Through a Glass, Darkly:¹ Reflections of Mutation From lacI Transgenic Mice

Corresponding author: Gregory R. Stuart, Centre for Environmental Health and the Department of Biology, University of Victoria, P.O. Box 3020 STN CSC, Victoria, BC V8W 3N5, Canada., gstuart{at}uvic.ca (E-mail)

Communicating editor: R. MICHOD

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

The study of mutational frequency (Mf) and specificity in aging Big Blue lacI transgenic mice provides a unique opportunity to determine mutation rates (MR) in vivo in different tissues. We found that MR are not static, but rather, vary with the age or developmental stage of the tissue. Although Mf increase more rapidly early in life, MR are actually lower in younger animals than in older animals. For example, we estimate that the changes in Mf are 4.9 x 10^-8 and 1.1 x 10^-8 mutations/base pair/month in the livers of younger mice (<1.5 months old) and older mice (1.5 months old), respectively (a 4-fold decrease), and that the MR are 3.9 x 10^-9 and 1.3 x 10^-7 mutations/base pair/cell division, respectively (~30-fold increase). These data also permit an estimate of the MR of GC AT transitions occurring at 5'-CpG-3' (CpG) dinucleotide sequences. Subsequently, the contribution of these transitions to age-related demethylation of genomic DNA can be evaluated. Finally, to better understand the origin of observed Mf, we consider the contribution of various factors, including DNA damage and repair, by constructing a descriptive mutational model. We then apply this model to estimate the efficiency of repair of deaminated 5-methylcytosine nucleosides occurring at CpG dinucleotide sequences, as well as the influence of the Msh2^-/- DNA repair defect on overall DNA repair efficiency in Big Blue mice. We conclude that even slight changes in DNA repair efficiency could lead to significant increases in mutation frequencies, potentially contributing significantly to human pathogenesis, including cancer.

THE use of transgenic rodents has greatly facilitated in vivo studies of the mechanisms of mutation, DNA repair, and carcinogenesis (KOHLER et al. 1991 ; MIRSALIS et al. 1994 ; MIRSALIS 1995 ; DE BOER and GLICKMAN 1998 ). While transgenic rodent mutagenicity assays provide a practical approach to the study of genotoxicity, they offer the additional advantage of providing novel insights into mechanisms of mutation. These observations often include unexplained or unexpected responses that reflect the true biological complexity inherent in mammalian systems and challenge our current understanding of these systems. Some recent examples of unpredicted results arising from transgenic rodent mutational assays include the following: (1) an apparent lack of correlation of 2-amino-1-methyl-6-phenylimidazo [4,5-b]pyridineor 2-amino-3,4-dimethylimidazo[4,5-f] quinoline-induced tissue adduct levels and mutagenicity with carcinogenicity in target tissues (OKONOGI et al. 1997 ; OCHIAI et al. 1998 ); (2) a higher-than-expected spontaneous mutation frequency (Mf) in mrkII transgenic mice encoding a lacI gene with reduced 5'-CpG-3' (CpG) content, due to increased frequency of GC AT transition mutations at the few remaining CpG sequences (SKOPEK et al. 1998 ); (3) a decline in lacI Mf during spermatogenesis in younger but not older mice (WALTER et al. 1998 ); and (4) chemoprotection by 2,3,7,8-tetrachlorodibenzo-p-dioxin against aflatoxin B₁-induced mutation in female but not male lacI transgenic rats (A. S. THORNTON-GLICKMAN, Y. ODA, G. R. STUART, J. HOLCROFT, J. G. DE BOER and B. W. GLICKMAN, unpublished results).

Mutations accumulate in a tissue-specific manner during the life span of an organism, contributing significantly to the risk of diseases including cancer. The study of the origin, frequency, and especially the specificity of mutation is a necessary first step toward understanding the fundamental molecular mechanisms responsible for mutation. We recently reported the changes in spontaneous Mf and mutational spectra (MS) with age in the lacI transgene recovered from liver, bladder, and brain of Big Blue mice (STUART et al. 2000 ). Those data enabled us to combine Mf data from aging mice with estimates of cellular turnover to calculate here, for the first time, mutation rates (MR) in young and adult animals. The data show that MR are not static, as might be inferred on the basis of numerous literature reports that quote a single value but, rather, appear to vary as a function of developmental age or level of proliferative activity of the tissue.

The validity of these estimates of MR is strengthened by the use of DNA sequencing to correct for nonindependent mutational events (i.e., clonal expansions) and the use of the well-characterized lacI transgene as the mutational target. DNA sequence analysis from this laboratory of nearly 20,000 lacI mutants recovered from Big Blue rodents indicates that ~410 of the 1080 (38%) nucleotides that encode the lacI gene may be recovered as mutations (http://eden.ceh.uvic.ca/sites.htm; B. W. GLICKMAN and J. G. DE BOER, unpublished data). These data enable the calculation of Mf per mutable nucleotide.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Experimental data:
Please refer to STUART et al. 2000 for detailed descriptions of the experimental treatments and sources of the Mf data. For convenience, the relevant Mf for liver and brain are provided in a footnote to Table 1.

Definitions:
Mutation frequency is defined as the prevalence of mutation in a gene at a specified age, corrected by DNA sequence analysis for possible nonindependent mutational events due to clonal expansion (DE BOER et al. 1996 , DE BOER et al. 1997 ). [Note that uncorrected Mf data provide a mutant frequency (MF).] Mutation rates (MR) are best described as the change in Mf (Mf) per cell division, although at times (e.g., for convenience of comparison to literature values when discussing MR of deaminations of 5-methylcytosine), MR are expressed as the Mf per unit time. Accordingly, increases in lacI Mf with time are better described as Mf rather than MR. A mutational spectrum (MS) describes the nature, nucleotide position, and frequency of mutations that have occurred within a gene (or a defined DNA sequence). Demethylation is used generically to describe the loss or disappearance of 5-methylcytosine (5MC) from mammalian genomic DNA that can occur by spontaneous hydrolytic deamination of 5MC or enzymatically by DNA (cytosine-5)-methyltransferase or DNA demethylase (refer to RESULTS AND DISCUSSION).

	RESULTS AND DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

Use of mutation frequency rather than mutant frequency:
In this article we have chosen to use Mf rather than MF to calculate MR, as the aging data described in STUART et al. 2000 (upon which the following discussion is based) reported Mf. Also, Mf may be more representative of the underlying, initial mutational events that contribute to MR, as opposed to MF that include biological processes (such as clonal expansion) that follow the initial mutational event. Finally, it is noted that the corrections from MF to Mf (STUART et al. 2000 ) were generally rather conservative; the clonal corrections were 9, 20, 14, 17, and 12% in liver and 38, 16, 15, 10, and 11% in brain aged 1.5, 6, 12, 18, and 25 months, respectively. Nevertheless, it is acknowledged that some laboratories prefer that all mutants be considered (e.g., DROST and LEE 1995 ; DRAKE et al. 1998 ; THOMPSON et al. 1998 ; HEDDLE 1999 ).

