Mlh1 mediates tissue-specific regulation of mitotic ... - Nature

2 downloads 0 Views 172KB Size Report
Oct 11, 2004 - Changshun Shao*,1, Li Deng1, Yanping Chen1, Raju Kucherlapati2, Peter J Stambrook3 and. Jay A Tischfield*,1. 1Department of Genetics ...
Oncogene (2004) 23, 9017–9024

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Mlh1 mediates tissue-specific regulation of mitotic recombination Changshun Shao*,1, Li Deng1, Yanping Chen1, Raju Kucherlapati2, Peter J Stambrook3 and Jay A Tischfield*,1 1 Department of Genetics, Rutgers University, 604 Allison Road, Piscataway, NJ 08854-8082, USA; 2Departments of Genetics and Medicine, Harvard Medical School and Harvard-Partners Center for Genetics and Genomics, Boston, MA 02115, USA; 3Department of Cell Biology, Neurobiology and Anatomy, University of Cincinnati College of Medicine, Cincinnati, OH 45267, USA

Mitotic recombination (MR) between chromosome homologs in somatic cells is a major pathway to the loss of heterozygosity (LOH), which may cause cancer if tumor suppressor genes are involved. MR can be suppressed by DNA sequence heterology (homeology) in hybrid mice from matings between species or between subspecies. We now report that MR is relatively suppressed in F1 hybrids between inbred strains C57BL/6 and 129S2. The frequency of MR in fibroblasts is lower in F1 hybrid mice than in either of the two parental strains. However, MR in T cells is not affected by strain background. Thus, relatively small genetic differences are capable of restricting MR in a tissue-specific manner. Using Mlh1deficient mice, we tested the role of mismatch repair in MR in two isogenic cell types. In fibroblasts of C57BL/ 6  129S2 F1 mice, the suppression of MR is alleviated in the absence of MLH1. In contrast, MR is not affected by Mlh1 status in T cells. The frequency of point mutations at the reporter gene loci Aprt and Hprt, on the other hand, is significantly increased in both T cells and fibroblasts of Mlh1/ mice. Thus, different cell types respond differently to MLH1 deficiency, and the contribution of MR to tumorigenesis may be tissue-dependent in the absence of mismatch repair. Oncogene (2004) 23, 9017–9024. doi:10.1038/sj.onc.1208148 Published online 11 October 2004 Keywords: Mlh1; mismatch repair; mitotic recombination; point mutation; mouse model

Introduction Loss of heterozygosity (LOH) at tumor suppressor genes is associated with the etiology of many cancers (Cavenee et al., 1983; Haigis et al., 2002). LOH can occur through one of at least six mechanistically distinct pathways, including mitotic recombination (MR), which involves the reciprocal exchange of chromatid segments between chromosome homologs (Cavenee et al., 1983; Haigis et al., 2002). By using mouse models, we and *Correspondence: C Shao or JA Tischfield E-mail: [email protected] or [email protected] Received 25 February 2004; revised 29 July 2004; accepted 30 July 2004; published online 11 October 2004

other groups have identified several factors that modulate MR in mammalian cells in vivo. First, the initiation of MR requires the physical proximity between homologs (Haigis and Dove, 2003). When the nucleolar organizing region on mouse chromosome 18 is absent from one homolog, due to a chromosomal translocation, homolog co-localization near the nucleolus is impaired and is accompanied by a decreased probability of MR and a reduced incidence of intestinal adenomas in ApcMin mice (Haigis and Dove, 2003). Second, MR is highly dependent on DNA sequence homology between the homologs (Shao et al., 2001). While MR causes between 60 and 80% of Aprt LOH in hybrid mice of inbred strains, such as 129S2  C3H/HeJ F1, it is minimally detectable in fibroblasts of hybrids of distantly related mouse strains, such as 129S2 versus CAST/Ei. When regional chromosome homology is reintroduced, however, a high level of MR can be restored in those regions (Shao et al., 2001). Third, studies on mice that are deficient in genes responsible for DNA damage response also led to the identification of genes that may regulate MR. For instance, MR is dramatically increased in Blm-deficient mice (Luo et al., 2000). Also, there is a modest increase of MR in fibroblasts, but not in T cells, of p53 null mice (Shao et al., 2000). These findings indicate that BLM and TP53 normally suppress MR. Interestingly, a variety of chemical mutagens, but not one dose of ionizing radiation, were found to induce MR in splenic T cells (Liang et al., 2002; Wijnhoven et al., 2003). We use Aprt þ / mice as a model system to study LOH (Shao et al., 1999). Somatic cell variants that have undergone in vivo LOH at Aprt are recoverable and expandable in vitro by virtue of their pre-existing resistance to adenine analogs such as 2,6- diaminopurine (DAP). The DAP-resistant (DAPr) colonies are then analysed to discriminate between genetic mechanisms that produced the in vivo loss of APRT activity. We showed previously that, while MR occurs frequently and accounts for majority of the LOH events in the hybrid progeny of crosses between classical mouse strains, it is significantly suppressed in the hybrids between a classical inbred strain and a feral strain (Shao et al., 1999, 2001). We further showed that this suppression is caused by chromosomal divergence (Shao et al., 2001). Here we tested whether MR can be further enhanced in inbred strains, where the two homologs

