RICHARD J. ALBERTINI, J. PATRICK O'NEILL,

AND JANICE A. NICKLAS

Somatic mutations in critical genes are initiating events in cancer. Although studies of oncogenes and tumor-suppressor genes are crucial for an understanding of pathogenesis, mutations in non-disease-related reporter genes may be more informative regarding the mutation process itself. One such reporter is the human hypoxanthine-guanine phosphoribosyltransferase (hprt) gene. In vivo hprt mutations in T lymphocytes can be assessed by either a short-term or a cloning assay. The latter permits isolation of mutant cells for molecular analyses. A worldwide database has now established the normal background mutant frequency (MF) for hprt . This background, some modifying factors, and the results of illustrative studies of the mutagenic effects of chemical and radiation exposures are summarized. A synopsis of the molecular spectra as currently defined for both background and radiation-induced in vivo hprt mutations is also presented. These spectra show that carcinogenic mutagenic mechanisms are captured in the hprt reporter gene, especially those that are operative in several of the lymphoid malignancies. The possibility of assessing in vivo hprt mutations in humans in cell types other than lymphocytes is considered.

Somatic mutations arise in normal individuals. Some are spontaneous mistakes due to replication errors, while others result from exposures to endogenous or exogenous mutagens. The vast majority of these mutations are probably harmless to the

individual. However, rarely, critical genes or genetic regions are involved, leading to disease. This is perhaps best illustrated in cancer, where somatic mutations are the initiating pathogenic events.

The study of genes of oncological significance holds the central position in current cancer research. While molecular analyses of mutations in cancer-relevant genes define the relationship between their malfunction and pathogenesis, studies of mutations in the harmless or reporter regions of the genome may be more useful for defining the mutation process itself. Part of the reason is that cancer-related genetic events are usually non-selectable, making it difficult to recognize generalized classes of rare events. More importantly, mutations in oncogenes or tumor-suppressor genes often alter cell proliferation, making it impossible to measure primary frequencies. It is these primary frequencies that are the important measurements both for defining mutagenic mechanisms and for public health.

Assays for in vivo arising mutations are of considerable value for monitoring human populations for potential health hazards due to environmental mutagens and carcinogens. This value is enhanced if the mutant cells can be recovered for molecular analyses. In this case, the in vivo mutants become material for basic research into mutational mechanisms.

The gene for hypoxanthine-guanine phosphoribosyltransferase (hprt), located at position Xq26, is constitutively expressed but dispensable for cell viability in virtually all tissues. The encoded HPRT enzyme phosphoribosylates hypoxanthine and guanine for reutilization through the purine salvage pathway. Without this activity, the de novo pathway of purine biosynthesis is used by the cell so that viability is maintained. HPRT phosphoribosylates the purine analogues 8-azaguanine (AG) and 6-thioguanine (TG) as well as its normal substrates. In the absence of this metabolic conversion, these analogues are not cytotoxic. Cells that lack HPRT activity are therefore resistant to killing by these agents, while HPRT-proficient cells are not.

Mutations of the hprt gene provide a selectable reporter event for in vivo somatic mutations in humans (Albertini et al., 1982, 1990; Morley et al., 1983). Two methods are available for assessing these mutations in peripheral blood T lymphocytes. One is a short-term phenotypic assay, which has the advantage of speed and potential for automation but the disadvantage of consuming the mutant cells, which are then not available for study (Albertini et al., 1981, 1988; Ammenheuser et al., 1988, 1991, 1994; Ostrosky-Wegman et al., 1988, 1990; Morley et al., 1983; Montero et al., 1991; Ward et al., 1994). The other is a cloning assay which allows for molecular analyses (Albertini et al., 1982, 1990; O'Neill et al., 1987, 1989). Issues of historical importance for in vivo mutation research such as demonstrating unequivocally the genetic basis of the measured endpoints (Albertini et al., 1985; Turner et al., 1985) and the relationship between mutational events and their mutant progeny (Nicklas et al., 1986, 1989) have been resolved for the hprt system using the cloning assay.