Liver growth and development:
Before MR can be calculated, it is necessary to estimate the number of cell divisions (in some cases, more specifically the number of rounds of DNA replication) that have occurred. Two-week-old CBA/C57BL mouse liver contains ~0.8 x 10⁸ hepatocytes, increasing 1.5-fold to ~1.2 x 10⁸ cells in 3-month-old mice (BRODSKY and URYVAEVA 1985 ). Noting that the liver of a young adult mouse weighs ~1.5 g, these numbers agree reasonably well with those provided by BUETOW 1985 , who reported that a 2-month-old male mouse liver contains ~1.16 x 10⁸ cells/cm³, with this number decreasing slightly to 0.98 x 10⁸ and 0.88 x 10⁸ cells/cm³ at ages 12 and 24 months, respectively. Cell proliferation decreases significantly, ~3.3-fold, in the liver of male mice from ages 10–13 weeks (2.5–3.2 months) (ELDRIDGE and GOLDSWORTHY 1996 ). Although the number of cells in the postnatal liver reach a plateau, DNA synthesis continues at a reduced rate throughout adulthood, resulting in an age-related increase in mean polyploidy (CARRIERE 1969 ; BRODSKY and URYVAEVA 1977 , BRODSKY and URYVAEVA 1985 ). Thus, mean DNA ploidy levels in mouse liver double from ages 1 week to 1 month, and thereafter increase steadily, doubling again by 24 months of age (BRODSKY and URYVAEVA 1977 ). This increase in liver polyploidy is accompanied by increase in liver weight but not cell number.

Calculations of MR should therefore consider DNA replication resulting from polyploidization in addition to that contributed by cellular proliferation, since mutations are established (i.e., fixed, as in fixation of mutations) during DNA replication (STUART et al. 2000 ). From the age of 2–3 weeks, it has been reported that each liver cell in the mouse enters the mitotic cycle from one to six times (three on the average), resulting in an 8- to 10-fold increase in liver mass and about a 3-fold increase in the number of cells. Mature hepatocytes are fully differentiated, self-maintaining cells with low proliferative rate and low, if any, rate of cell elimination from the population during the life of the mouse. The liver cells in newborn mice are diploid but polyploidy levels increase in young animals (URYVAEVA 1981 ). In adult mice most, if not all, mitoses are polyploid.

The relatively small increases in the postnatal number of liver cells is reflected by the slow growth rate of this tissue. During normal growth, hepatocytes rarely divide—even in young rapidly growing animals, 2–12 days pass between successive mitoses, and several months in adults (SCHULTZE et al. 1978 ; URYVAEVA 1981 ). Mouse hepatocytes are regarded generally as having a turnover time of 480–620 days (CAMERON 1971 ).

Brain growth and development:
In the mouse, proliferative activity associated with brain development appears to be largely completed by 3–4 weeks after birth (KORR 1980 ). Accordingly, the number of cells in the whole brain of the mouse stabilizes at 0.85 x 10⁸ nuclei (cells) by 1 month of age, with no significant change in this value up to 36 months of age (FRANKS et al. 1974 ; BUETOW 1985 ). The adult mouse brain is practically mitotically quiescent, except for a small population of glial cells (KORR 1980 ; BOWMAN 1985 ). DNA polyploidy levels are also known to remain low, mostly diploid, in the adult brain (WINICK et al. 1972 ).

Liver mutation frequency and rate:
The lacI Mf at conception (0.7 months before birth) is zero, since an inherited mutation in any one of the estimated 40 lacI transgenes present on mouse chromosome 4 (DYCAICO et al. 1994 ) would result in a spontaneous Mf of 2.5 x 10^-2, >500-fold higher than the spontaneous Mf of ~4–5 x 10^-5 normally observed at age 1.5 months (HEDDLE 1998 ). Therefore, we were able to calculate that the increase in Mf (Mf) during the 2.2-month period from conception (Mf of zero) to age 1.5 months postnatal (4.3 x 10^-5) was 2.0 x 10^-5 mut/lacI transgene/month. Since ~410 lacI nucleotide positions result in mutants recoverable in the Big Blue assay, the Mf in animals <1.5 months old was therefore 4.9 x 10^-8 mutations/bp/month.

It has been noted previously that spontaneous mutations in somatic cells appear to accumulate steadily throughout adult life (CURTIS 1971 ). This result was confirmed in our mutational studies of Mf in liver of mice that were >1.5 months old (STUART et al. 2000 ). The least squares plot for the Mf in liver of mice >1.5 months old (Figure 1 in STUART et al. 2000 ) gave a slope of 0.45 x 10^-5 mut/lacI transgene/month (R = 0.987). Dividing by the 410 lacI nucleotide positions recoverable as mutants in the Big Blue assay, the Mf in animals >1.5 months old was therefore 1.1 x 10^-8 mut/bp/month. On the basis of these values, the Mf in liver in younger mice (<1.5 months old) increased approximately fourfold faster than in mice >1.5 months old.

We also calculated the MR (the Mf per cell division) by using the estimates of cellular proliferation and DNA replication provided above. Since adult liver proliferates slowly but mean polyploidy levels increase, the Mf with age in this tissue results primarily from DNA replication in nondividing cells, resulting in fixation of DNA lesions (during translesion bypass) or DNA mispairs as mutations (STUART et al. 2000 ). Therefore, to facilitate both the calculation of MR in adult liver and to simplify the discussion that follows, polyploidizing DNA replications were considered to be functionally equivalent to cellular divisions.

Since the livers of younger mice, aged 1.5–2 months, contain ~1.3 x 10⁸ cells, we estimated that ~27 cell divisions (DNA replications) had occurred (2²⁷ = 1.3 x 10⁸) during the period from conception to age 1.5 months. Dividing the Mf of 4.3 x 10^-5 at age 1.5 months by 410 mutable lacI nucleotide positions and 27 cell divisions gave an MR of ~3.9 x 10^-9 mut/bp/division in mice aged up to 1.5 months. The estimate of the number of cell divisions assumed that cell death (for example, due to apoptosis) was negligible. However, it may be noted that the magnitude of this MR estimate is rather modestly affected over a wide range of cell divisions; for example, using half as many (13) and twice as many (54) cell divisions resulted only in an approximately twofold change upward or downward, respectively, in the calculated MR.