Mlh1-mediated suppression of somatic mutation C Shao et al

9018

Mus musculus 129S2

Oncogene

C3H/HeJ

musculus

C57BL/6J

castaneus Inbred laboratory strains

Mus spretus

b

MR is suppressed in a tissue-specific manner in hybrids of inbred mouse strains

In

Inb

re 57 d st BL ra , 1 ins 29 S2 Int ers ) (12 tr 9S ain 2 X hy C5 brid 7B s L)

Sequence divergence

(C

The Aprt þ / model has been used to demonstrate that, in hybrids of two closely related laboratory strains, such as 129S2  C3H/HeJ F1 or 129S2  C57BL/6J F1, the majority of DAPr clones are produced by MR. In contrast, in hybrids of distantly related strains, such as 129S2 x SPRET/Ei F1, MR is significantly reduced (Shao et al., 2001), indicating the dependence of MR on chromosome homology. Although different laboratory strains were derived from a small pool of founders, they have diverged from each other for nearly a hundred years. Significant DNA sequence differences have accumulated, as evidenced by the high level of polymorphism in microsatellite repeats (Dietrich et al., 1994), by the rapid amplification of retrotransposons (DeBerardinis et al., 1998), and by the widespread occurrence of single-nucleotide polymorphisms (Wade et al., 2002). Thus, while chromosome homologs in 129S2  C57BL/6J F1, designated Chr8129 and Chr8C57, respectively, are likely to have a higher degree of homology than homologs in 129S2  SPRET/Ei F1, they still are less ideal substrates for homologous recombination than homologs in inbred strains where the two chromosomes are identical (e.g. Chr8129 versus Chr8129). This suggests that MR in inbred mouse strains should be higher than in F1 hybrids between inbred strains (Figure 1). To test this proposition, we measured the frequency of DAPr clones from Aprt þ / F1 mice derived from 129S2  C57BL/6J (or C57BL/6J  129S2) crosses and from 129S2 and C57BL/6J Aprt þ / congenic mice. As shown in Figure 2, the frequency of DAPr fibroblast clones from either of the two inbred strains was higher than that from the F1 hybrid mice. In all mice, the large majority of fibroblast clones had lost the untargeted Aprt allele and were designated as class I (Table 1). Class II includes all mechanisms where APRT activity is lost but the untargeted Aprt allele remains (Shao et al., 1999). As all simple sequence repeat (SSR) markers in the inbred strains are homozygous, the contribution of MR to the class I mutants could not be unequivocally determined in these mice. However, given

domesticus

te (12 rspe 9S cifi 2X ch SP ybri RE ds T)

Results

a

Level of MR

are identical, by measuring the frequency of MR in an inter-strain mouse hybrid, 129S2  C57BL/6 F1 and in its inbred parents. We found that the frequency of MR in fibroblasts, but not in T cells, is significantly lower in the hybrid than in either of the parental strains, substantiating the notion that sequence divergence can suppress MR. We also tested whether the mismatch repair gene Mlh1 plays a role in regulating MR. We found that suppression of MR in hybrid mice is alleviated in fibroblasts in the absence of MLH1. We also found that point mutations are elevated in both fibroblasts and T cells of Mlh1/ mice, but that the mutational spectrum is different from that observed with Pms2/ mice.

Figure 1 Inverse correlation of sequence divergence and MR. (a) A partial phylogenetic tree showing the evolution of the inbred laboratory mouse strains. About two-thirds of the genome of the laboratory strains are believed to be derived from the subspecies Mus musculus domesticus, and the rest from Mus musculus musculus subspecies (Wade et al., 2002). Inbred laboratory strains are more closely related to each other than to wild-derived mice. (b) Predicted decrease of MR as sequence divergence between homologs increases

that almost all class I clones recovered from 129S2 x C57BL/6J, 129S2 x C3H/HeJ hybrids (Shao et al., 1999, 2000, 2001) unequivocally arise by MR, it is highly likely that most of the class I clones in the 129S2 and C57 congenic mice are also caused by MR. This finding suggests that even a relatively low level of DNA sequence divergence, such as that between closely related inbred strains, can suppress MR. In contrast to fibroblasts, the frequency with which DAPr T-cell clones arise is not significantly different between congenic C57BL/6J mice and 129S2 x C57 F1 hybrids, and neither is the frequency of MR (Table 2). Thus, unlike fibroblasts, the sequence divergence between the Chr8129 and Chr8C57 does not reduce the frequency of MR in splenic T cells. These data are entirely consistent with our earlier observation that there is less suppression of MR between Aprt and centromere in T cells compared to fibroblasts derived from hybrids of distantly related strains (Shao et al., 2001).

