Minichromosome maintenance helicase paralog MCM9 is dispensible for DNA replication but functions in germ-line stem cells and tumor suppression

Abstract

Effective DNA replication is critical to the health and reproductive success of organisms. The six MCM2–7 proteins, which form the replicative helicase, are essential for high-fidelity replication of the genome. Many eukaryotes have a divergent paralog, MCM9, that was reported to be essential for loading MCM2–7 onto replication origins in the Xenopus oocyte extract system. To address the in vivo role of mammalian MCM9, we created and analyzed the phenotypes of mice with various mutations in Mcm9 and an intronic DNA replication-related gene Asf1a. Ablation of Mcm9 was compatible with cell proliferation and mouse viability, showing that it is nonessential for MCM2–7 loading or DNA replication. Mcm9 mutants underwent p53-independent embryonic germ-cell depletion in both sexes, with males also exhibiting defective spermatogonial stem-cell renewal. MCM9-deficient cells had elevated genomic instability and defective cell cycle reentry following replication stress, and mutant animals were prone to sex-specific cancers, most notably hepatocellular carcinoma in males. The phenotypes of mutant mice and cells suggest that MCM9 evolved a specialized but nonessential role in DNA replication or replication-linked quality-control mechanisms that are especially important for germ-line stem cells, and also for tumor suppression and genome maintenance in the soma.

The eukaryotic Mcm gene family consists of eight genes, Mcm2 to -7, and in a subset of organisms, Mcm8 and Mcm9. These genes encode a highly conserved, ∼200-amino acid “MCM domain” containing Walker A and Walker B AAA+ ATPase motifs. MCM2 to -7 (hereafter, MCM2-7) form the DNA replicative helicase (1–3). The activities of MCM2-7 are controlled to prevent rereplication during a cell cycle (4). In both yeast and mice, each of the MCM2-7 proteins are essential, despite having structural similarity (5, 6). Dysfunction of, or decreases in MCMs can cause genome instability and cancer (6–8).

Mcm8 and Mcm9 are both present in diverse eukaryotes, although yeast lack them and Drosophila melanogaster has only Mcm8 (9, 10). Drosophila Mcm8 is involved in meiotic recombination (9), and human MCM8 was suggested to enable recruitment of CDC6 to replication origins (11). CDC6 is needed for loading MCM2-7 onto replication origins, and formation of the prereplication complex (pre-RC).

Mouse Mcm9 was first described as encoding a 386 amino acid, C-terminal–truncated paralog of the MCM2-7 proteins (12). This isoform (MCM9^S) contains a partial MCM domain (Fig. 1) containing the Walker A-type ATPase motif, but not a Walker B domain. A larger form (MCM9^L) was predicted to contain both the Walker A and B motifs required for helicase activity in MCM2-7 (13). However, the canonical Walker A motif in MCMs is GDP[G/S]×[S/A]KS, but in MCM8 and -9 it is GDPG[L/T]GKS. The latter is more similar to the canonical AAA+ ATPase consensus of GxxGxGK[S/T]. The MCM9 Walker B motif is also atypical (CCIDEFNSL compared with the more canonical motif of MCM2-7 [C/V][C/L]IDEFDKM).

Fig. 1.

Mcm9 locus structure and isoforms. (A) Mcm9 has 2 isoforms: Mcm9^s and Mcm9^L. The former uses an alternative splice site in exon 7 that results in termination. Triangles depict the insertion sites of the gene traps. XE518 created a deletion that includes Asf1a. Locations of polymorphic SNPs used for measuring deletion size in the Mcm9^XE518 allele are indicated. These are: (Left to Right) rs51772485, rs46087546, rs51382030, rs51176035, rs47659858, and rs49316622. The Asf1a gene trap is located 640-bp downstream of exon 2. (B) MCM9 isoforms and the gene-trap alleles. MCM9 contains a zinc finger, CDK site, and the MCM domain. (C) Tissue expression of Mcm9 and Asf1a by semi-qRT-PCR. Primer pairs used in the top four panels are sets 1 to 4, respectively, as shown in A. (D) qRT-PCR analysis of Mcm9 and Asf1a in mutant MEFs (n = 3). AWO, Mcm9^AWO655; XG, Mcm9^XG743. Error bars indicate SD.