We estimated the extent of cellular proliferation (DNA replication) in livers older than 1.5 months as follows. First, the relatively steady increase in Mf (Mf) with age in the adult liver suggested that the balance between DNA replication (fixation of DNA mispairs or DNA lesions as mutations) and DNA repair was maintained during this period. Second, BRODSKY and URYVAEVA 1977 found that mean polyploidy levels in mouse liver increased only about twofold from ages 1 to 24 months. Third, the turnover time for hepatocytes is 480–600 days (1.3–1.6 years; CAMERON 1971 ), indicating that the population of liver cells is replaced ~1.5 times over 2 years. These observations collectively suggest that from ages 1.5 to 25 months there are probably less than two cell divisions (or DNA replications) on average per cell.

In liver from mice aged 1.5–25 months, by assuming that two cell divisions (DNA replications) had occurred, we calculated the MR to be ~1.3 x 10^-7 mut/bp/div[= (1.1 x 10^-8 mut/bp/month)(23.5 months)/(2 div)]. Assuming that three cell divisions had occurred only reduced the MR 1.5-fold, to 8.6 x 10^-8 mut/bp/div.

For convenience, the liver Mf and MR values calculated above are summarized in Table 1.

Brain mutation frequency and rate:
Since cellular proliferation in the mouse brain is essentially completed by 1 month of age, resulting in 0.85 x 10⁸ cells, it can be estimated that ~27 cell divisions have occurred (2²⁶ = 0.67 x 10⁸; 2²⁷ = 1.3 x 10⁸). Using this value and the brain Mf at 1.5 months of 2.9 x 10^-5, we calculated that the MR in mouse brain from conception (-0.7 months) to age 1.5 months was 1.1 x 10^-6 mut/lacI/div, or 2.6 x 10^-9 mut/bp/div. This compares with the Mf during this period of 3.2 x 10^-8 mut/bp/month.

The small but statistically significant increase in brain Mf between ages 1.5 and 6 months (Table 1) is consistent with the known low proliferative capacity of brain tissue. The small increase of brain Mf of 1.6-fold suggests that some DNA replication had occurred; thus, it was assumed that the brain cell population probably underwent less than one doubling during this period. Therefore, by using the Mf at 6 months of 4.6 x 10^-5 and assuming one cell division, we calculated that in brain aged 1.5–6 months the Mf was 9.2 x 10^-9 mut/bp/month, and the MR during this period was ~1.7 x 10^-5 mut/lacI/div (or 4.2 x 10^-8 mut/bp/div).

Finally, since there was no significant Mf in brain greater than 6 months old, the Mf and the MR in brains older than 6 months were practically zero (not detectable). This result is consistent with the fact that adult brain is essentially mitotically quiescent, and thus the contribution to MR from proliferative mechanisms is negligible.

These brain Mf and MR data are also summarized in Table 1.

Distinguishing between Mf and MR:
Why should we distinguish between Mf and MR? Most data from mutagenicity assays report Mf (the prevalence of mutations) at a specified point in time, while most studies that discuss mutation with regard to human health or evolution report MR (change in prevalence over time). Therefore, we feel that to better understand data from model systems, including the Big Blue assay, we should also understand the nuances between Mf (the accumulated mutational burden) and MR (the rate of increase in Mf per cell division, or less preferably, unit of time).

According to the data in Table 1, the increase in Mf (Mf) in liver of younger mice (<1.5 months old) occurred more than fourfold faster than in older (1.5 month) mice. Conversely, the MR in younger mice (<1.5 month) was ~33-fold slower than in older mice (1.5 month). The simplest explanation for this seemingly paradoxical relationship between Mf and MR is that in developing tissues DNA replication probably contributes more to increases in Mf (e.g., through generation of replication-dependent DNA mismatches) than does DNA damage (nonreplication-dependent premutagenic lesions, e.g., hydrolytic deamination of 5MC). This is supported by the observation that the Mf that had accumulated during the first 2.2 months of life (i.e., from conception to 1.5 months postnatal) required an additional 10.5 months to double in frequency. Thus, in younger animals the Mf is greater than in older animals. Conversely, during the period of rapid cellular proliferation in developing mouse liver, the spontaneous mutations that occurred were relatively quickly partitioned among progeny daughter cells, resulting in a lower MR in younger animals.

Another way of describing the changes in Mf and MR with age is to note that (1) Mf appear to be proportional to cellular proliferation (Mf div); (2) MR appear to be inversely proportional to cellular proliferation [MR 1/ div; such a relationship was speculated previously by DROST and LEE 1995 in their interesting article on germline MR]; and (3) cellular proliferation varies inversely with age (div 1/ age). By substituting expression 3 into expressions 1 and 2, we see that as age increases, Mf decreases and MR increases, consistent with observed values.

The data summarized in Table 1 clearly indicate that MR in a tissue are not constant throughout the life span of an animal. Interestingly, however, for comparable developmental periods of growth the values for the Mf as well as the MR in liver and brain are remarkably similar. Whether or not this similarity is coincidental remains to be determined.

Deamination of 5-methylcytosine:
The most prevalent spontaneous mutation, >25% of all mutations in mammalian tissues, are GC AT transitions occurring at CpG sequences (STUART et al. 2000 ). These mutations are generally attributed to hydrolytic deamination of 5-methylcytosine (5MC) bases that are present at CpG sequences (COULONDRE et al. 1978 ; BIRD 1980 ; COOPER and YOUSSOUFIAN 1988 ; COOPER and KRAWCZAK 1989 ). The lacI gene contains 190 CpG sequences, considering both DNA strands (FARABAUGH 1978 ); of these sequences, 84/190 (44%) have been recovered as mutants in the Big Blue lacI assay (B. W. GLICKMAN and J. G. DE BOER, unpublished data; http://www.eden.ceh.uvic.ca/sites.htm). Using the Mf data from our aging study (STUART et al. 2000 ), we calculated the Mf of GC AT mutations at CpG sequences in liver of mice aged 1.5–25 months to be 0.19 x 10^-5 mut/lacI/month, corresponding (after dividing by 84 recoverable CpG mutations per lacI transgene) to 2.3 x 10^-8 deaminations/5MC/month.

The 5MC deamination MR determined in mouse liver was compared to those determined in vitro and in vivo in double-stranded DNA from different taxonomic groups (Table 2). Several observations are immediately apparent. First, the rate of spontaneous hydrolytic deamination of 5MC measured in vitro is amply sufficient to account for the MR observed in vivo. The very low in vivo deamination rates indicate that repair of GT mispairs must be highly efficient, perhaps >99% (YANG et al. 1996 ). Second, the significant lowering of MR from Escherichia coli to mice, apes, and humans, respectively, indicates that MR are not uniform among different taxonomic groups or evolutionary time (WILSON and JONES 1983 ; LI and TANIMURA 1987 ; WILSON et al. 1987 ; MATSUO et al. 1993 ; LI et al. 1996 ). While these generalizations are not new, the reasonably accurate determination of this rate in transgenic rodents strengthens the validity of these observations.