This chapter describes both methods for assessing hprt mutations arising in vivo in human T cells. Quantitative results from our laboratory and elsewhere are presented, with emphasis on results using the cloning assay. The bulk of the chapter considers the molecular bases of hprt mutations using results from several laboratories but primarily from our own. The importance of characterizing T-cell receptor (TCR) genes among mutant isolates, allowing mutations to be analyzed in the context of in vivo clonality, is detailed. Finally, the degree to which carcinogenic mutagenic mechanisms are captured by the reporter hprt mutations is examined. This capture determines the validity of using these mutations as surrogates for cancer-causing genetic events in humans.

The T lymphocytes of human peripheral blood are isolated by density centrifugation, cryopreserved, and then stored in liquid nitrogen. Cryopreservation is required for the short-term assays so that cycling T cells in the peripheral blood do not distort results (see below) (Albertini et al., 1981). The measured endpoint in the short-term assays is T cells' in vitro resistance to TG inhibition during DNA synthesis when stimulated by phytohemagglutinin (PHA). Culture intervals are short—42 hours; that is, 24 hours before and 18 hours after addition of label in the absence (control) or presence (test) of TG. The label is either ³H-thymidine for enumeration of resistant cells by authoradiography (Albertini et al., 1981; Ammenheuser et al., 1988) or 5-bromodeoxyuridine (Montero et al., 1991) for the scoring of TG-resistant cells by fluorescence differences. The labeling index of test cultures divided by the total number of enumerable cells in these cultures (labeling index of controls × total cells) gives the variant frequency (VF). VF is assumed to be the frequency of mutants in the peripheral blood. The term "variant" is used because, strictly speaking, mutation cannot be demonstrated because resistant cells are consumed in the assay. The short-term assays are thus phenotypic assays for mutation.

Peripheral blood T lymphocytes are obtained in the same way as for the short-term assay. Cryopreservation in this case is optional but is often convenient so that groups of assays can be performed at the same time. The measured endpoint in the cloning assay is clonal growth in the presence of TG. The T cells are cultured in limiting dilutions in the wells of microtiter dishes in the absence (control) or presence (test) of TG and scored microscopically in 8 to 14 days (O'Neill et al., 1987, 1989). Control plates contain few cells per well, typically 1, 2 or 5. Test wells contain larger numbers per well, typically 10⁴ or 2 × 10⁴, as only rare cells

will be capable of growth in TG. The wells also contain a source of growth factor (interleukin-2), PHA, and irradiated accessory cells. Cloning efficiencies are determined from the Poisson relationship P₀ = e^-x, where x is the average number of clonable cells per well. The ratio of the cloning efficiency in the presence of TG to the cloning efficiency in its absence gives the mutant frequency (MF). The term "mutant" is used because the genetic basis of the resistant cells can be demonstrated. Growing colonies can be isolated and propagated in vitro for molecular and other studies.

Initial estimates of frequencies of hprt mutations in humans came from the autoradiographic short-term assay (Strauss and Albertini, 1977, 1979). These reports, mostly from our laboratory, gave VF values in the range of 10^-4, which were clearly in error. The early studies did not recognize the fact that infrequent cycling T cells in the peripheral blood may continue DNA synthesis in vitro in the presence of TG, even if non-mutant. Although these "phenocopies" are eventually killed by TG, they do incorporate label during the early hours of culture in the short-term assays and thus are scored as variants. As outlined elsewhere, cryopreservation removes this effect (Albertini et al., 1981). Reports of the last several years are based on methods that remove the phenocopy effect and do reflect the numbers of true variants (Albertini et al., 1981, 1988; Ammenheuser et al., 1988, 1991, 1994; Ostrosky-Wegman et al., 1988, 1990; Montero et al., 1991; Ward et al., 1994). However, the initial reports demonstrate the potential for error inherent in phenotypic assays unless special precautions are taken.