The first major functional study of MCM9 reported that in Xenopus egg extracts, MCM9 interacts with CDT1 to load MCM2-7 onto replication origins, and also counteracts the inhibitory effects of Geminin upon CDT1 for replication licensing (14). Thus, MCM9 was deemed essential for pre-RC formation and DNA replication. Here, we report that Mcm9 is not required for pre-RC formation or DNA replication in mice, but is important for germ-line stem-cell maintenance and proliferation, genome stability, and cancer prevention.

Results

Genomic Structure of the Mouse Mcm9 Locus and mRNA Isoforms.

Although RefSeq and published Mcm9 gene models are available (12, 13), we performed additional bioinformatic analyses and targeted RT-PCR to derive the correct structure, as shown in Fig. 1 (see also Fig. S1 and SI Results). Mcm9 produces two alternative C-terminal mRNA isoforms by alternative use of a splice site within exon 7. Skipping this site results in a short isoform (Mcm9^S) encoding a protein of 386 aa. Splicing to exon 8 creates a longer isoform (Mcm9^L), encoding an 1,134-aa protein (Fig. 1A). MCM9^s contains only half of the conserved MCM domain containing the Walker A motif (Fig. 1B).

The seventh intron contains another gene, Asf1a (Antisilencing function 1 homolog A), with a transcriptional orientation antisense to Mcm9 (Fig. 1A). ASF1a is a histone chaperone with diverse functions, including the regulation of replication fork progression under normal and replication stressed conditions (15, 16). Evidence suggests it does so by interacting with MCM2-7 via a histone H3-H4 bridge to transfer histones from dissociated nucleosomes to newly forming nucleosomes in the wake of the replication fork (15).

Mcm9 Isoforms Are Expressed Ubiquitously in Mice.

Expressed sequence tag (EST) data in Unigene show that Mcm9 is expressed in many tissues, but does not distinguish between the two isoforms. RT-PCR analysis of Mcm9^S and Mcm9^L revealed expression of both in all tissues and through early development (Fig. 1C). Levels of transcription in the gene body were assessed by analyzing GRO-Seq (global run-on sequencing) data from ES cells and mouse embryonic fibroblasts (MEFs) (17). This method determines the amount and locations of engaged RNA polymerase on a gene at the time of the assay (18). Run-on transcripts are produced across the entire Mcm9 gene body. The expression levels of mature transcripts indicate that the amounts of each isoform produced are similar, based on the exon sequence counts of Mcm9^S and Mcm9^L (19) (Fig. S2). Asf1a had similar amounts of engaged RNA pol II (Fig. S2).

MCM9 Is Not Required for DNA Replication.

To determine the in vivo function of the MCM9 isoforms, we generated mice containing three different gene-trap insertions within Mcm9 (Fig. 1 A and B). We also generated a mutant allele of Asf1a (Asf1a^ALO673) to explore potential regulatory or functional relationships between the two genes.

The Mcm9^XE518 insertion actually caused a deletion beginning downstream of Mcm9 exon 4, extending through the entire Asf1a gene, and terminating upstream of the final Mcm9 exon (Fig. 1A). Thus, Mcm9^XE518 is null for Asf1a and presumably Mcm9. Intercrosses of Mcm9^XE518/+ mice failed to produce newborn homozygotes (P << 0.001 by χ²) Table S1), indicating the allele is embryonically lethal. Timed matings revealed that Mcm9^XE518/XE518 embryos were viable at embryonic day (E) 9.5 but were smaller and developmentally arrested at a stage resembling E8.5 embryos (Fig. 2B). These embryos contained no detectable Asf1a mRNA or Mcm9 transcripts downstream of the gene-trap insertion site (Fig. 2E).

Fig. 2.

Midgestation lethality in MCM9- and ASF1A-deficient embryos. (A–D) E9.5 embryos. (Scale bars, 500 μM.) (E) RT-PCR of E9.5 embryos. Primers for Mcm9^S and Mcm9^L are sets 2 and 3, respectively, from Fig. 1A. Primers (XE518WTF and XE518GTR) for Mcm9^XE are specific for the gene-trap fusion transcript. (F) RT-PCR/sequencing of polymorphic coding SNPs (rs51382030 for Asf1a; rs51772485 for Mcm9). The WT allele here is C3H, which is identical to the parental 129-based gene trap chromosome. XE, Mcm9^XE518 ; AL, Asf1a^AL0673.

