in Preston, 1999b).

The HPRT mutation assay can assess the frequency of induced mutations in stem cells or other precursor cells because a proportion of the mutations are induced as nonlethal events. The transmissible proportion will be greater for chemicals that do not induce large deletions; this will include the majority of nonradiomimetic chemicals. Induction of mutations in lymphocyte precursor cells will lead to clonal expansion of mutations in the peripheral pool. However, assessment of the T-cell antigen receptor status of the mutant clones permits a correction for clonal expansion. The population of cells derived from any particular stem cell has a unique antigen receptor status (Albertini and Hayes, 1997). The GPA assay can similarly be used for the assessment of chronic exposures or for estimating exposures at some long time after exposure (Albertini and Hayes, 1997). The predictive value of the assay for adverse health outcome appears to be limited, but it can provide an estimate of exposure.

The potential for cytogenetic endpoints being predictive of relative cancer risk has been addressed in recent reports from the European Study Group on Cytogenetic Biomarkers and Health (Hagmar et al., 1998a,b; Bonassi et al., 2004; Norppa et al., 2006). The groups selected for cytogenetic studies consisted of individuals with reported occupational exposure and unexposed controls. The association between cancer and the frequency of unstable chromosome aberrations in the study groups was not based on exposure status, but rather on the relative frequency of chromosome aberrations, namely, low (1–33 percentiles), medium (34–66 percentiles), and high (67–100 percentiles). In general, the higher the relative frequency of unstable aberrations, the greater the risk of cancer death for all tumors combined. The authors make it clear that there is insufficient information on exposure for it to be used as a predictor of cancer development. In fact, the data indicate that individuals with higher frequencies of chromosome aberrations for whatever reason (genetic or environmental) are as a group at greater risk of dying from cancer. This is very different from concluding that exposures to mutagens that result in a higher frequency of chromosome aberrations in peripheral lymphocytes leads to an increased risk of cancer, especially for specific tumor types. The relevance of exposure to mutagenic chemicals in these studies by Hagmar et al. (1998a,b) is uncertain because there was no association between increased SCE frequencies, which may be indicative of higher exposure levels, and increased cancer mortality.

This latter concern was addressed by the same group (Bonassi et al., 2000) in a more recent study. The study again showed that there was a significantly increased risk for subjects with a high level of chromosome aberrations compared to those with a low level in both Nordic and Italian cohorts. Of particular relevance to risk assessment was the observation that the relationships were not affected by the inclusion of occupational exposure level or smoking. The risk for high versus low levels of chromosome aberrations was similar in individuals heavily exposed to carcinogens and in those who had never, to their knowledge, been exposed to any specific environmental carcinogen. These data highlight the need to use caution when considering the relevance of chromosome aberration data from human biomonitoring studies in cancer risk assessment.

In the last 15 or so years, the field of genetic toxicology has moved into the molecular era. The potential for advances in our understanding of basic cellular processes and how they can be perturbed is enormous. The ability to manipulate and characterize DNA, RNA, and proteins has been at the root of this advance in knowledge. However, the development of sophisticated molecular biology does not in itself imply a corresponding advance in the utility of genetic toxicology and its application to risk assessment. Knowing the types of studies to conduct and knowing how to interpret the data remain as fundamental as always. Measuring finer and finer detail can perhaps complicate the utility of the various mutagenicity assays. There is a need for genetic toxicology to avoid the temptation to use more and more sophisticated techniques to address the same questions and in the end make the same mistakes as have been made previously. How successful we are in designing informative studies based on the most recent molecular techniques perhaps cannot be judged at this time. However, the following examples of recent approaches to obtaining data for enhancing our ability to use noncancer (genotoxicity) data in a mechanistically based cancer (and genetic) risk assessment process provide some encouragement. Several recent developments (eg, the refinement of FISH painting, the high-throughput adverse pathway detection approach, and the use of ultrahigh throughput DNA sequencing technologies) have already been described in the appropriate assay sections above because they are currently in general use.