Mlh1-mediated suppression of somatic mutation C Shao et al

9019

MR between divergent homologs is elevated in fibroblasts but not in T cells of Mlh1/ mice We backcrossed C57BL/6 Mlh1/ knockout mice (Edelmann et al., 1996) to 129S2 Aprt/ mice for five generations. In this manner, (N5)129S2 Mlh1 þ /Aprt þ / mice were generated, in which chromosome 8 harboring the null Aprt allele is entirely derived from the 129S2 background, as determined by analysis of SSR polymorphisms. These mice were then crossed to C57BL/6 Mlh1 þ / mice to produce hybrid mice that are heterozygous for Aprt, along with each of the possible Mlh1 genotypes. The frequency of DAPr variants from Mlh1/  hybrid mice increased about six fold for fibroblasts (Table1 and Figure 3) and about 3.5-fold for T cells (Table 2). Allele-specific PCR and analysis of polymorphic SSRs revealed that the frequency of MR in fibroblasts increased in proportion to the increase in mutant frequency (Table 1). In contrast, the increase in T-cell mutant frequency was exclusively due to an increase in class II clones, with no evidence of elevated MR (Table 2). Thus, these data indicate that Mlh1 suppresses MR between divergent homologs in fibroblasts but not in T cells. Consistent with the observation that mismatch repair suppresses MR when chromosome homologs have divergent sequences, the frequency of MR was independent of mismatch repair status in inbred strains of mice where the chromosome homologs are identical. The frequency of DAPr fibroblast colonies from C57BL Mlh1/ was similar to that of C57BL Mlh1 þ / þ mice (Figures 2 and 3, Table 1). Splenic T cells from C57BL/ 6J Mlh1/ mice, like those from 129S2  C57BL/6J F1 Mlh1/ mice, displayed a fourfold increase in DAPr

1000

100

P = 0.0037

10 P = 0.0004

Frequency of DAPr fibroblasts (x 10-6)

10000

129

C57

1 129xC57

Figure 2 MR is enhanced in fibroblasts of inbred mice compared to hybrids of inbred strains. Each point represents the frequency of DAPr fibroblasts from one ear. The bars indicate median values. The majority of the DAPr clones are caused by MR. Points at the bottom represent ears from which no DAPr colonies were recovered

Table 1

Mitotic recombination in fibroblasts is modulated by chromosome homology and Mlh1 status

Genotype

Strain

No. of ears

Cloning efficiency (mean7s.e.)

DAPr frequency (  106)a

No. of DAPr colonies analysed

No. of MR (or class I)

% MR (or class I)

MR frequency (  106)b

Mlh1+/+ Mlh1+/+ Mlh1+/+

129  C57 129 C57

18 21 30

1.5870.08 1.5470.08 1.1870.09

47 183 132

10 50 35

8 38 31

80 76 89

37.6 139.1 117.5

Mlh1+/+ Mlh1+/ Mlh1/ Mlh1/

(N5)129  C57 (N5)129  C57 (N5)129  C57 C57

26 18 26 20

1.7370.12 1.3570.03 1.6370.13 1.4170.08

30* 57 178* 151

22 21 71 41

17 15 50 28

77 71 70 68

23.1 40.7 124.6 102.7

Median. bCalculated by multiplying DAPr frequency and the percentage of MR. MR, mitotic recombination. *P ¼ 0.0007, Mann–Whitney U-test

a

Table 2 MR in T cells is not affected by chromosome homology and Mlh1 status Genotype

Strain

No. of mice

Cloning efficiency (mean7s.e.)

DAPr frequency (  106)a

No. of DAPr colonies analysed

No. of MR (or class I)

% MR (or class I)

MR frequency (  106)b

Mlh1+/+ Mlh1+/+ Mlh1/ Mlh1/

129  C57 C57 (N5)129  C57 C57

13 12 10 7

8.4471.36 6.6171.13 6.5971.04 6.1570.94

21.6 14.7 74.5 66.5

66 44 113 73

39 31 23 7

59.1 70.5 20.4 9.6

12.8 10.4 15.2 6.4

a

Median. bCalculated by multiplying the DAPr frequency and the percentage of MR Oncogene

Mlh1-mediated suppression of somatic mutation C Shao et al

9020

colonies compared with Mlh1 þ / þ mice, and the increase was caused entirely by class II variants, whereas the frequency of class I (representative of MR) remained the same (Table 2). Point mutation is elevated in both T cells and fibroblasts of Mlh1/ mice We have previously shown that point mutations were increased for both T cells and fibroblasts of Pms2/ mice, while MR was unaffected (Shao et al., 2002). As in Pms2/ mice, the elevation in the frequency of DAPr T cells in Mlh1/ mice was entirely due to an increase in class II clones, about 10-fold in the latter case (Table 2). 10000