Because Asf1a is also disrupted in the Mcm9^XE518 allele, the lethal phenotype could be a result of mutation of either gene or a combination of both. To distinguish between these possibilities, we created mice bearing the Asf1a^ALO673 gene-trap allele, which is predicted to truncate 129 C-terminal amino acids from the 209-aa protein. Homozygotes died at midgestation, with a phenotype similar to Mcm9^XE518/XE518 embryos (Fig. 2D). The phenotype is likely caused by loss of Asf1a alone because normal Mcm9 transcripts are produced in cis from the Asf1a^ALO673 allele (Fig. 2F). As expected, Mcm9^XE518 failed to complement the lethality of Asf1a^ALO673 (Fig. 2C and Table S1). These data demonstrate that neither gene is required for cell division (and presumably DNA replication) during the first half of mouse gestation, a period of dramatic cell proliferation. However, these data do not reveal the developmental role of MCM9, as the lethal phenotype of Asf1a in the Mcm9^XE518 allele obscures or precludes potential contemporaneous or subsequent roles.

Because ASF1 physically associates with MCM2-7 and facilitates nucleosome reassembly during DNA replication and DNA repair (15, 20), processes that may overlap with MCM9, the question arises as to whether the physical arrangement of these two genes (which is conserved in mammals and chickens but apparently not zebrafish, based on current genome assemblies) is entirely coincidental. It is possible that there is a regulatory relationship between the genes, either with respect to their sharing common enhancer elements or modulators of chromatin structure, or in still more complex ways involving the complementarity of their transcripts or the interplay of the mechanics of divergent transcription complexes. However, transcription of Mcm9 through the Asf1a gene is not required for Asf1a transcription in cis (Fig. 1D), and intact Asf1a is not needed for Mcm9 transcription (Fig. 2F). These results indicate that if there is indeed a regulatory relationship between the genes, it is likely to be related to sharing control sequences, identical regional chromosomal structure, or posttranscriptional processing.

Mice with Severe Depletion or Elimination of Mcm9 Are Viable.

To identify the in vivo roles of each Mcm9 isoform, two other gene-trap alleles were used that do not ablate Asf1a. Mcm9^XG743 is located downstream of exon 7 and Asf1a mRNA levels are ∼75% of WT (Fig. 1D). Mcm9^XG743 is predicted to allow production of the entire MCM9^S polypeptide, either as a fusion to β-geo or as the endogenous protein (Fig. 1B). A slight amount (<3%) of mRNA was detected that corresponds to splicing over the gene trap (exon 7–8 products). Mcm9^S and fusion transcripts were also reduced from WT levels (Fig. 1D). These results indicate that Mcm9^XG743 is severely hypomorphic or null for Mcm9^L, and potentially hypomorphic for Mcm9^S. Heterozygote intercrosses produced Mcm9^XG743/XG743 offspring at the expected Mendelian ratio (P = 0.98) (Table S1). The Mcm9^AW0655 allele, which effectively terminates transcripts upstream of the entire MCM domain (Fig. 1 A, B, and D), thus appears null for Mcm9 but leaves Asf1a transcription intact (Fig. 1D). Mcm9^{AWO655/AWO655} and Mcm9^AWO655/XE518 animals are completely viable (Table S1). Therefore, MCM9 is unnecessary for DNA replication.

MCM9 Deficiency Causes Germ-Cell Loss.

Despite being grossly normal, Mcm9^XG743 male homozygotes had markedly smaller testes than WT littermates, a differential that was evident at puberty and widened over time (Fig. 3A). Despite a consequent 67% decrease in epididymal sperm concentration at 12 wk of age, these mutant males were fertile. Testis histology revealed three notable seminiferous tubule abnormalities (Fig. 3 B–D). The most striking abnormality was progressive germ-cell depletion. The second was the presence of seminiferous tubule sections (∼5%) lacking spermatogonia (the adult germ-line stem cells), but containing differentiated meiotic spermatocytes or postmeiotic spermatids (Fig. 3D, arrows). The third class of abnormal tubule cross sections (5%) contained a cohort of spermatocytes arrested in meiosis (Fig. 3D, Inset). These abnormalities were also observed in Mcm9^AWO655 male homozygotes and compound heterozygotes (Fig. S3). Notably, mice with severe depletion of MCM2-7 are fertile and do not exhibit these germ-cell loss phenotypes (Fig. S3) (6, 8), suggesting that the germ-cell effects in Mcm9 mutants likely occur by mechanisms distinct from MCM2-7 helicase defects.