Advances in Cytogenetics

Until quite recently, the analysis of chromosome alterations relied on conventional chromosome staining with DNA stains such as Giemsa or on the process of chromosome banding. Both approaches require considerable expenditure of time and a rather high level of expertise. However, chromosome banding does allow for the assessment of transmissible aberrations such as reciprocal translocations and inversions with a fairly high degree of accuracy. Knowing the induction frequency of such aberrations is very important, given that they are generally not lethal to the cell and constitute by far the major class observed in inherited genetic defects and a significant fraction of the alterations observed in tumors. In addition, because stable aberrations are transmissible from parent to daughter cell, they represent accumulated effects of chronic exposures. The more readily analyzed but cell-lethal, nontransmissible aberrations such as dicentrics and deletions reflect only recent exposures and then only when analyzed at the first division after exposure. A more detailed discussion of these factors can be found in Preston (1999b).

The relative ease with which specific chromosomes, specific genes, and chromosome alterations can be detected has been radically enhanced by the development of FISH (Trask et al., 1993; Speicher and Carter, 2005). In principle, the technique relies on amplification of DNA from particular genomic regions such as whole chromosomes or gene regions and the hybridization of these amplified DNAs to metaphase chromosome preparations or interphase nuclei. Regions of hybridization can be determined by the use of fluorescent antibodies that detect modified DNA bases incorporated during amplification or by incorporating fluorescent bases themselves during amplification. The fluorescently labeled, hybridized regions are detected by fluorescence microscopy, and the signal can be increased in strength by computer-enhanced processes. The level of sophistication has increased so much that all 24 different human chromosomes (22 autosomes, X and Y) can be individually detected (Macville et al., 1997), as can all mouse chromosomes (Liyanage et al., 1996) (Figs. 9-5 and 9-6). Most recently, M-FISH (multicolor FISH) and M-BAND (multicolor banding) have been developed for defining chromosome abnormalities (Mackinnon and Chudoba, 2011) (Fig. 9-9). Alterations in tumors can also be detected on a whole-genome basis (Coleman et al., 1997; Veldman et al., 1997; Bridge and Cushman-Vokoun, 2011). The following example highlights the ability to construct breakpoint profiles of specific tumor types (Trost et al., 2006). In this example, a detailed analysis by spectral karyotyping of specific breakpoints in a set of primary myelodysplastic syndrome and acute myeloid leukemia samples revealed recurrent involvement of specific chromosome bands that contained oncogenes or tumor-suppressor genes. The aim will be to attempt to reveal the possible prognostic significance of the subgroups linked to these specific markers.

There is an extensive literature on the use of FISH for karyotyping tumors and in gene mapping but less on its utility for genetic toxicology studies, especially the assessment of stable chromosome aberrations at long periods after exposure or after long-term exposures. Three particular studies do, however, serve to exemplify the use of FISH in genetic toxicology.

Lucas et al. (1992) demonstrated that stable chromosomal aberrations could be detected in individuals decades after exposure to atomic bombs in Japan. How these frequencies relate to frequencies at the time of exposure is not known with any certainty, given the fact that induced frequencies were not measured because appropriate techniques were not available at that time.

Studies by Tucker et al. (1997, 2005) provided some assessment of the utility of FISH for the analysis of radiation-induced, stable chromosome alterations at various times after exposure. The frequency of reciprocal translocations induced by γ-rays in rat peripheral lymphocytes decreased with time after exposure, reaching a plateau at four days that was 55% to 65% of the induced frequency and with a dose dependency (Tucker et al., 1997). Similar results were obtained for human samples (Tucker et al., 2005). These results suggest that reciprocal translocations fall into two classes, stable and unstable (cell-lethal). It is quite possible that these “unstable” translocations are lost because of the presence of other cell-lethal damage in the same cell. Additional work is required to clarify this conclusion and to extend the studies to the effects of chemicals. In addition, a recent study by Tucker and Luckinbill (2011) demonstrated how FISH translocation analysis could be used to detect extremely low doses of ionizing radiation—a few centigray per year. Such analysis could be used to estimate responses at very low doses for utility in risk assessments at similarly low doses.