Table 16.1 summarizes typical quantitative VF and MF values obtained in studies in several laboratories over the last decade. Mean frequencies are remarkably similar for young adults whether determined by short-term or by cloning assay. This does not necessarily mean that these two kinds of measurement detect exactly the same thing. It is quite possible that different subpopulations of T cells are being assayed by each method, that the different endpoints (i.e., single round of DNA synthesis versus continual clonal growth) reflect different kinds of mutants, or that sibling mutants (see below) are represented differently in the two kinds of assay. Despite this, however, both the short-term and the cloning assays show a reproducible age effect in that VF and MF both increase approximately 2–5% per year of life, at least after the teen-age years (Albertini et al., 1988; Robinson et al., 1994). Both kinds of assays have been used to determine MF values during fetal life as reflected in placental cord blood samples (McGinniss et al., 1990; Ammenheuser et al., 1994; Manchester et al., 1995). These are about tenfold lower than what is observed for young adults. The cloning assay also has been used to carefully determine hprt MF values in normal children (Finette et al., 1994). These frequencies rise rather steeply until approximately age 16, after which the 2–5% per year rise with age obtains. The several factors that affect background hprt mutant frequencies using the cloning assay have recently been reviewed and

analyzed in detail with results from four laboratories (Cole and Skopek, 1994; Robinson et al., 1994). As reflected in Table 16.1, the various inherited human DNA repair deficiency states show the expected elevations in in vivo VF or MF values that most likely reflect an increased rate of mutation.

A major stimulus for developing assays of in vivo somatic mutations in humans is the desirability of monitoring for environmental mutagenesis. Several studies of populations exposed to ionizing radiation or to chemical mutagens have been conducted by laboratories worldwide (Seifert et al., 1987; Ammenheuser et al., 1988, 1991; Messing et al., 1989; Nicklas et al., 1990, 1991a; Branda et al., 1991; Bridges et al., 1991; Natarajan et al., 1991; Tates et al., 1991; Perera et al., 1992, 1993, 1994; Ward et al., 1994). Although some have been negative, reflecting either a true lack of mutations or insensitivity of the assays, numerous studies have been positive, indicating that these measurements do detect exogenously induced gene mutations. The issue of sensitivity is now being addressed by field studies that measure hprt mutations, which are clearly biomarkers of effect, in the same individuals in whom other biomarkers, such as those reflecting exposure, are being assessed. It is only through such studies that the sensitivities and ultimate use of these assays for human genotoxicity monitoring can be established.

Three examples of monitoring humans for increases in hprt mutation frequency as a result of exposure to a mutagenic agent will be summarized here. The first is a study of patients receiving ¹³¹I conjugated antibodies for cancer therapy (Nicklas et al., 1990, 1991a). A single treatment or multiple treatments resulted in significant increases in hprt MF and these elevated MF persisted for at least 24 months after treatment. In addition, the increases in MF were proportional to the amount of radiation received as a single dose. The second example is a study of foundry workers exposed primarily to polyaromatic hydrocarbons (Perera et al., 1993, 1994). There was a significant correlation between hprt MF and DNA adducts. This was the first report linking in vivo exposure to a mutagenic agent with the level of DNA adducts and the frequency of somatic cell mutation. The third example is a study of ethylene-oxide-exposed workers (Tates et al., 1991). Two populations that differed in exposure level were studied. While increases in sister chromatid exchanges (SCE) were found in both groups, increases in hprt MF were found only in the higher exposure group. These results demonstrate the difference in sensitivity between a biomarker of exposure (SCE) and a biomarker of effect (hprt). The level of detection of hprt mutations is being actively studied in a variety of human populations exposed to mutagenic agents.