Fig. 3.

Mcm9 mutations cause germ-cell depletion and loss of spermatogonial stem cells. (A and B) Quantification of testis weights and histological abnormalities. *Significantly different from WT; **significant decrease in mutant testis size over time (n = 10). Abbreviations in B for seminiferous tubule cross sections that contain or exihibit the following: GMC, giant multinucleated cells; MEI, meiotic arrest; MW, missing wave of spermatogenesis; GCD, germ-cell depletion; Norm, normal. (C and D) H&E-stained histological sections of testes from 12-wk-old mice of the indicated genotypes. WT, wild-type; XG, Mcm9^XG743 (magnification: 200×). Error bars indicate SD. Seminiferous tubules marked with an asterisk (*) are devoid of germ cells; those marked with a caret (^) exhibit meiotic arrest (Inset); and those indicated by arrows are depleted of spermatogonia but undergoing a final wave of spermatogenesis. Circled tubules are examples of normal spermatogenesis. (E and F) H&E-stained histological sections of ovaries from 24-wk-old mice of the indicated genotypes. (G–L) Immunofluorescence of 1-dpp testes (G and H) and ovaries (J and K) of the indicated genotypes (magnification: 200×). MVH (green) stains germ cells and DAPI stains nuclei (blue). (I and L) Germ-cell counts from MVH staining data at 1 dpp (n = 3). *Significant difference vs. WT (see Materials and Methods). XG, Mcm9^XG743/XG743; p53, Trp53^−/−; p21, Cdkn1a^−/−. Error bars indicate SD.

Young Mcm9^XG743/XG743 females were fertile but also exhibited germ-cell loss. Mutant 6- to 12-wk ovaries had fewer total follicles and were nearly devoid of primordial follicles. Oocytes were almost completely absent at 24 wk (Fig. 3 E and F). Most Mcm9^{AWO655/AWO655} females were infertile and exhibited ovarian hyperplasias and tubulostromal adenomas consistent with premature ovarian failure (see below).

The sex-independent shortfall of germ cells in peripubertal animals is suggestive of a defect early in the germ lineage, possibly during the expansion or establishment of germ-line stem-cell pools during embryonic development. To test this, control and Mcm9^XG743/XG743 gonads were serially sectioned and probed with the germ-cell–specific marker MVH (mouse vasa homolog). This process revealed a 72% decrease of gonocytes in nascent seminiferous tubules (Fig. 3 G–I), and 82% fewer oocytes in newborn ovaries (Fig. 3 J–L). Therefore, the shortfall in germ cells occurred during gestation. To determine if germ-cell loss occurs via TRP53-pathway–mediated elimination, Mcm9^XG743/XG743 Trp53^−/− and Mcm9^XG743/XG743 Cdkn1a^−/− (Cdk1na = p21) animals were bred and their germ-cell numbers were scored at 1 d postpartum (dpp). There was no rescue of germ-cell depletion in either of these compound mutants (Fig. 3 I and L). Curiously, TRP53 deficiency actually further decreased the numbers of germ cells in Mcm9^XG743/XG743 newborn males but not females (Fig. 3 I and L).

MCM9 Deficiency Does Not Impact MCM2-7 Chromatin Loading, Has Minimal Impact on Genome Stability, but Predisposes to Cancer.

Mutations reducing MCM2-7 levels cause genomic instability and cancer susceptibility in mice (7, 8) as a consequence of decreased pre-RC formation (6, 21, 22). Because it was reported that MCM9 is essential for MCM2-7 loading onto replication origins and thus pre-RC formation in Xenopus oocyte extracts (14), we tested whether MCM9 depletion impacts MCM2-7 chromatin loading in Mcm9 ^{AWO655/AWO655} MEFs. No difference in chromatin-bound MCM2, MCM4, and MCM7 was observed (Fig. 4 A and B). Furthermore, in contrast to Xenopus oocyte extracts immunodepleted for MCM9, no decrease of CDT1 was observed in mutant MEFs (Fig. 4 A and B).