Frequency of DAPr fibroblasts ( x 10-6)

Mlh1+/+

Mlh1-/-

Mlh1-/-

1000

100

10

We sequenced 86 class II clones and detected point mutations in 77. We further confirmed the increase in point mutation in T cells of Mlh1/ mice by analysis of mutations at the X-linked Hprt locus. Cells with Hprt mutations are recoverable by their resistance to 6thioguanine (6-TG). Mlh1/ mice showed an even greater increase in 6-TG-resistant (6-TGr) T cells compared to DAPr T cells (Table 3). The Hprt mutant frequency is more than 30-fold higher in Mlh1/ mice than in Mlh1 þ / þ mice, suggesting that Hprt is more susceptible to point mutation than Aprt in an Mlh1/ background, probably due to its large size and its 6 G repeat. This finding is again consistent with our previous observations with Pms2/ mice (Shao et al., 2002). As illustrated in Table 1, both class I and class II clones contributed to the increase in frequency of DAPr fibroblasts in Mlh1/ mice. Point mutations in Aprt were detected in three of six class II clones sequenced. We estimated the mutant frequency of 6-TGr skin fibroblasts from Mlh1/ mice. While only a single 6-TGr clone was detected in 23 ears of wild-type mice (pooled MF ¼ 0.6  106), at least one 6-TGr clone was detected in 13 of 18 ears of Mlh1/ mice (pooled MF ¼ 23.3  106). Thus, both MR and point mutation contribute to the increased mutant frequency in fibroblasts of hybrid Mlh1/ mice. It is noteworthy that Mlh1 þ / mice are predisposed to develop lymphomas and tumors of the GI tract (Edelmann et al., 1999), yet they do not have a higher frequency of point mutation in T cells than Mlh1 þ / þ mice. Thus, it appears that the increased propensity of Mlh1 þ / mice to develop lymphomas is not attributable to elevated mutation frequency or increased MR activity, and that loss of wild-type Mlh1 allele is a prerequisite to carcinogenesis. Spectrum of point mutations in Mlh1 and Pms2 null mice

1 C57

(N5)129 X C57

Figure 3 Lack of MLH1 enhances MR in fibroblasts of (N5)129S2  C57BL hybrids. Each point represents the frequency of DAPr fibroblasts from one ear. The bars indicate median values. Points at the bottom represent ears from which no DAPr colonies were recovered

Table 3

MLH1 and PMS2 may or may not be in the same repair complex, depending upon the type of lesion. Therefore, we asked whether the spectrum of point mutation in T cells from mice deficient in MLH1 is similar or different than that in Pms2-deficient mice. In Table 4, we summarize 65 independent Aprt point mutations detected in T cells of Mlh1/ mice. A detailed account of those mutations is given in Supplementary Table 1. Point mutations previously reported for Pms2/ mice (Shao et al., 2002) are also included in Table 4 for comparison. As in Pms2/ mice, the majority of the mutations in Mlh1/ mice were single base substitutions (69%). Frameshift mutations account for 28%. Two

Increased Hprt mutations in T cells of Mlh1/ mice

Genotype

Strain

No. of mice

Cloning efficiency (mean7s.e.)

6-TGr frequency (  106) (mean7s.e.)

6-TGr frequency (  106) (median)

Mlh1+/+ Mlh1+/ Mlh1/ Mlh1/

(N5)129  C57 (N5)129  C57 (N5)129  C57 C57

11 12 11 7

6.370.7 9.471.7 6.671.2 6.271.3

4.171.3 2.971.1 151.2724.6 2307107.8

3.6 1.3 180.8 192.2

Oncogene

Mlh1-mediated suppression of somatic mutation C Shao et al

9021 Table 4 Distribution of point mutations in T cells of Mlh1/ and Pms2/ mice Mlh1/ (%) Transitions T>C A>G C>T G>A

11 3 12 15

(16.9) (4.6) (18.5) (23.1)

C to T, whereas in Mlh1/ mice different base substitutions are more evenly distributed (Table 4).