Although sensitivity is not considered in this paper, the biology of the hprt mutational response in T cells is relevant to the interpretation of recent attempts to detect radiation-induced somatic mutations in atomic bomb survivors (Hakoda et al., 1988a, b, 1989; Hirai et al., 1996). Interpretation must include considerations of the survival of mutants and the difference between mutants and mutations. Most in vivo hprt mutations in T cells in adults arise in the peripheral lymphoid tissue rather than in the true stem cell compartment. In fact, without thymic function which

is largely absent in adults, T-cell maturation cannot occur. This plus empirical evidence indicate that the lifespan of T-cell mutants will be less than the life of the individual and, for some, may be a period of months to years (Ammenheuser et al., 1988). This can be advantageous, depending on the purpose of a human monitoring study. However, somatic mutations in other than the stem cell compartments are not useful for detecting a mutational response that occurred more than half of a human lifetime in the past. It is in this light that the reports of hprt mutations in the T cells of atomic bomb survivors should be interpreted. The slight increase in MFs found in the cohort of individuals with high radiation exposures compared to non-exposed controls has been interpreted as relative insensitivity of the hprt reporter gene to ionizing radiation. In contrast, the response to acute ionizing radiation in the form of radiotherapy of cancer is quite robust (Nicklas et al., 1990, 1991a; Sala-Trepat et al., 1990; Ammenheuser et al., 1991). These superficially discordant observations are easily reconciled when it is recognized that mutants are not the same as mutations. The infrequent elevations in MF in individual high-exposure survivors probably result from clonal amplifications of the mutants that descend from stem cell mutations in persons who were young at the time of the bomb blast. Traces of most of the mutations induced by the bombs which occurred in the peripheral lymphocyte tissues of adults have long disappeared in the 40+ years between exposure and assay. Therefore, there probably is no simple causal relationship between the slope of the MF values in the atomic bomb survivors and their radiation exposures. More will be said below of mutants versus mutations.

The ability to isolate and propagate mutant colonies from cloning assays allows characterization of the underlying hprt mutations. A substantial armamentarium of methods is now available for these molecular analyses (Table 16.2). Mutation events ranging from single-base changes to deletions and translocations of the hprt gene can be identified and mutational spectra under a variety of circumstances described. However, a precise distinction between mutants and mutations, briefly considered above for its quantitative implications, is even more important for defining mutational spectra, which describe specific kinds of mutations among all mutations. Molecular characterization of another set of genes in mutant isolates [i.e., the T-cell receptor (TCR) genes] is used to define in vivo clonality, sibling mutants, and primary independent hprt mutational events.

Cells of the immune system recognize the universe of antigens via their surface receptors—that is, surface immunoglobulins (Ig) on B lymphocytes and TCRs on T lymphocytes. There is an enormous diversity of these antigen-specific receptors which is generated at the somatic level by rearrangements of germ-line encoded

variable (V), diversity (D), junctional (J), and constant (C) regions of the Ig and TCR genes. This process of rearrangement proceeds in an orderly fashion during lymphocyte ontogeny and is mediated by an enzyme activity system termed V(D)J recombinase. DNA cleavage by V(D)J recombinase is directed by highly conserved consensus sequences consisting of a heptamer and nonamer separated

by 12 or 23 bases. V(D)J recombinase–mediated rearrangements are characterized by certain hallmarks in the junctional regions—namely, nibbling back from the point of incision, the presence of palindromic or P nucleotides, and the insertion of non-templated bases, presumably by terminal deoxynucleotidyl transferase (TdT) activity. These all contribute to the hypervariability of the junctional regions which confer the requisite diversity for antigen binding. At the molecular level, the Ig or TCR gene rearrangement patterns in B or T cells, respectively, constitute molecular signatures of clonality. For the T cells, there are four TCR genes: the α and β genes, located at chromosome positions 14q11 and 7q35, respectively, which are co-expressed on the majority of cells; and the γ and δ genes, located at chromosome positions 7p15 and 14q11, respectively, which are co-expressed on the remainder.