Caveats associated with the MR determined for deamination of 5MC in the lacI transgene include the belief that the bacterially derived lacI transgene is fully methylated (KOHLER et al. 1990 ; SCRABLE and STAMBROOK 1997 ; DE BOER and GLICKMAN 1998 ; YOU et al. 1998 ), and is therefore nontranscribed (PROVOST and SHORT 1994 ). The transgenes are stably integrated into the mouse genome as a tandem array of ~40 copies at a single locus on chromosome 4 (DYCAICO et al. 1994 ). Also, the density of CpG sequences in the lacI transgene is higher than the average density of these sequences in the mammalian genome (DE BOER and GLICKMAN 1998 ). Nevertheless, MR determined in the lacI transgene are likely to be reasonably accurate estimates of the average rate occurring throughout the mouse genome for several reasons. First, mutations in the lacI transgene are thought to be neutral and therefore confer no selective growth advantage or disadvantage to the cell. Second, the consideration of MR per mutable site (base pair) effectively normalizes the data, with respect to under- or overrepresentation of CpG sequences throughout the genome. Finally, although this conclusion is sometimes debated, it appears that DNA repair activity is not significantly different in the lacI transgene compared to endogenous mammalian genes, as similar mutational responses have been observed in the lacI transgene compared to the endogenous mouse genes Dlb-1 (TAO et al. 1993 ) and Hprt (SKOPEK et al. 1995 ; WALKER et al. 1996 ). As well, the change in the lacI spontaneous mutational spectrum observed in Msh2^-/-lacI cotransgenic mice indicates that lacI transgenes respond as predicted to changes in DNA repair function (ANDREW et al. 1997 ).

Since Mf in the lacI transgene can be ~10-fold higher than those at selected endogenous loci such as Hprt (SKOPEK et al. 1995 ), generalizations on the basis of studies involving the lacI transgene are sometimes debated. However, differences in Mf among different genes reflect, in part, differences in the relative sizes of the genes (or perhaps more specifically, the number of nucleotide positions that are recoverable as mutants), as well as the nature of the assay itself. The lacI gene is a highly sensitive mutational target, with 38% of the nucleotide positions recoverable as mutants, which undoubtedly contributes to elevated Mf. Also, as mentioned, the CpG content of the lacI transgene is somewhat higher than in endogenous loci. Nevertheless, since ~40–50% of lacI spontaneous mutations involve base substitutions at CpG dinucleotide sequences, these sites likely account for a less than twofold elevation in Mf in the lacI transgene relative to endogenous mammalian loci. Finally, we believe that the consideration of Mf (or MR) per base pair may normalize the mutational data, regardless of the locus examined.

Demethylation of DNA:
Since demethylation of genomic DNA, principally involving the disappearance of 5MC nucleotides, has been reported to occur at high frequency in aging mammalian cells (GAMA-SOSA et al. 1983 ; WILSON and JONES 1983 ; HOAL-VAN HELDEN and VAN HELDEN 1989 ; MAZIN 1994 ), we decided to determine the influence of demethylation on the Mf for deamination of 5MC and vice versa. Demethylation (we use the term generically to describe any disappearance of 5MC and, consequently, the reduction in overall DNA methylation levels) potentially occurs by three routes. (1) Hydrolytic deamination of 5MC can yield thymine directly, resulting in premutagenic GT mispairs (COULONDRE et al. 1978 ; COOPER and KRAWCZAK 1989 ). (2) Immediately after DNA replication, cytosines present in the nascent DNA strands are unmethylated. During the methylation of hemimethylated CpG sequences, DNA (cytosine-5)-methyltransferase covalently bonds at the C6 position of cytosines, greatly labilizing the amino group, resulting in deamination events before methylation at the C5 position (SHEN et al. 1992 ; STEINBERG and GORMAN 1992 ; LAIRD and JAENISCH 1996 ). Thus, cytosines at CpG sequences may be converted directly to thymines, again resulting in premutagenic GT mispairs. These GT mispairs, however they arise, are recognized and repaired with high but not absolute efficiency by thymine DNA glycosylase (BROWN and JIRICNY 1987 ; BROOKS et al. 1996 ; MARIETTA et al. 1998 ), resulting in a low but finite accumulation of GC AT transitions with each round of DNA replication. (3) DNA demethylase, a novel enzyme that specifically recognizes 5MC residues at CpG sequences in mammalian DNA, has been described recently (BHATTACHARYA et al. 1999 ; RAMCHANDANI et al. 1999 ). The product of this enzymatic reaction is a normal (i.e., nonmutagenic) GC base pair. Since 5MC residues in mammalian genomes occur most frequently at CpG sequences, we can use the mutational data from lacI transgenic rodents to evaluate the mechanism principally involved in the demethylation of aging mammalian cells.

Published estimates of the rate of demethylation of mouse and rat genomes vary greatly, even for a single species and tissue such as mouse liver (GAMA-SOSA et al. 1983 ; WILSON and JONES 1983 ; SINGHAL et al. 1987 ; WILSON et al. 1987 ; HOAL-VAN HELDEN and VAN HELDEN 1989 ; TAWA et al. 1990 ; KANUNGO and SARAN 1992 ; MAZIN 1994 , MAZIN 1995 ). The reasons for these discrepancies are not obvious but might involve one or more of the following: differences in species, age, strain, and tissues; dietary factors affecting DNA methylation levels; or choice of analytical method (DRAHOVSKY and BOEHM 1980 ; REIN et al. 1998 ). Nevertheless, the estimates of the rate of demethylation are very large; e.g., WILSON et al. 1987 determined that the rate of loss of 5MC from the genome of C56BL/6J mouse liver was 0.012% per month. The demethylation rate reported by MAZIN 1994 in mice (unspecified tissue) of 0.033% per cell per day (or 0.99%/cell/month) is ~80-fold greater than that reported by WILSON et al. 1987 . Similar demethylation rates may be inferred from the data described by HOAL-VAN HELDEN and VAN HELDEN 1989 , who reported a 46% decrease in the percentage of 5MC in liver of rats from 1 day before birth to 6 months of age.