Pms2/ (%) 26 3 1 6

(55.3) (6.4) (2.1) (12.8)

Transversions Large deletions (>10 bp)

4 (6.2) 2 (3.1)

3 (6.4) 0

2 bp deletions At dinucleotide run Elsewhere

2 (3.1) 0

1 (2.1) 0

1 bp deletions At mononucleotide run Elsewhere

15 (23.1) 0

4 (8.5) 0

1 bp insertions At mononucleotide run Elsewhere Total

1 (1.5) 0 65 (100)

3 (6.4) 0 47 (100)

Figure 4 Distribution of frameshift mutations in five exons of Aprt gene. Exons are shown in upper case, introns (partial) are shown in lower case. Mono- and dinucleotide runs with frameshift mutations are underlined. m, deletion; ., insertion

mutants have deletions larger than 10 bp. Frameshift mutations are clustered exclusively within runs of monoand dinucleotides (Figure 4), consistent with previous observations that mutational hotspots occur at monoand dinucleotide runs in tumor suppressor genes in mismatch repair-deficient cells (Markowitz et al., 1995; Huang et al., 1996; Rampino et al.,1997). However, the distribution of single base substitutions is significantly different between Pms2/ and Mlh1/ mice. In Pms2/ mice, 72% of all transitions are T to C, and only 3% are

Discussion Tissue-specific suppression of MR by chromosomal divergence Our Aprt þ / mouse model has allowed us to determine the frequency and types of the somatic mutations that accumulate in vivo in two isogenic cell types. Previously, we evaluated the effects of chromosomal divergence on MR by comparing interspecific (129S2  SPRET/Ei F1) and intersubspecific hybrids (129S2  CAST/Ei F1) with hybrids of classical inbred strains, for example, 129S2  C3H/HeJ. Of these crosses, interspecific hybrids possess the greatest DNA sequence divergence of chromosome homologs, and thus may be the least permissive for MR. Inbred strains manifest far greater sequence identity of homologs. Hybrids between classical inbred strains, with a low level of chromosome sequence divergence, lie between the two extremes. If suppression of MR is dependent upon the degree of sequence divergence, hybrids between classical inbred strains, though more permissive for MR than interspecific hybrids, should be less permissive than the parental inbred strains. Indeed, we found that fibroblasts from C57BL/6  129S2 F1 mice have a lower frequency of MR than fibroblasts from either of their two parental strains (Figure 1). Thus, low level of sequence divergence has an inhibitory effect on MR, and can lead to reduced LOH. While MR is greatly reduced in fibroblasts from interspecific hybrids and less so from interstrain hybrids, a reduction of MR in T cells was detectable only in 129S2  SPRET/Ei F1 hybrids (Shao et al., 2001). Thus, MR in T cells, compared to fibroblasts, appears to be less inhibited by chromosomal DNA divergence. The mechanism(s) underlying this tissue-specific suppression of MR between homeologous chromosomes is unknown. It has been shown that the proximity between chromosome regions is required for MR and chromosomal translocation (Nikiforova et al., 2000; Haigis and Dove, 2003; Roix et al., 2003; Tischfield and Shao, 2003), and the positioning of chromosomes in a nucleus may vary between cell types (Nikiforova et al., 2000; Roix et al., 2003). Thus, the possibility exists that this tissue-specific phenotype is caused by differences in relative positioning of chromosome 8 homologs in the nuclei in different tissues. If the two copies of chromosome 8 (where Aprt is located) are positioned farther apart at certain stages of T-cell development, they may have a reduced chance of physical contact, and consequently a reduced frequency of recombination regardless of the degree of sequence homology. Another possibility is that the enzyme repertoire for the repair of DNA strand breaks varies between tissues so that nonhomologous end joining is preferred over homologous recombination in certain cell types. Oncogene

Mlh1-mediated suppression of somatic mutation C Shao et al

9022

Elevation of point mutation and MR in the absence of MLH1 MLH1 is a key component of the MMR machinery that plays a primary role in genome maintenance. MLH1 mutations are observed in a large proportion of cases of human hereditary nonpolyposis colorectal carcinoma (HNPCC) (Peltomaki and Vasen, 1997). Epigenetic silencing by hypermethylation of the MLH1 promoter is frequently found in sporadic colon cancers (Herman et al., 1998; Veigl et al., 1998). The correlation between lack of MMR and higher tumor incidence implicates a higher likelihood of tumor suppressor gene mutations as the mechanism underlying tumorigenesis (Loeb, 1991). Indeed, frameshift mutations have been identified in several tumor suppressor genes in MMR-deficient cancer cells (Markowitz et al., 1995; Huang et al., 1996; Rampino et al., 1997). Previous studies of bacteria, yeast, and mammalian cells also indicate that MMR proteins play a key role in restricting recombination between divergent DNA sequences. MMR-deficient yeast showed an increased frequency of homeologous recombination (Harfe and Jinks-Robertson, 2000). Gene-targeting experiments showed that a higher recombination frequency was achieved with Msh2-deficient ES cells than with wildtype cells using vectors containing mismatches (de Wind et al., 1995; Elliott and Jasin, 2001). Consistent with those findings, we observed an increase of MR between divergent homologs in fibroblasts of Mlh1/ mice. However, in contrast to reports of increased point mutation, there has been no report indicating an increased MR leading to LOH at tumor suppressor gene loci in tumors of HNPCC patients or MMRdeficient mice. Several factors may contribute to this discrepancy. First, not all MMR genes play a role in regulating MR. For example, we observed a significant increase in base substitution and frameshifts in hybrid Pms2/ mice, but found no increase in MR (Shao et al., 2002). This observation is consistent with the report that enhancement of recombination between divergent DNA sequences was much less in yeast pms1 (equivalent of Pms2 in mammals) mutants compared with mlh1 or msh2 mutants (Datta et al., 1996). Thus, MMR genes each have distinctive roles in regulating recombination. Second, the increase in MR in the absence of MMR may be less dramatic than the increase in point mutation, especially in cases where the gene is large and is located near the centromere. The Aprt gene is relatively small, about 3 kb in genomic sequence, and it is located near the telomere, which may skew the mutational spectrum toward MR. In spite of that, the frequency of fibroblast mutants derived from MR is only sixfold higher in hybrid Mlh1/ mice than that in Mlh1 þ / þ mice. In sharp distinction, Hprt mutants in Mlh1/ mice, which are caused by point mutations, occur at 30–40-fold greater frequency compared to Mlh1 þ / þ mice. Thus, a modest increase in MR in the absence of MMR may be masked by a more dramatic increase in point mutation. Third, defects in homologous recombination repair (HRR) genes in MMR-deficient tumors may disrupt Oncogene