Fig. 4.

Loss of MCM9 does not alter MCM2-7 or CDT1 levels, however, leads to mild genomic instability and cell-cycle defects under replication stress. (A) Western blots of mutant and WT MEFS showing detergent soluble vs. chromatin bound levels of indicated proteins. (B) Quantification of data in A. (C) Micronucleus levels in erythrocytes. (D) Chromatid breaks in mutant MEFS. (E) Mcm9 mutant MEFs undergo premature senescence. The y-axis values were taken every 3 d upon passage. (F) Mutant MEFs exhibit a delay in cell-cycle entry following APH-induced replication stress. Primary data are in Fig. S3. MEFs were serum-starved to synchronize at G0/G1, serum was then added, then measurements were taken. AW, Mcm9^AWO655; XG. Mcm9^XG743. The P value is based on t testing. UN, measurements 20 h after serum addition, but untreated with APH; 20 hr APH, same as previous, but with APH treatement; 3hr Rel or 24hr Rel, same as previous, but 3 h or 24 h after APH was removed from media (n = 3). Error bars indicate SD.

To test for somatic chromosome instability, we measured peripheral blood micronucleus levels, an indicator of chromosome instability that is elevated in Mcm2-7–deficient mice (6, 8). Relative to WT, micronucleus levels were marginally higher in Mcm9^XG743/XG743 but not Mcm9^{AWO655/AWO655} mice (Fig. 4C). We also observed a 1.6-fold increase in metaphase chromatid breaks in Mcm9^XG743/XG743 MEFs (Fig. 4D).

Although Mcm9 disruption had no apparent impact on MCM2-7 homeostasis in MEFs and only slightly increased chromosomal instability, we aged Mcm9 homozygotes, compound heterozygotes, and control siblings to assess potential long-term health consequences. All mutant males developed tumors by 1 y of age (vs. 32–33% of heterozygote and WT controls) (Table 1). The most remarkable difference compared with controls was in the incidence of hepatocellular carcinoma (HCC; 8 of 13 mutants vs. 0 of 28 controls). Furthermore, eight of eight affected animals had multiple HCCs (Table S2). MCM9-deficient females were prone to ovarian tumors (10 of 22 vs. 0 of 12 in controls) by 1 y of age (Table 1).

Table 1.

Tumor incidence in MCM9-deficient mice

Genotype	Age (wk)	#	N	HCC (males)/OvT (females)	HA (males)/OvH (females)	Oth	Tumor-free
Males
WT	40–55	9	≥N5	0	2	1	67%
Mcm9GT/+	40–55	19	≥N4	0	6	1	68%
Mcm9GT/GT	40–55	13	≥N4	8	4	1	0%
Females
Controls	31–66	12	≥N2	0	1	1	92%
Mcm9GT/GT	32–56	7	≥N2	4	2*	1	29%
Mcm9GT/GT	9–25	15	<N3	6	3	2	47%

 N = number of backcross generations into strain C3H; HCC, hepatocellular carcinoma; HA, hepatocellular adenoma; Oth, other tumors; OvT, ovarian tumor; OvH, ovarian hyperplasia; WT, wild-type; GT, Mcm9^XE518, Mcm9^XG743, or Mcm9^AWO655; #, number of mice. *OvH in contralateral ovary to tumor. In females, “Controls” include both heterozygotes and WT animals. Data on all individuals are presented in Table S2.

MCM9 Deficiency Does Not Perturb the Cell Cycle, but Does Delay Cell Cycle Reentry Following Replication Stress.

To determine if Mcm9 deficiency alters the cell cycle as does depletion of MCM2-7 (8, 23), Mcm9^XG743/XG743 primary MEFs (made from C3H congenic embryos) were evaluated under normal or replication stressed conditions. Continuous cultures of early passage cells were no different from WT with respect to proliferation, BrdU incorporation, and cell-cycle profile (Fig. S4 A–E). However, the mutant cultures underwent senescence prematurely (by passage 5), ultimately producing immortalized survivors (Fig. 4E). When Mcm9^XG743/XG743 and WT MEFs were synchronized at G0/G1 by serum starvation, then released (by serum addition) in the presence of aphidicolin (APH), APH-treated (but not untreated) mutant cells exhibited a delay in progression from G0/G1 through S-phase (Fig. 4F and Fig. S4G). These results confirm that MCM9 is dispensable for unperturbed DNA replication in mouse cells, but its absence accelerates senescence and confers susceptibility to DNA replication stress.