Considering the published demethylation data and our MR for GC AT transitions at CpG sequences, what is the potential influence of demethylation on observed Mf? The mouse genome contains ~7 x 10⁹ DNA base pairs per diploid nucleus, with ~1 mol% of the cytosines in the liver of C57BL/6 mice methylated as 5MC (GAMA-SOSA et al. 1983 ; TAWA et al. 1990 ). [It should be noted that some laboratories report somewhat higher, possibly age-related, values for the 5MC content in young mouse liver. For example, WILSON et al. 1987 reported that 3% of the cytosines in 4-wk-old C57BL/6J mouse liver were present as 5MC.] Therefore, assuming that cytosines comprise about one-quarter of the nucleotides in the mouse genome with 1 mol% of these present as 5MC, there are ~1.8 x 10⁷ 5MC per diploid mouse nucleus. Multiplying this value by the 5MC deamination rate determined in the lacI transgene of adult mice (2.3 x 10^-8 deam/5MC/month; Table 1) indicates that overall, ~0.4 deaminations of 5MC are expected to occur per month per diploid mouse genome due to spontaneous hydrolytic (and mutagenic) deamination of 5MC.

On the basis of the 5MC content of mouse liver estimated above and the demethylation rate reported by WILSON et al. 1987 of 0.012% per month, we estimate that in aging adult mouse liver demethylation results in the disappearance of ~2.2 x 10³ 5MC per diploid nucleus per month, a rate that is ~5.5 x 10³-fold greater than that which occurs due to the deamination of 5MC determined on the basis of data from the lacI transgene. That is, if the demethylation rate determined by WILSON et al. 1987 was due solely to hydrolytic deamination of 5MC to thymine [or possibly, DNA (cytosine-5)-methyltransferase-mediated conversion of cytosine to thymine during methylation of hemimethylated CpG sequences], the Mf observed in the lacI transgene would be ~1.3 x 10^-4 deam/5MC/month (= 1.1 x 10^-2 deam/lacI transgene/month), an extraordinarily high Mf. On the basis of these calculations, demethylation of genomic DNA at the levels described by Gama-Sosa, Wilson, and others cannot be attributable to spontaneous hydrolytic or enzyme-mediated (mutagenic) deamination events. The recent discovery of a mammalian DNA demethylase specific for 5MC (BHATTACHARYA et al. 1999 ; RAMCHANDANI et al. 1999 ) therefore provides the most reasonable explanation, at present, for the nonmutagenic hypomethylation of DNA. As mentioned, this enzyme directly converts 5MC to cytosine, resulting in GC base pairs.

A descriptive mutational model:
As noted previously (STUART et al. 2000 ; DISCUSSION), mutations occur at different rates in mouse liver and brain tissue, especially in adult tissues. Furthermore, the Mf for the most common spontaneous mutation, deamination of 5MC present at CpG sequences, closely parallels the larger, overall spontaneous Mf. Since these transition mutations are presumably of specific origin, hydrolytic deamination of 5MC, the spontaneous rate of deamination of 5MC in vitro is expected (on the basis of thermodynamic considerations, i.e., Arrhenius kinetics) to be constant at a given temperature. Thus, deviations from this rate observed in vivo in different tissues must reflect biological influences, including DNA repair.

Observed MR (and Mf) likely reflect a balance between the formation of premutagenic DNA lesions and mispairs, DNA replication (fixation of mutations during translesion bypass, misincorporation at the replication fork), DNA repair, and cell death. Consideration of the effects of these processes on mutation leads to the description of mutation as a function (f),

(1)

that can be expressed as an equation

(2)

(It is noted that somewhat analogous models have been proposed by others; e.g., BURKHART and MALLING 1993 ; DRAKE et al. 1998 ; and HOLMQUIST 1998 .) In our model, which may be amended or refined as required, P is the probability of the indicated process. The DNA repair term accounts for the repair of preexisting, premutagenic DNA damage and mispairs. The mutagenic trans-lesion bypass term reflects the contribution of DNA replication past the DNA lesion that results in mutation; if there is no misincorporation or if the lesion blocks replication (which is likely to be a lethal event), then there is no contribution to the mutation rate. The cell death term is included in the event that a mutational treatment results in significant apoptotic or necrotic cell death that might affect the observed MR; under normal circumstances, this effect is believed to be negligible. Finally, it is noted that nucleotides are sometimes misincorporated during replication of nondamaged DNA; however, these mispairs do not immediately contribute to the MR but rather contribute subsequently via the other terms in the equation.

It follows that

(3)

Interestingly, under equilibrium (steady-state) conditions, DNA replication contributes a linear increase to observed Mf according to the formula

(4)

where i is the number of cell divisions (or DNA replications) that have occurred. For example, assuming an initial population of 10⁶ largely wild-type cells but containing 100 lacI mutants and having a MR of 10^-5 new mutants per cell division (or DNA replication), after 5 cell divisions the initial Mf of 1.00 x 10^-4 will have increased to 1.25 x 10^-4. (The Mf increases are 1.05 x 10^-4, 1.10 x 10^-4, 1.15 x 10^-4, 1.20 x 10^-4, and 1.25 x 10^-4 after 1, 2, 3, 4, and 5 divisions, respectively.) Thus, ratios of Mf such as spontaneous vs. induced reflect overall differences in MR.

In time, sufficient data might be accumulated for each of the factors inherent in Equation 2 to permit the quantitative evaluation of mutation a priori. Nevertheless, in the interim this model might serve as a useful model for evaluating qualitatively or semiquantitatively the relative contributions of various factors to Mf observed in vivo or in vitro.

To illustrate, we applied our descriptive mutational model to calculate the efficiency of repair of GT mispairs in mouse liver DNA arising due to deamination of 5MC. Since mouse hepatocytes are relatively long lived, and apoptotic indices in adult mouse liver are typically ~0.007% (JAMES et al. 1998 ), we may ignore the cell death term. The translesion bypass term can probably be eliminated since unrepaired GT mispairs will be faithfully replicated, more or less. Therefore, substituting the rate of hydrolytic deamination of 5MC determined in vitro (Table 2) and the Mf due to GC AT at CpG sequences (Table 1) into Equation 2

we calculate that DNA repair of GT mispairs in mouse liver was ~98.5–99.9% efficient, depending on the in vitro MR chosen. (Note that in this instance, we needed to use the Mf value from Table 1 rather than the MR for the units to cancel.) Similarly, we predict that since the human 5MC deamination rate is slower than in mice, the efficiency of repair of GT mispairs is higher, ~99.99–99.9996%, on the basis of the data from Table 2.