The diversity of TCR gene rearrangements is so vast that identity for all rearranged genes in two or more cells or isolates from an individual indicates their in vivo clonal lineage dating from the time of these rearrangements. Thus, two or more hprt mutant isolates with identical TCR gene rearrangements are clonal descendants of the same T-cell progenitor. If the hprt changes are also identical at the molecular level, which is by far the usual case, these are sibling mutants deriving from the same in vivo hprt mutation. The decidedly unusual occurrence of different hprt changes in isolates with identical TCR gene rearrangements may indicate an in vivo clone with a mutator phenotype. In any event, this observation indicates that the hprt mutational events arose in vivo after the TCR gene rearrangements, that is, in a post-thymic cell in the periphery. The much more common finding of one or more hprt mutant isolates with identical hprt changes plus one or more non-mutant or wild-type isolates from the same individual, all with the same TCR gene rearrangements, indicates the same post-thymic peripheral origin of the in vivo hprt mutation. It is from such observations that we know that most in vivo hprt T-cell mutations arise in the peripheral lymphoid tissue rather than in the stem cell compartment in adults.

It should be obvious that analyses of TCR gene rearrangements among hprt mutant isolates, by identifying sibling mutants, provides a basis for correcting the in vivo mutant frequencies determined by cloning assays to in vivo mutation frequencies. When this was analyzed for 413 wild-type and 1,736 hprt mutant isolates from 58 individuals, in vivo clonality occurred among mutants but not wild-type isolates (O'Neill et al., 1994). Thirty-five (60%) individuals had at least one instance of clonality among their mutant isolates. Despite this, adjusting the measured MF values by clonality to obtain the mutation frequency only had a major effect on nine samples, all of which had MF values of > 40 × 10^-6. Table 16.3 gives illustrative results from this study. These differences, in vivo clonality among individuals may explain the large inter-individual variability in MF values, even after age corrections, compared to the relatively low level Poisson variability seen for intra-individual MF values determined by repeat assays (Robinson et al., 1994).

Molecular analyses at differing levels of resolution have now given results on thousands of background (''spontaneous") human in vivo arising hprt T-cell mutants (reviewed in Albertini et al., 1990; Cariello and Skopek, 1993; and Cole and Skopek, 1994). Our laboratory alone has analyzed approximately 1,600 isolates by Southern blot and, more recently, several hundred more by multiplex PCR of genomic DNA and by RT-PCR and sequencing of the cDNA product. Many laboratories are analyzing hprt T-cell mutants, and a computerized database of published results is available to all investigators in this field (Cariello et al., 1992; Cariello, 1994). The following is based on these sources.

The molecular spectrum for background in vivo hprt mutations in human T lymphocytes includes approximately 15% with gross structural alterations such as deletions, insertions or other rearrangements, and 85% with "point mutations" (Nicklas et al., 1989). The gross changes were defined primarily by Southern blot analyses and therefore do not include most involving less than 300 basepairs. The proportion of gross alterations when defined by multiplex PCR is somewhat smaller, probably because generally only deletion events are identified. "Point mutations" are all other kinds of events, not simply base changes. It must be emphasized that this overview applies only to the background mutants in adults. Studies of background mutants in placental cord blood, which reflect mutations arising in the fetus, show approximately 75-85% with large structural alterations, often of one kind, with the remainder being the smaller mutations. This reversal of the adult pattern persists in young children until approximately age five (Finette et al., 1996). The discussion of carcinogenic mutagenic mechanisms below will consider these fetal and childhood mutations in detail.

Another way to characterize the human background in vivo mutation spectrum is to concentrate on the "point mutations." Although data are available from several sources, those in the computerized database essentially reflect the current findings. Table 16.4 gives the results on 254 mutations. Of these, 81 or 32% are base substitutions in coding regions, with somewhat more transversions than transitions, and 19 or 7% are base substitutions in introns that affect splicing. The most frequent class of change is base substitution (81/165 for sequence-defined mutations) followed by frameshifts and small genomic deletions (23/165 each). The remaining 89 mutations have genomically undefined alterations in cDNA ("splice" mutations). Several investigators are describing sites of increased mutation ("hot spots"), but these results are still preliminary. However, a high frequency of both transitions and transversions at G₁₉₇ has been reported by several laboratories (Cariello and Skopek, 1993).