Discussion

Mcm9 encodes at least two isoforms; however, the collection of mutant alleles reported here shows that neither is necessary for pre-RC formation or DNA replication in mice or mouse cells. This result was surprising, given a report that MCM9 is essential for loading MCM2-7 and thus formation of pre-RCs and DNA replication in a Xenopus egg extract system (14). In this widely-used model, sperm DNA added to the extract gets converted into chromatin, and a single round of replication ensues in a putatively physiological manner. That study (14) found that MCM9 binds to chromatin in an origin recognition complex-dependent manner, and by virtue of interacting directly with the essential helicase loading factor CDT1, enables MCM2-7 assembly into pre-RCs. Evaluation of MCM9's role was determined by immunodepletion, and supplementation of immunodepleted extracts with in vitro-produced Mcm9 mRNA partially rescued MCM2-7 loading and the block in DNA replication.

We can offer theories on these contradictory data. First, amphibians and mammals may differ in the need for MCM9 in DNA replication, such that whereas it is essential in Xenopus, it has acquired a more specialized (e.g., germ-line stem cells) but nonessential role in mammals. In this regard, it should not be overlooked that the Xenopus oocyte is a specialized cell. Many eukaryotes lack Mcm9 (and Mcm8), which supports the idea that MCM9 evolved to have a specialized role, rather than an essential one for DNA replication in some species. Second, the Xenopus egg extract system may not be entirely accurate in recapitulating all aspects of DNA replication and its regulation. Third, it is possible that, despite rigorous controls (14), the MCM9 immunodepletion had an unknown, deleterious effect upon pre-RC formation. Furthermore, because CDT1 interacts directly with MCM2-7 in vivo (24), MCM9 would not be required to mediate the essential CDT1:MCM2-7 interaction needed for loading. Whatever the reason for the disparity in results, the present study underscores the importance of in vivo genetic ablation studies to reveal the true physiological role of genes in the context of a whole animal.

Although Mcm9 is ubiquitously expressed, its ablation appears to affect only a subset of tissues in the mouse under normal laboratory conditions. There were two major phenotypes: germ-cell depletion and cancer susceptibility. The observed germ-cell loss appears to occur at two stages. First, the shortfall observed at birth must have its roots in: (i) the specification or proliferation of the primordial germ cell (PGC) lineage; (ii) migration of the PGCs to the genital ridges and primitive gonads; or (iii) proliferation within the primitive gonads. This germ-cell proliferation happens rapidly, expanding the number from ∼100 to >20,000 (25). Alternatively, PGCs may get depleted because of a failure in maintenance after migration to the primitive gonads. The second stage of loss is specific to males, where there appears to be a defect in spermatogonial self-renewal during adulthood. How MCM9 deficiency causes these defects is not clear. Although MCM9 is related to the MCM2-7 replicative helicase proteins, we do not believe that overall DNA replication per se is impaired, because severely hypomorphic MCM2-7 mutations do not have germ-cell defects. Rather, we conjecture that given the observed phenotypes of cancer susceptibility, genomic instability, and replication stress sensitivity, that MCM9 deficiency causes a subtle defect in repair of replication-induced damage to which rapidly proliferating germ-line stem cells are especially sensitive. If true, this defect might trigger cell death as a means to prevent transmission of a compromised genome to offspring. However, like DNA damage- or replication stress-induced apoptosis of ES cells (26), such death would have to occur via a TRP53-independent pathway, as indicated by our genetic studies. Alternatively, the progressive loss of spermatogonial stem cells in adult male Mcm9 mutants points to stem-cell maintenance or proliferation defects. Future experiments will be geared toward identifying the stages at which germ cells are being lost during development, and what cell cycle and damage checkpoint pathways might be activated to cause their elimination or failure to proliferate.