In a second application of our mutational model, the recent report of spontaneous mutations in the lacI transgene of DNA-repair-proficient and Msh2^-/- mice (ANDREW et al. 1997 ) provides the opportunity to estimate the contribution of this DNA repair pathway to observed Mf. In the three tissues examined (small intestine, thymus, and brain), MF were elevated ~11-, 15- and 5-fold, respectively, in Msh2^-/- mice compared to control animals. Using the thymus as an example and assuming that the differences in the observed MF in control and Msh2^-/- mice were attributable solely to the DNA repair defect, Equation 2 may be simplified as a pair of equations

that can be rearranged by substitution of terms to give

Thus, if we make the assumption that DNA repair in the control thymus was 99.99% efficient, then the equation suggests that DNA repair in the Msh2^-/- thymus would have been 99.848% efficient, a very small change in overall DNA repair. (Note that under these conditions, DNA repair efficiency in the Msh2 thymus would apparently drop to zero as overall DNA repair efficiency in the control thymus drops to 93.4%.) Whether or not these conclusions are correct remains to be validated. As well, it cannot be excluded that small differences in the extent of DNA replication or cell death in Msh2^-/- mice relative to control animals also contribute significantly to the differences in spontaneous Mf observed in these animals. Nevertheless, the model predicts that subtle differences in the efficiency of DNA repair might profoundly influence observed Mf. If this is true, one implication is that mutator phenotypes (LOEB 1991 ) could be attributable to small perturbations in the efficiency of DNA repair.

	FOOTNOTES

¹ For now we see through a glass, darkly; but then face to face: now I know in part; but then shall I know even as also I am known. 1 Corinthians 13:12.

	ACKNOWLEDGMENTS

We acknowledge Drs. Jan Drake, Susan Lewis, James Neel, and Richard Setlow for commenting on this manuscript prior to submission, and Dr. Richard Michod and the anonymous reviewers for their comments on the submitted manuscript. The support of The Cancer Research Society, Inc. (Montreal, Quebec, Canada) for G.R.S. is gratefully acknowledged.

Manuscript received December 1, 1999; Accepted for publication March 27, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS AND DISCUSSION LITERATURE CITED

ANDREW, S. E., A. H. REITMAIR, J. FOX, L. HSIAO, and A. FRANCIS et al., 1997  Base transitions dominate the mutational spectrum of a transgenic reporter gene in MSH2 deficient mice. Oncogene 15:123-129[Medline].

BHATTACHARYA, S. K., S. RAMCHANDANI, N. CERVONI, and M. SZYF, 1999  A mammalian protein with specific demethylase activity for mCpG DNA. Nature 397:579-583[Medline].

BIRD, A. P., 1980  DNA methylation and the frequency of CpG in animal DNA. Nucleic Acids Res. 8:1499-1504[Abstract/Free Full Text].

BOWMAN, P. D., 1985 Aging and the cell cycle in vivo and in vitro, pp. 117–136 in CRC Handbook of Cell Biology of Ageing, edited by V. J. CRISTOFALO. CRC Press, Boca Raton, FL.

BRODSKY, V. Y., and I. V. URYVAEVA, 1985 The biological significance of polyploidy and polyteny, pp. 175–233 in Genome Multiplication in Growth and Development. Biology of Polyploid and Polytene Cells. Cambridge University Press, Cambridge.

BRODSKY, W. Y. and I. V. URYVAEVA, 1977  Cell polyploidy: its relation to tissue growth and function. Int. Rev. Cytol. 50:275-332[Medline].

BROOKS, P. J., C. MARIETTA, and D. GOLDMAN, 1996  DNA mismatch repair and DNA methylation in adult brain neurons. J. Neurosci. 16:939-945[Abstract/Free Full Text].

BROWN, T. C. and J. JIRICNY, 1987  A specific mismatch repair event protects mammalian cells from loss of 5-methylcytosine. Cell 50:945-950[Medline].

BUETOW, D. E., 1985 Cell numbers vs. age in mammalian tissues and organs, pp. 1–115 in CRC Handbook of Cell Biology of Ageing, edited by V. J. CRISTOFALO. CRC Press, Boca Raton, FL.

BURKHART, J. G. and H. V. MALLING, 1993  Mutagenesis and transgenic systems: perspective from the mutagen, N-ethyl-N-nitrosourea. Environ. Mol. Mutagen. 22:1-6[Medline].

CAMERON, I. L., 1971 Cell proliferation and renewal in the mammalian body, pp. 45–85 in Cellular and Molecular Renewal in the Mammalian Body, edited by I. L. CAMERON and J. D. THRASHER. Academic Press, New York.

CARRIERE, R., 1969  The growth of liver parenchymal nuclei and its endocrine regulation. Int. Rev. Cytol. 25:201-277[Medline].

COOPER, D. N. and M. KRAWCZAK, 1989  Cytosine methylation and the fate of CpG dinucleotides in vertebrate genomes. Hum. Genet. 83:181-188[Medline].

COOPER, D. N. and H. YOUSSOUFIAN, 1988  The CpG dinucleotide and human genetic disease. Hum. Genet. 78:151-155[Medline].

COULONDRE, C., J. H. MILLER, P. J. FARABAUGH, and W. GILBERT, 1978  Molecular basis of base substitution hotspots in Escherichia coli.. Nature 274:775-780[Medline].

CURTIS, H. J., 1971  Genetic factors in aging. Adv. Genet. 16:305-324[Medline].

DE BOER, J. G. and B. W. GLICKMAN, 1998  The lacI gene as a target for mutation in transgenic rodents and Escherichia coli.. Genetics 148:1441-1451[Abstract/Free Full Text].

DE BOER, J. G., H. ERFLE, J. HOLCROFT, D. WALSH, and M. DYCAICO et al., 1996  Spontaneous mutants recovered from liver and germ cell tissue of low copy number lacI transgenic rats. Mutat. Res. 352:73-78[Medline].

DE BOER, J. G., H. ERFLE, D. WALSH, J. HOLCROFT, and J. S. PROVOST et al., 1997  Spectrum of spontaneous mutations in liver tissue of lacI transgenic mice. Environ. Mol. Mutagen. 30:273-286[Medline].

DRAHOVSKY, D. and T. L. J. BOEHM, 1980  Enzymatic DNA methylation in higher eukaryotes. Int. J. Biochem. 12:523-528[Medline].

DRAKE, J. W., B. CHARLESWORTH, D. CHARLESWORTH, and J. F. CROW, 1998  Rates of spontaneous mutation. Genetics 148:1667-1686[Abstract/Free Full Text].

DROST, J. B. and W. R. LEE, 1995  Biological basis of germline mutation: comparisons of spontaneous germline mutation rates among Drosophila, mouse, and human. Environ. Mol. Mutagen. 25(Suppl. 26):48-64.

DYCAICO, M. J., G. S. PROVOST, P. L. KRETZ, S. L. RANSOM, and J. C. MOORES et al., 1994  The use of shuttle vectors for mutation analysis in transgenic mice and rats. Mutat. Res. 307:461-478[Medline].