A major reason for characterizing the background mutational spectrum at hprt is for comparison with mutational spectra determined in individuals who have been exposed to environmental mutagens. The rationale is that different mutagens or classes of mutagens should induce characteristic mutational changes which then can be used to define the nature of the exposure. In this sense, hprt mutations

serve as restricted biomarkers of exposure, providing specificity by defining the offending mutagen. Clearly, exposures of human fibroblasts to agents such as benz[a]pyrene in vitro result in an hprt mutational spectrum characterized by G->T transversions (R.H. Chen et al., 1990). It is significant that mutations in the p53 gene in lung cancers from smokers but not from nonsmokers are also characterized by such changes (Suzuki et al., 1992; Hollstein et al., 1994). As regards hprt, the early results from studies of chemical exposures in humans have been mixed, probably because insufficient numbers of mutants have been studied (Vrieling et al., 1992; Burkhardt-Schultz et al., 1993; Cole and Skopek, 1994). By contrast, hprt mutations in humans exposed to ionizing radiation clearly show a spectrum that becomes increasingly dominated by large alterations such as deletions as a function of radiation dose (Nicklas et al., 1990, 1991a). Table 16.5 gives the overall frequency of large-alteration hprt mutations in cancer patients receiving irradiation from an internal ¹³¹I emitter, as compared to the background frequencies of such events in normals and in pre-treatment patients, and to the frequency of such mutations in human T lymphocytes exposed in vitro to 300 cGy external-beam acute -irradiation. Clearly, an induced mutational spectrum can be differentiated from the background spectrum, indicating that at least some degree of specificity regarding mutagen exposures can be achieved.

Although the original reason for defining in vivo hprt mutational spectra was to provide specificity for human monitoring, another, perhaps more important reason, is emerging. The original reason viewed hprt mutations as measures of exposure. However, these mutations really are biomarkers of effect, and changes in their frequencies and kinds may have a human health significance beyond simply reflecting mutagen exposures. Hprt or other specific reporter gene mutations may identify

similar events occurring elsewhere in the genome, some with pathogenic consequences. If reporter gene mutations are surrogates for mutations in cancer-relevant genes, increases in frequencies may indicate increases in cancer risk. For this to be true, mutagenic mechanisms with carcinogenic potential must be recorded in the reporter gene. The emerging reason for defining in vivo hprt mutational spectra therefore is to determine if mutations here, measured in a gene and tissue of convenience, are valid surrogates for cancer-causing mutations arising in target tissues.

Hprt cannot undergo the homologous somatic recombination as often seen in cancers because of its X-chromosomal location. It does, however, capture a variety of other carcinogenic mutagenic mechanisms. Large deletions are seen as well as translocations (Nicklas et al., 1989). These are common in tumors. Almost 25% of internal hprt deletions have breakpoints in DNA sequences with high homology to topoisomerase II consensus cleavage sequences (Rainville et al., 1995). Similar chromosome breakpoints are often seen in leukemia. Hprt even undergoes fusions with downstream elements resulting in fusion transcripts (Lippert et al., 1997)—an occurrence frequently encountered in cancer. All of the above are general mutagenic mechanisms.