the process of MR. Several MMR-deficient tumor cell lines were found to be defective in HRR, and one was demonstrated to be due to a frameshift mutation in XRCC2 (Mohindra et al., 2002). Cells with defective HRR may undergo rapid karyotypic evolution, thus acquiring a growth advantage and predominance in the tumor cell population as carcinogenesis progresses. Fourth, as we have shown in this study, the restriction of recombination by MMR may be limited to certain tissues. The spectrum of Apc gene mutation in intestinal adenomas of B6Mlh1/ mice has been characterized (Kuraguchi et al., 2000; Shoemaker et al., 2000). In a sharp contrast to intestinal adenomas of B6 Apcmin Mlh1 þ / þ mice where Apc þ was predominantly lost by MR, a great majority of the tumors in Apcmin Mlh1/ mice were caused by intragenic mutations in Apc (Kuraguchi et al., 2000; Shoemaker et al., 2000). Interestingly, most frameshifts in Apc are clustered at dinucleotide or mononucleotide runs (Kuraguchi et al., 2000), very much resembling the spectrum of Aprt mutations that we observed in the DAPr T cells of Mlh1/ mice. As the two homologs are identical in the B6 Apcmin mice, it was impossible to determine whether MLH1 had mediated a suppression of MR between divergent homologs in the intestinal cells. Nevertheless, the increase in point mutations in the absence of Mlh1 is so robust that MR, even if it is increased in the absence of Mlh1, may be insignificant as a contributing factor to the increased tumorigenesis. Differential activity of MLH1 and PMS2 proteins We and others reported that T to C transitions predominate in DAPr mutants in Pms2-/- mice (Shao et al., 2002; Shin and Turker, 2002; Shin et al., 2002), accounting for 72% of all base substitutions in DAPr T cells (Shao et al., 2002). Here we showed a relatively even distribution of all mutation types in Mlh1/ mice (Table 4). Such a difference, together with evidence for different roles in regulating MR between homeologous sequences, suggests functional differences between the two members of a heterodimer required for full MMR. The pleiotropic phenotypes displayed by Mlh1/ mice (Edelmann et al., 1996; Prolla et al., 1998), including a high incidence of a broader spectrum of tumors, a high mutation rate, and meiotic defects in both sexes, indicates a more general role of MLH1 compared to some other MMR genes. Our data are consistent with a recent study in which yeast Mlh1 and Pms1 (the homolog of mammalian Pms2) are each shown to have different DNA-binding activities in the absence of dimerization (Hall et al., 2003). While substitutions of conserved amino acids in MLH1 result in a mutator phenotype in haploid cells and strongly reduce DNA binding by the MLH1, and by MLH1–PMS1 heterodimer, similar substitutions in Pms1 have little effect, suggesting that MLH1 and PMS1 differ in their interactions with DNA. It has also been shown that the absence of MLH1 can abolish MutL enzyme activity, while the absence of PMS2 produces only a

Mlh1-mediated suppression of somatic mutation C Shao et al

9023

partial MutL reduction (Yao et al., 1999). It is possible that different MutL heterodimeric complexes may each have a preference for specific mismatches such that PMS2 complexed with MLH1 may be more specialized for the correction of T:G mismatches. We showed that the types of somatic mutation and the degree to which each accumulates in the absence of MLH1 are remarkably different between two isogenic cell types. While T cells primarily utilize the MMR system to correct base pair mismatches during DNA replication, fibroblasts apparently rely on the same system both for the correction of mismatches and for the suppression of homeologous recombination. Interestingly, embryonic stem cells exhibit a significant reduction in the frequency of both MR and point mutations compared to their differentiated fibroblast derivatives (Cervantes et al., 2002). All these findings point to the necessity for a more comprehensive approach towards investigating mutagenesis in different mammalian cell types in vivo.