ELDRIDGE, S. R. and S. M. GOLDSWORTHY, 1996  Cell proliferation rates in common cancer target tissues of B6C3F1 mice and F344 rats: effects of age, gender, and choice of marker. Fundam. Appl. Toxicol. 32:159-167[Medline].

FARABAUGH, P. J., 1978  Sequence of the lacI gene. Nature 274:765-769[Medline].

FRANKS, L. M., P. D. WILSON, and R. D. WHELAN, 1974  The effects of age on total DNA and cell number in the mouse brain. Gerontologia 20:21-26[Medline].

GAMA-SOSA, M. A., R. M. MIDGETT, V. A. SLAGEL, S. GITHENS, and K. C. KUO et al., 1983  Tissue-specific differences in DNA methylation in various mammals. Biochim. Biophys. Acta 740:212-219[Medline].

HEDDLE, J. A., 1998  The role of proliferation in the origin of mutations in mammalian cells. Drug Metab. Rev. 30:327-338[Medline].

HEDDLE, J. A., 1999  On clonal expansion and its effects on mutant frequencies, mutation spectra and statistics for somatic mutations in vivo.. Mutagenesis 14:257-260[Free Full Text].

HOAL-VAN HELDEN, E. G. and P. D. VAN HELDEN, 1989  Age-related methylation changes in DNA may reflect the proliferative potential of organs. Mutat. Res. 219:263-266[Medline].

HOLMQUIST, G. P., 1998  Chronic low-dose lesion equilibrium along genes: measurement, molecular epidemiology, and theory of the minimal relevant dose. Mutat. Res. 405:155-159[Medline].

JAMES, S. J., L. MUSKHELISHVILI, D. W. GAYLOR, A. TURTURRO, and R. HART, 1998  Upregulation of apoptosis with dietary restriction: implications for carcinogenesis and aging. Environ. Health Perspect. 106(Suppl. 1):307-312.

KANUNGO, M. S. and S. SARAN, 1992  Methylation of DNA of the brain and liver of young and old rats. Indian J. Biochem. Biophys. 29:49-53[Medline].

KOEBERL, D. D., C. D. K. BOTTEMA, R. P. KETTERLING, P. J. BRIDGE, and D. P. LILLICRAP et al., 1990  Mutations causing hemophilia B: direct estimate of the underlying rates of spontaneous germ-line transitions, transversions, and deletions in a human gene. Am. J. Hum. Genet. 47:202-217[Medline].

KOHLER, S. W., G. S. PROVOST, P. L. KRETZ, M. J. DYCAICO, and J. A. SORGE et al., 1990  Development of a short-term, in vivo mutagenesis assay: the effects of methylation on the recovery of a lambda phage shuttle vector from transgenic mice. Nucleic Acids Res. 25:3007-3013.

KOHLER, S. W., G. S. PROVOST, A. FIECK, P. L. KRETZ, and W. O. BULLOCK et al., 1991  Spectra of spontaneous and mutagen-induced mutations in the lacI gene in transgenic mice. Proc. Natl. Acad. Sci. USA 88:7958-7962[Abstract/Free Full Text].

KORR, H., 1980  Proliferation of different cell types in the brain. Adv. Anat. Embryol. Cell Biol. 61:1-72[Medline].

LAIRD, P. W. and R. JAENISCH, 1996  The role of DNA methylation in cancer genetic and epigenetics. Annu. Rev. Genet. 30:441-464[Medline].

LI, W.-H. and M. TANIMURA, 1987  The molecular clock runs more slowly in man than in apes and monkeys. Nature 326:93-96[Medline].

LI, W.-H., D. L. ELLSWORTH, J. KRUSHKAL, B.-H. CHANG, and D. HEWETT-EMMETT, 1996  Rates of nucleotide substitution in primates and rodents and the generation-time effect hypothesis. Mol. Phylogenet. Evol. 5:182-187[Medline].

LOEB, L. A., 1991  Mutator phenotype may be required for multistage carcinogenesis. Cancer Res. 51:3075-3079[Free Full Text].

MARIETTA, C., F. PALOMBO, P. GALLINARI, J. JIRICNY, and P. J. BROOKS, 1998  Expression of long-patch and short-patch DNA mismatch repair proteins in the embryonic and adult mammalian brain. Brain Res. Mol. Brain Res. 53:317-320[Medline].

MATSUO, K., O. CLAY, T. TAKAHASHI, J. SILKE, and W. SCHAFFNER, 1993  Evidence for erosion of mouse CpG islands during mammalian evolution. Somat. Cell Mol. Genet. 19:543-555[Medline].

MAZIN, A. L., 1994  Enzymatic DNA methylation as a mechanism of ageing. Mol. Biol. 28:11-31.

MAZIN, A. L., 1995  Life span prediction from the rate of age-related DNA demethylation in normal and cancer cell lines. Exp. Gerontol. 30:475-484[Medline].

MIRSALIS, J. C., 1995  Transgenic models for detection of mutations in tumors and normal tissues of rodents. Toxicol. Lett. 82–83:131-134.

MIRSALIS, J. C., J. A. MONFORTE, and R. A. WINEGAR, 1994  Transgenic animal models for measuring mutations in vivo.. Crit. Rev. Toxicol. 24:255-280[Medline].

OCHIAI, M., K. ISHIDA, T. USHIJIMA, T. SUZUKI, and T. SOFUNI et al., 1998  DNA adduct level induced by 2-amino-3,4-dimethylimidazo[4,5-f]-quinoline in Big Blue mice does not correlate with mutagenicity. Mutagenesis 13:381-384[Abstract/Free Full Text].

OKONOGI, H., T. USHIJIMA, X. B. ZHANG, J. A. HEDDLE, and T. SUZUKI et al., 1997  Agreement of mutational characteristics of heterocyclic amines in lacI of the Big Blue mouse with those in tumor related genes in rodents. Carcinogenesis 18:745-748[Abstract/Free Full Text].

PROVOST, G. S. and J. M. SHORT, 1994  Characterization of mutations induced by ethylnitrosourea in seminiferous tubule germ cells of transgenic B6C3F₁ mice. Proc. Natl. Acad. Sci. USA 91:6564-6568[Abstract/Free Full Text].

RAMCHANDANI, S., S. K. BHATTACHARYA, N. CERVONI, and M. SZYF, 1999  DNA methylation is a reversible biological signal. Proc. Natl. Acad. Sci. USA 96:6107-6112[Abstract/Free Full Text].

REIN, T., M. L. DEPAMPHILIS, and H. ZORBAS, 1998  Identifying 5-methylcytosine and related modifications in DNA genomes. Nucleic Acids Res. 26:2255-2264[Abstract/Free Full Text].