There is rather striking evidence that hprt also captures a much more specific carcinogenic mutagenic mechanism. Human lymphoid malignancies frequently show nonrandom chromosome rearrangements with one of the breakpoints near one of the Ig genes in B-cell malignancies or one of the TCR genes in T-cell malignancies (Finger et al., 1986; Boehm and Rabbits, 1989; Tycko and Sklar, 1990; Breit et al., 1993). The precise site of this breakpoint is often the heptamer-nonamer consensus recognition signal sequence (RSS) that directs canonical rearrangements of these genes. The other breakpoint required to produce the characteristic chromosome translocation, which is usually in the region of an oncogene, is often in a cryptic consensus heptamer. Junctional regions of the new, translocated chromosome also frequently bear the hallmarks of a V(D)J-recombinase-mediated event, so named because of the enzyme activity that directs the rearrangement of these characteristic regions of the Ig and TCR genes, as described above. Therefore, this specific maturational Ig and TCR gene rearrangement mechanism frequently proceeds illegitimately with carcinogenic consequences.

A characteristic translocation t(1;14)(p32;q11) occurs in approximately 3% of all childhood T-cell acute lymphoblastic leukemias (T-ALL) (Q. Chen et al., 1990; Bernard et al., 1991). In the leukemic cells, most of the breakpoints on chromosome 14 cluster in the Dδ-Jδ region of the TCR δ locus, while those on chromosome 1 are in a 1kb 5' region of the so called tal-1 (for T-cell acute leukemia) gene. The tal-1 gene codes for a protein thought to be important in early hematopoietic cells. Its disregulation by the t(1;14) translocation probably produces an oncogene function in the genesis of T-ALL.

In an additional 20-30% of childhood T-ALL, a submicroscopic deletion of 90kb of the tal-1 5' region juxtaposes its coding sequences to the first non-coding

exon of an upstream sil (for stem cell leukemia interrupting locus) gene (Aplan et al., 1990, 1991; Brown et al., 1990; Breit et al., 1993). This puts the fused sil-tal-1 gene under the transcriptional control of the sil gene promoter (Bernard et al., 1991; Aplan et al., 1991, 1992). Again, disregulation of tal-1 with leukemogenic consequences occurs. Importantly, the 5' breakpoint of this deletion (in the sil gene) occurs in only one V(D)J recombinase heptamer recognition sequence, while the 3' breakpoint occurs in one of four locations, each however in a V(D)J heptamer-nonamer recognition sequence (Bernard et al., 1991; Aplan et al., 1992; Breit et al., 1993). The junctional sequences of these deletions show the hallmarks of V(D)J recombinase activity (Aplan et al., 1991; Bernard et al., 1991; Breit et al., 1993). Therefore, in both the submicroscopic deletion and the t(1;14) translocation, illegitimate V(D)J recombinase activity produces a leukemogenic somatic mutation.

As implied above, the background hprt mutational spectrum in the human fetus, as determined in placental cord blood T cells, differs markedly from the background spectrum found in adults. In the fetus, up to 85% of the hprt mutations show gross structural alterations of the gene, with up to 50% of these being a seemingly identical exon 2-3 deletion (McGinniss et al., 1989). Because these mutations arise in the fetus at a time when T lymphocytes are undergoing thymic differentiation, it seemed reasonable that some aberration of the TCR gene rearrangement mechanism was involved. This, in fact, was what was found by sequence analysis of the breakpoint and junctional regions of 18 hprt deletion mutants isolated from 13 newborns. All mutants have a single 5' breakpoint in intron 1 at base 21971. However, three different 3' breakpoints in intron 3 are observed in the group of 18 mutants, occurring either at base 22250 (15 instances), base 20156 (two instances), or base 22570 (one instance) (Fuscoe et al., 1991). The breakpoint site in intron 1 is in a V(D)J recombinase heptamer, while each of the intron 3 breakpoint sites is in a V(D)J recombinase heptamer-nonamer RSS. Therefore, there is complete analogy between the breakpoint sites for the hprt mutations and those for the sil-tal fusion gene in childhood T-ALL. All of the hprt mutants showed the characteristics of a V(D)J-mediated recombination, i.e., heptamer or heptamer-nonamer RSS, nibbling back, P-nucleotides, and non-templated additions.