FISH methods have also allowed for an accurate and sensitive assessment of chromosomal alterations present in tumors. The particular advance that makes this assessment feasible is known as comparative genomic hybridization (CGH) (Kallioniemi et al., 1992). CGH results in the ability to identify the role of chromosomal structural and numerical alterations in tumor development. The genomic instability present in all tumor types appears to have a specific genetic basis, as shown elegantly for colon cancer by Vogelstein and colleagues (Cahill et al., 1998). For CGH, tumor and control DNAs are differentially labeled with fluorescence probes and cohybridized to normal metaphase chromosome preparations. The ratio of the fluorescence intensities of hybridized tumor and control DNA indicates regions of normal genomic content as well as those regions that are over- or underrepresented in tumors. The CGH method is being adapted for automated screening approaches using biochips (Solinas-Toldo et al., 1997; Hosoya et al., 2006). Assessing genetic alterations such as specific gene deletions in single metastatic tumor cells is feasible using a slightly different but complementary approach (Pack et al., 1997, 2005).

The types of FISH approaches described here undoubtedly indicate the direction in which cytogenetic analysis will proceed. The types of data collected will affect our understanding of how tumors develop. Data on the dose–response characteristics for a specific chromosomal alteration as a proximate marker of cancer can enhance the cancer risk assessment process by describing effects of low exposures that are below those for which tumor incidence can be reliably assessed. Cytogenetic data of the types described above can also improve extrapolation from data generated with laboratory animals to humans.

Molecular Analysis of Mutations and Gene Expression

With the advent of molecular biology techniques, the exact basis of a mutation at the level of the DNA sequence can be established. In many cases, the genetic basis of human disease can be determined even though human genes have long DNA sequences and a complex genomic arrangement. Molecular biology techniques have also enabled a distinction to be made between background mutations and those induced by specific agents. The latter observations are addressed by analyzing the mutational spectra in target genes in laboratory organisms and in humans (DeMarini, 2000; Hemminki and Thilly, 2004). For reasons of inherent sensitivity of available methods, the genes analyzed for mutations are ones for which mutated forms can be selected. The confounding factor of many normal cells, which far outnumber a few mutant cells in an exposed cellular population, can be removed by mutant selection approaches. Methods to overcome the drawback of only being able to study selectable genes have been developed, and particular ones such as ligation-mediated PCR are close to the required sensitivity level (Albertini and Hayes, 1997; Makrigiorgos, 2004; Yeh et al., 2006). For example, COLD-PCR (co-amplification at lower denaturation temperature-PCR) was developed to selectively amplify minority alleles from mixtures of wild-type and mutation-containing sequences (Li and Makrigiorgos, 2009).

A giant step forward in the ability to detect and characterize mutations at both the DNA and RNA level has been provided by the development of chip technology (Southern, 1996) and array-based assay systems (Wodicka et al., 1997). With hybridization of test DNAs to oligonucleotide arrays, specific genetic alterations or their cellular consequences can be determined rapidly and automatically (Houlston and Peto, 2004; Vissers et al., 2005). Cost remains a limiting factor, but the potential for assessing specific cellular changes following chemical exposure is enormous. Perhaps of even greater practical importance to the ability to detect mutations is that in the past few years the advent of second- and third-generation ultrahigh throughput DNA sequencing techniques has enormously enhanced the ability to assess mutations quantitatively at the whole genome level in a very short period of time. Such techniques have been used, for example, to characterize the genetic alterations in a number of tumor types (Stratton et al., 2009). Until quite recently, alterations in gene expression following specific exposures or for specific genotypes were analyzed gene by gene. Such an approach makes it difficult to assess changes in gene expression that occur in a concerted fashion. Recent advances using cDNA microarray technologies have allowed the measurement of changes in expression of hundreds or even thousands of genes at one time (Harrington et al., 2000; Elvidge, 2006). The level of expression at the mRNA level is measured by the amount of hybridization of isolated cDNAs to oligonucleotide fragments from known genes or expressed sequence tags (ESTs) on a specifically laid out grid. Although this technique holds great promise for establishing a cell's response to exposure to chemical or physical agents in the context of normal cellular patterns of gene expression, it remains to be established how to analyze the vast amounts of data that are being obtained and what magnitude of change in gene expression constitutes an adverse effect as far as cellular phenotype is concerned. Extrapolating the responses to organs and whole animals represents a challenge still to be addressed. These microarray-based techniques are now being replaced by massively parallel sequencing or ultrahigh throughput sequencing approaches that can quantitatively assess gene expression changes in response to exposures. Such sequencing-based techniques have the great advantage that they are based on molecule counting approaches rather than on hybridization, thereby making them more quantitative and able to detect very low level transcripts (Blencowe et al., 2009). There are parallel efforts in the area of proteomics and metabolomics whereby changes in a broad range of cellular proteins can be assessed in response to endogenous or exogenous factors, potentially leading to the development of biomarkers of effect (Aebersold et al., 2005; Robertson, 2005; Griffin, 2006; McGregor and Souza, 2006). The biggest hurdle currently is the relative paucity of sequence data available for the world of proteins and their multiple posttranslational modifications. Certainly progress is rapid, and so methodologies akin to gene expression assessment are likely to be close at hand (Xie et al., 2011).