Interestingly, Mcm9 mutant mice were susceptible to distinct sex-specific cancers. Males were highly prone to HCC. In humans, HCC is more common in males than females (27). The majority of HCC mouse models are driven by viral induction, drug treatments, deregulated oncogenes, or a combination of these (28, 29). We are aware of only one germ-line mutation (Abcb4/Mdr2) that causes a high incidence of HCC formation with <1 y latency (30). Therefore, the Mcm9 mutants may provide a useful model for genetic susceptibility to HCC. Mutant females were prone to ovarian tumors, although these may be related to germ-cell depletion and premature ovarian failure (31). In conclusion, MCM9 depletion drives neoplasia by a mechanism that does not seem to involve a major increase in chromosome instability or disrupted loading of MCM2-7 onto pre-RCs. However, the delayed ability of mutant MEFs to reenter the cell cycle following APH treatment suggests that certain cell types may have a sensitivity to replication stress, and this may play a role in the cancer susceptibilities.

Materials and Methods

Mice.

Gene-trap–bearing ES cell lines (from 129 substrains) were obtained from BayGenomics (XG743 and XE518) and The Sanger Institute (AWO655 and ALO673). Chimeras were generated by microinjection of the ES cells into C57BL/6J blastocysts using standard procedures. Following germ-line transmission, alleles were backcrossed into C3HeB/FeJ^a (“C3H”). Exact insertion sites of gene-trap vectors were determined by “primer walking,” as previously described (6, 8). Genotyping was performed either by PCR amplification of the neo gene within the vector, by insertion-specific assay, or using polymorphic flanking microsatellite markers D10Mit20 and D10Mit194 that are polymorphic between 129 and C3H (Table S3). The use of mice in this study was approved by Cornell’s Institutional Animal Care and Use Committee, Protocol #2004-0038 to J.C.S.

Histology and Immunohistochemistry.

For basic histology, tissues were fixed in 4% paraformaldehyde overnight, paraffin-embedded, sectioned, and stained with H&E. For germ-cell counts, 10-μM sections of 1-d-old gonads were immunostained as previously described (32). Antibodies: Rabbit anti-DDX4/MVH (Abcam ab13840; 1:250); goat anti-rabbit Alexa 488 conjugate (Molecular Probes A11008; 1:1,000). Germ cells were counted in three sections from the midportion of each gonad and averaged. The data were analyzed using one-way ANOVA with Bonferroni correction (Prism software package). The resulting P values were used to determine significance (P < 0.05).

RT-PCR and cDNA Analysis.

Semiquantitative PCR analysis of various mouse tissues (Fig. 1C) was performed on the Mouse Multiple Tissue cDNA panel from Clontech (636745). Real-time RT-PCR was performed as previously described (6, 8). Oligonucleotide primers are listed in Table S3.

MEF Growth Studies.

MEF growth analyses and metaphase spreads were performed as previously described (33). For senescence assays, cells were counted every 3 d and replated at 5 × 10⁵. All MEFs were derived from C3HeB/FeJ congenic (N10) embryos and were primary cultures.

Micronucleus Assays.

Micronucleus assays were performed essentially as previously described (34).

Isolation of Protein Fractions.

For Western analyses of chromatin bound vs. nonchromatin-bound proteins, we used a Triton-100 detergent fractionation protocol. Briefly, MEFs were trypsinized, washed twice in cold PBS, resuspended by vortexing in 1 mL TX-NE (320 mM sucrose, 7.5 mM MgCl₂, 10 mM Hepes, 1% Triton X-100, plus protease inhibitor), and incubated on ice for 30 min. Nuclei were pelleted (500 × g, 3 min) and the supernatant, containing proteins of cell membrane, cytosolic, and free forms of MCMs, was designated as the detergent “soluble” fraction (23, 35). The nuclear pellet was resuspended in 0.5 mL RIPA, liberating the chromatin fraction containing nuclear scaffold proteins, DNA, and chromatin binding forms of MCMs. Antibodies used were as follows. MCM2: ab31159 (Abcam); MCM7: ab2360 (Abcam); β-actin: A1978 (Sigma); Fibrillarin: ab5821 (Abcam), CDT1: 06–1295 (Millipore), GAPDH: 6C5 (Advanced Immunochemical).