More recent studies have indicated that the V(D)J recombinase–mediated hprt mutations remain frequent during childhood, being the most common single kind of spontaneous mutational event in normal children up to age five (Finette, 1996). The age-frequency distribution of V(D)J recombinase–mediated hprt mutations shows a striking similarity to the age-frequency distribution of childhood acute T-ALL. Therefore, this pathogenic, mutagenic mechanism is captured by hprt as a harmless mistake, although it can give rise to serious disease when it occurs in other genetic regions. The hprt gene clearly is a valid surrogate for at least those mutational events that give rise to the lymphoid malignancies in childhood.

Somatic mutations in the hprt gene arise as harmless errors in humans and can be measured in peripheral blood T lymphocytes. As the mutant cells are not consumed in the cloning assay for these events, the underlying mutations can be analyzed at the molecular level. Several investigations have now shown that exposures to both chemical and physical mutagens increase hprt MF values; the sensitivity of this biomarker for detecting such exposures is being defined. The observation that the T-cell hprt mutations arise in adults in the peripheral lymphoid tissue rather than in the stem cell compartment indicates that the mutant cells will have a lifetime less than the lifetime of the individual, probably best measured in months or years. In an attempt to provide specificity for assessing mutagen exposures, molecular mutational spectra are being described for both the spontaneous mutations and for those associated with a variety of genotoxic agents. It clearly has been shown that ionizing radiation shows a characteristic mutational spectrum, inducing a high frequency of hprt mutations with gross structural alterations. Definition of mutational spectra mandates identification of in vivo clonality of mutants. Characterization of TCR gene rearrangement patterns allows this identification, and permits calculation of mutation frequencies from measured MF values. Hprt molecular mutational spectra are being defined to provide specificity for human monitoring and, more recently, for determining if mutagenic mechanisms with carcinogenic potential are reported by this gene. Studies in newborns and children have shown that the most frequently arising spontaneous hprt mutation in this age group is a large deletion mediated by the V(D)J recombinase system. This is also the mutational mechanism responsible for a wide variety of lymphoid malignancies, including T-ALL of childhood. Hprt, therefore, captures precisely this mutagenic mechanism. It is perhaps significant that this particular mutation is reported in T lymphocytes that are the target cells for the malignancies of concern. The question regarding the relevance of reporter mutations in tissues of convenience for predicting pathological events in tissues of concern remains open.

This latter point too is being addressed by studies of in vivo hprt mutations in humans. Myeloid stem cells (CD34+) from human bone marrow or from peripheral blood, especially from placental cord blood, may be cloned in methylcellulose (Fauser and Messner, 1978). Differentiated myeloid and erythroid cell colonies are recognized after several days in the presence of critical cytokines. It is possible to incorporate TG in the methylcellulose and select for TG-resistant colony growth and differentiation. Such colonies are present and hprt mutant frequencies can be determined. Analyses of DNA extracted from these colonies by either RT-PCR or multiplex PCR followed by sequencing have demonstrated that they are true hprt mutants. As for the T-cell system, a molecular mutational spectrum is being defined. There will soon be a myeloid stem cell system analogous to the T-cell system for identifying, quantifying, and characterizing hprt mutations in this target tissue. The validity of hprt as a surrogate gene for detecting matched mechanisms that give rise to the more common adult hematopoietic malignancy can be assessed.

This chapter has reviewed the experience to date with hprt mutations in humans. The different assay systems are being exploited practically for mutagenicity monitoring. There is every expectation that such monitoring can identify human environmental health hazards before they cause disease. The cells recovered from the mutagenicity studies are providing material for basic studies, which are beginning to define mechanisms in humans that influence the mutational process itself.

We are grateful for the technical support of Linda M. Sullivan and Timothy C. Hunter and the secretarial assistance of Inge Gobel. This work was supported by the US Department of Energy (FG0287ER60502), the National Institutes of Health (5PO1-ES05294-05), and the American Cancer Society (CB-45 and CN-141). DOE and NIH support does not constitute endorsement of the views expressed.