SCHULTZE, B., A. M. KELLERER, C. GROSSMANN, and W. MAURER, 1978  Growth fraction and cycle duration of hepatocytes in the three-week-old rat. Cell Tissue Kinet. 11:241-249[Medline].

SCRABLE, H. and P. J. STAMBROOK, 1997  Activation of the lac repressor in the transgenic mouse. Genetics 147:297-304[Abstract].

SHEN, J.-C., W. M. RIDEOUT, III, and P. A. JONES, 1992  High frequency mutagenesis by a DNA methyltransferase. Cell 71:1073-1080[Medline].

SHEN, J.-C., W. M. RIDEOUT, III, and P. A. JONES, 1994  The rate of hydrolytic deamination of 5-methylcytosine in double-stranded DNA. Nucleic Acids Res. 22:972-976[Abstract/Free Full Text].

SINGHAL, R. P., L. L. MAYS-HOOPES, and G. L. EICHHORN, 1987  DNA methylation in aging of mice. Mech. Ageing Dev. 41:199-210[Medline].

SKOPEK, T. R., K. L. KORT, and D. R. MARINO, 1995  Relative sensitivity of the endogenous hprt gene and lacI transgene in ENU-treated Big Blue B6C3F1 mice. Environ. Mol. Mutagen. 26:9-15[Medline].

SKOPEK, T., D. MARINO, K. KORT, J. MILLER, and M. TRUMBAUER et al., 1998  Effect of target gene CpG content on spontaneous mutation in transgenic mice. Mutat. Res. 400:77-88[Medline].

STEINBERG, R. A. and K. B. GORMAN, 1992  Linked spontaneous CGTA mutations at CpG sites in the gene for protein kinase regulatory subunit. Mol. Cell. Biol. 12:767-772[Abstract/Free Full Text].

STUART, G. R., Y. ODA, J. G. DE BOER, and B. W. GLICKMAN, 2000  Mutation frequency and specificity with age in liver, bladder and brain of lacI transgenic mice. Genetics 154:1291-1300[Abstract/Free Full Text].

TAO, K. S., C. URLANDO, and J. A. HEDDLE, 1993  Comparison of somatic mutation in a transgenic versus host locus. Proc. Natl. Acad. Sci. USA 90:10681-10685[Abstract/Free Full Text].

TAWA, R., T. ONO, A. KURISHITA, S. OKADA, and S. HIROSE, 1990  Changes of DNA methylation level during pre- and postnatal periods in mice. Differentiation 45:44-48[Medline].

THOMPSON, J. N., JR., R. C. WOODRUFF, and H. HUAI, 1998  Mutation rate: a simple concept has become complex. Environ. Mol. Mutagen. 32:292-300[Medline].

URYVAEVA, I. V., 1981  Biological significance of liver cell polyploidy: an hypothesis. J. Theor. Biol. 89:557-571[Medline].

WALKER, V. E., N. J. GORELICK, J. L. ANDREWS, T. R. CRAFT, and J. G. DE BOER et al., 1996  Frequency and spectrum of ethylnitrosourea-induced mutation at the hprt and lacI loci in splenic lymphocytes of exposed lacI transgenic mice. Cancer Res. 56:4654-4661[Abstract/Free Full Text].

WALTER, C. A., G. W. INTANO, J. R. MCCARREY, C. A. MCMAHAN, and R. B. WALTER, 1998  Mutation frequency declines during spermatogenesis in young mice but increases in old mice. Proc. Natl. Acad. Sci. USA 95:10015-10019[Abstract/Free Full Text].

WILSON, V. L. and P. A. JONES, 1983  DNA methylation decreases in aging but not in immortal cells. Science 220:1055-1057[Abstract/Free Full Text].

WILSON, V. L., R. A. SMITH, S. MA, and R. G. CUTLER, 1987  Genomic 5-methyldeoxycytidine decreases with age. J. Biol. Chem. 262:9948-9951[Abstract/Free Full Text].

WINICK, M., J. A. BRASEL, and P. ROSSO, 1972  Nutrition and cell growth. Curr. Concepts Nutr. 1:49-97[Medline].

YANG, A. S., M. L. GONZALGO, J.-M. ZINGG, R. P. MILLAR, and J. D. BUCKLEY et al., 1996  The rate of CpG mutation in Alu repetitive elements within the p53 tumor suppressor gene in the primate germline. J. Mol. Biol. 258:240-250[Medline].

YOU, Y. H., A. HALANGODA, V. BUETTNER, K. HILL, and S. SOMMER et al., 1998  Methylation of CpG dinucleotides in the lacI gene of the Big Blue transgenic mouse. Mutat. Res. 420:55-65[Medline].

ZHANG, X. and C. K. MATHEWS, 1994  Effect of DNA cytosine methylation upon deamination-induced mutagenesis in a natural target sequence in duplex DNA. J. Biol. Chem. 269:7066-7069[Abstract/Free Full Text].

This article has been cited by other articles:

				V. Gorbunova, A. Seluanov, Z. Mao, and C. Hine Changes in DNA repair during aging Nucleic Acids Res., December 3, 2007; 35(22): 7466 - 7474. [Abstract] [Full Text] [PDF]

				J. Wang, K. D. Gonzalez, W. A. Scaringe, K. Tsai, N. Liu, D. Gu, W. Li, K. A. Hill, and S. S. Sommer From the Cover: Evidence for mutation showers PNAS, May 15, 2007; 104(20): 8403 - 8408. [Abstract] [Full Text] [PDF]

				B. L. Parsons, R. R. Delongchamp, F. A. Beland, and R. H. Heflich Levels of H-ras codon 61 CAA to AAA mutation: response to 4-ABP-treatment and Pms2-deficiency Mutagenesis, January 1, 2006; 21(1): 29 - 34. [Abstract] [Full Text] [PDF]

				D G R Evans, E R Maher, and M E Baser Age related shift in the mutation spectra of germline and somatic NF2 mutations: hypothetical role of DNA repair mechanisms J. Med. Genet., August 1, 2005; 42(8): 630 - 632. [Abstract] [Full Text] [PDF]

				A. Seluanov, D. Mittelman, O. M. Pereira-Smith, J. H. Wilson, and V. Gorbunova DNA end joining becomes less efficient and more error-prone during cellular senescence PNAS, May 18, 2004; 101(20): 7624 - 7629. [Abstract] [Full Text] [PDF]

				Y.-X. Fu and H. Huai Estimating Mutation Rate: How to Count Mutations? Genetics, June 1, 2003; 164(2): 797 - 805. [Abstract] [Full Text] [PDF]