Materials and methods Mice We maintained 129S2 Aprt mutant mice in our colony (Engle et al., 1996; Stambrook et al., 1996). The Aprt þ / mice of the congenic C57BL/6 strain were generated by backcrossing 129S2 Aprt þ / to C57BL/6 for 12 generations. Assay of polymorphic SSR markers indicated that the chromosomes 8 are homozygous for C57BL/6 markers for the proximal 60 cM and for the most telomeric loci D8Mit156 and D8Mit56 in

(N12) C57BL/6 mice. Mlh1 þ / mice of congenic C57BL/6 strain are described elsewhere (Edelmann et al., 1996). We backcrossed C57BL/6 Mlh1 þ / mice to 129S2 for five generations to obtain (N5)129S2 Mlh1 þ /Aprt þ / (or Aprt/). 129S2  C57BL/6 Aprt þ / mice with different genotypes of Mlh1 ( þ / þ , þ /, /) were obtained by crossing (N5)129S2 Mlh1 þ /Aprt þ / (Aprt/) to C57BL/6Mlh1 þ / mice. Some 129S2  C57BL/6 hybrid mice were also derived from crosses between C57BL/6 wild-type mice and 129S2 Aprt/ mice. Isolation and molecular characterization of mutant cell clones The DAPr mutant fibroblast clones were isolated as described (Shao et al., 1999). 6-TGr mutant fibroblast clones were recovered as described (Shao et al., 2002). DAPr and 6-TGr Tcell clones were isolated as described (Shao et al., 2000). The DAPr clones were first divided into two classes by absence (class I) or physical retention (class II) of the original Aprt þ (untargeted) allele (Shao et al., 1999). Class I clones were further characterized with SSR markers along chromosome 8 to determine their mechanism(s) of origin. Clones that exhibited LOH at loci distal to Aprt but remained heterozygous at loci proximal to Aprt were interpreted to be derivatives of MR, as was previously demonstrated more rigorously (Shao et al., 1999). We sequenced the DNA of all five Aprt exons, introns 1, 3 and 4, and part of the promoter region of class II clones for detection of point mutation. Clones with the same point mutation were considered to be sibs for purposes of analysis. Acknowledgements This work is supported by US NIH Grant R01ES011633 to JAT.

References Cavenee WK, Dryja TP, Phillips RA, Benedict WF, Godbout R, Gallie BL, Murphree AL, Strong LC and White RL. (1983). Nature, 305, 779–784. Cervantes RB, Stringer JR, Shao C, Tischfield JA and Stambrook PJ. (2002). Proc. Natl. Acad. Sci. USA, 99, 3586–3590. Datta A, Adjiri A, New L, Crouse GF and Jinks-Robertson S. (1996). Mol. Cell. Biol., 16, 1085–1093. DeBerardinis RJ, Goodier JL, Ostertag EM and Kazazian HH. (1998). Nat. Genet., 20, 288–290. de Wind N, Dekker M, Berns A, Radman M and te Riele H. (1995). Cell, 82, 321–330. Dietrich WF, Miller JC, Steen RG, Merchant M, Damron D, Nahf R, Gross A, Joyce DC, Wessel M, Dredge RD, Marquis A, Stein LD, Goodman N, Page DC and Lander ES. (1994). Nat. Genet., 7, 220–245. Edelmann W, Cohen PE, Kane M, Lau K, Morrow B, Bennett S, Umar A, Kunkel T, Cattoretti G, Chaganti R, Pollard JW, Kolodner RD and Kucherlapati R. (1996). Cell, 85, 1125–1134. Edelmann W, Yang K, Kuraguchi M, Heyer J, Lia M, Kneitz B, Fan K, Brown AM, Lipkin M and Kucherlapati R. (1999). Cancer Res., 59, 1301–1307. Elliott B and Jasin M. (2001). Mol. Cell. Biol., 21, 2671–2682. Engle SJ, Stockelman MG, Chen J, Boivin G, Yum MN, Davies PM, Ying MY, Sahota A, Simmonds HA, Stambrook PJ and Tischfield JA. (1996). Proc. Natl. Acad. Sci. USA, 93, 5307–5312.