The move in the field of genetic toxicology is away from the “yes/no” approach to hazard identification and much more toward a mechanistic understanding of how a chemical or physical agent can produce adverse cellular and tissue responses. In turn such knowledge can be used for the development of informative bioindicators representing the key events along the pathway from initial interactions with cells to adverse outcome (Jarabek et al., 2009). The move is clearly toward analysis at the whole genome level and away from single gene responses. The challenges are apparent and the solutions are being identified at a rapid pace.

The field of genetic toxicology has had an overall life of about 70 years and has undergone several rebirths during this period. Genetic toxicology began as a basic research field with demonstrations that ionizing radiations and chemicals could induce mutations and chromosome alterations in plant, insect, and mammalian cells. The development of a broad range of short-term assays for genotoxicity served to identify many mutagens and address the relationship between mutagens and cancer-causing agents, or carcinogens. The inevitable failure of the assays to be completely predictive resulted in the identification of nongenotoxic carcinogens. In the 1980s, genetic toxicology began to move more toward providing insights into a better understanding of the mutagenic mechanisms underlying carcinogenicity and heritable effects. With this improved understanding, genetic toxicology studies began to turn away from hazard identification alone and move toward quantitative risk assessment. Major advances in our knowledge of mechanisms of cancer formation have been fueled by truly amazing progress in molecular biology.

Genetic toxicology has begun to take advantage of the knowledge that cancer is a genetic disease with multiple steps, many of which require a mutation. The identification of chromosome alterations involved in tumor formation has been facilitated greatly by the use of FISH. The ability to distinguish between background and induced mutations, in some cases, can be achieved by mutation analysis at the level of DNA sequence. Second- and third-generation ultrahigh throughput sequencing technologies have allowed such whole genome mutation analysis to be accomplished in a very short time. Key cellular processes related to mutagenesis have been identified, including multiple pathways of DNA repair, cell cycle controls, and the role of checkpoints in ensuring that the cell cycle does not proceed until the DNA and specific cellular structures are checked for fidelity. These observations have enhanced our knowledge of the importance of genotype in susceptibility to cancer. Recent developments in genetic toxicology have greatly improved our understanding of basic cellular processes and alterations that can affect the integrity of the genetic material and its functions. The ability to detect and analyze mutations in mammalian germ cells continues to improve and can contribute to a better appreciation for the long-term consequences of mutagenesis in human populations. Improvements in the qualitative assessment of mutation in somatic cells and germ cells have been paralleled by advances in the ability to assess genetic alterations quantitatively, especially in ways that enhance the cancer and genetic risk assessment process (Preston, 2005).

The authors thank Drs David DeMarini and Andrew Kligerman for their valuable comments as part of the review of this chapter.

This document has been reviewed in accordance with the USEPA policy and approved for publication but does not necessarily reflect EPA policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.