Haigis KM, Caya JG, Reichelderfer M and Dove WF. (2002). Proc. Natl. Acad. Sci. USA, 99, 8927–8931. Haigis KM and Dove WF. (2003). Nat. Genet., 33, 33–39. Hall MC, Shcherbakova PV, Fortune JM, Borchers CH, Dial JM, Tomer KB and Kunkel TA. (2003). Nucleic Acids Res., 31, 2025–2034. Harfe BD and Jinks-Robertson S. (2000). Annu. Rev. Genet., 34, 359–399. Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, Markowitz S, Willson JK, Hamilton SR, Kinzler KW, Kane MF, Kolodner RD, Vogelstein B, Kunkel TA and Baylin SB. (1998). Proc. Natl. Acad. Sci. USA, 95, 6870–6875. Huang J, Papadopoulos N, McKinley AJ, Farrington SM, Curtis LJ, Wyllie AH, Zheng S, Willson JK, Markowitz SD, Morin P, Kinzler KW, Vogelstein B and Dunlop MG. (1996). Proc. Natl. Acad. Sci. USA, 93, 9049–9054. Kuraguchi M, Edelmann W, Yang K, Lipkin M, Kucherlapati R and Brown AM. (2000). Oncogene, 19, 5755–5763. Liang L, Shao C, Deng L, Mendonca MS, Stambrook PJ and Tischfield JA. (2002). Mutat. Res., 502, 69–80. Loeb LA. (1991). Cancer Res., 51, 3075–3079. Luo G, Santoro IM, McDaniel LD, Nishijima I, Mills M, Youssoufian H, Vogel H, Schultz RA and Bradley A. (2000). Nat. Genet., 26, 424–429. Markowitz S, Wang J, Myeroff L, Parsons R, Sun L, Lutterbaugh J, Fan RS, Zborowska E, Kinzler KW and Vogelstein B. (1995). Science, 268, 1336–1338. Oncogene

Mlh1-mediated suppression of somatic mutation C Shao et al

9024 Mohindra A, Hays LE, Phillips EN, Preston BD, Helleday T and Meuth M. (2002). Hum. Mol. Genet., 11, 2189–2200. Nikiforova MN, Stringer JR, Blough R, Medvedovic M, Fagin JA and Nikiforov YE. (2000). Science, 290, 138–141. Peltomaki P and Vasen HF. (1997). Gastroenterology, 113, 1146–1158. Prolla TA, Baker SM, Harris AC, Tsao JL, Yao X, Bronner CE, Zheng B, Gordon M, Reneker J, Arnheim N, Shibata D, Bradley A and Liskay RM. (1998). Nat. Genet., 18, 276–279. Rampino N, Yamamoto H, Ionov Y, Li Y, Sawai H, Reed JC and Perucho M. (1997). Science, 275, 967–969. Roix JJ, McQueen PG, Munson PJ, Parada LA and Misteli T. (2003). Nat.Genet., 34, 287–291. Shao C, Deng L, Henegariu O, Liang L, Raikwar N, Sahota A, Stambrook PJ and Tischfield JA. (1999). Proc. Natl. Acad. Sci. USA, 96, 9230–9235. Shao C, Deng L, Henegariu O, Liang L, Stambrook PJ and Tischfield JA. (2000). Proc. Natl. Acad. Sci. USA, 97, 7405–7410. Shao C, Stambrook PJ and Tischfield JA. (2001). Nat. Genet., 28, 169–172. Shao C, Yin M, Deng L, Stambrook PJ, Doetschman T and Tischfield JA. (2002). Oncogene, 21, 2840–2845.

Shin CY and Turker MS. (2002). DNA Repair (Amst), 1, 995–1001. Shin CY, Mellon I and Turker MS. (2002). Oncogene, 21, 1768–1776. Shoemaker AR, Haigis KM, Baker SM, Dudley S, Liskay RM and Dove WF. (2000). Oncogene, 19, 2774–2779. Stambrook PJ, Shao C, Stockelman M, Boivin G, Engle SJ and Tischfield JA. (1996). Environ. Mol. Mutagen., 28, 471–482. Tischfield JA and Shao C. (2003). Nat. Genet., 33, 5–6. Veigl ML, Kasturi L, Olechnowicz J, Ma AH, Lutterbaugh JD, Periyasamy S, Li GM, Drummond J, Modrich PL, Sedwick WD and Markowitz SD. (1998). Proc. Natl. Acad. Sci. USA, 95, 8698–8702. Wade CM, Kulbokas III EJ, Kirby AW, Zody MC, Mullikin JC, Lander ES, Lindblad-Toh K and Daly MJ. (2002). Nature, 420, 574–578. Wijnhoven SW, Sonneveld E, Kool HJ, van Teijlingen CM and Vrieling H. (2003). Carcinogenesis, 24, 139–144. Yao X, Buermeyer AB, Narayanan L, Tran D, Baker SM, Prolla TA, Glazer PM, Liskay RM and Arnheim N. (1999). Proc. Natl. Acad. Sci. USA, 96, 6850–6855.

Supplementary Information accompanies the paper on Oncogene website ( http://www.nature.com/onc).

Oncogene