Snm1-Deficient Mice Exhibit Accelerated Tumorigenesis and Susceptibility to Infection

Abstract

The eukaryotic SNM1 gene family has been implicated in a number of cellular pathways, including repair of DNA interstrand cross-links, involvement in VDJ recombination, repair of DNA double-strand breaks, and participation in cell cycle checkpoint pathways. In particular, mammalian SNM1 has been shown to be required in a mitotic checkpoint that causes arrest of cells in prophase prior to chromosome condensation in response to spindle poisons. Here, we report on the phenotype of a knockout of Snm1 in the mouse. Snm1^−/− mice are viable and fertile but exhibit a complex phenotype. Both homozygous and heterozygous mice show a decline in survival compared to wild-type littermates. In homozygous mutant males, this reduction in survival is principally due to bacterial infections in the preputial and mandibular glands and to a lesser extent to tumorigenesis, while in homozygous and heterozygous females, it is due almost solely to tumorigenesis. The high incidence of bacterial infections in the homozygous mutant males suggests an immune dysfunction; however, examinations of T- and B-cell development and immunoglobulin class switching did not reveal a defect in these pathways. Crossing of Snm1 mutant mice with a Trp53 null mutant resulted in an increase in mortality and a restriction of the tumor type to lymphomas, particularly those of the thymus. Taken together, these findings demonstrate that Snm1 is a tumor suppressor in mice that in addition has a role in immunity.

The SNM1 gene family is represented by five mammalian members including SNM1, SNM1B, Artemis, ELAC2, and CPSF73 (9, 13, 29). The commonality among these genes is the SNM1 domain, which contains a metallo-β-lactamase fold and an appended β-CASP sequence (4, 19). The β-CASP domain is predicted to function as a nucleic acid binding domain, and the metallo-β-lactamase fold has been shown to possess DNA endonuclease activity upon interaction between Artemis and the DNA-dependent protein kinase catalytic subunit (16). Outside of the SNM1 domain, there is no significant sequence homology between the five mammalian SNM1 family members. At least two of these family members, Artemis and ELAC2, have been implicated as tumor suppressors. Artemis mutations in human patients result in a severe combined immunodeficiency (SCID) syndrome in which both T and B cells fail to mature (19). In addition, Artemis-deficient cell lines are sensitive to ionizing radiation (IR) (6, 20, 21), which has been attributed to both defects in the double-strand break (DSB) repair pathway of nonhomologous end joining (15, 16, 22) and cell cycle checkpoint signaling (33). The SCID and IR-sensitive phenotypes have been reconstructed in a knockout of Artemis in the mouse, and in addition, when combined with Trp53 deficiency, Artemis was revealed to have tumor suppressor activity (24, 25). A second SNM1 family member, ELAC2, has been implicated in prostate cancer susceptibility in humans (29), although this conclusion has proven to be controversial (27).

In budding yeast, the single SNM1 gene is involved in mediating resistance to interstrand cross-linking drugs, such as nitrogen mustard or mitomycin C (MMC), but not to other forms of DNA damage (10, 26). The yeast protein has also been shown to possess exonuclease activity (14) and appears to function in the repair of DNA DSBs that occur during processing of interstrand cross-links (2, 17). However, the function of mammalian Snm1 seems quite distinct from that of the yeast protein. While it has been shown that human Snm1 localizes to sites of DSBs after exposure of cells to either IR or cross-linking agents (23), Snm1-deficient cells exhibited no hypersensitivity to IR and only minor hypersensitivity to MMC (9). Thus, the function of Snm1 at sites of DSBs remains unclear. In addition, our recent findings have demonstrated that Snm1 is involved in a mitotic cell cycle checkpoint that arrests cells in prophase in response to spindle poisons, such as nocodazole and taxol (1). This checkpoint appears to correspond to the pathway that was originally discovered and found defective in Chfr-deficient cells (28). Consistent with a defect in the prophase checkpoint, both Snm1- and Chfr-deficient cells are highly hypersensitive to spindle poisons.

In this report, we describe the phenotype of mice in which Snm1 has been disrupted by gene targeting. Snm1^−/− mice are viable and fertile but exhibit decreased long-term survival compared to wild-type littermates. The two identified causes of this decreased survival are accelerated tumorigenesis and susceptibility to infection.

MATERIALS AND METHODS

Generation of Snm1 mutant ES cells and mice.

A mouse Snm1 cDNA clone (IMAGE clone identification no. 533047) was used to screen a lambda phage mouse (129/SvEv) genomic library to obtain a fragment of the Snm1 locus. The Snm1 targeting construct was designed to replace exons 2 to 7 with a loxP-flanked PGKneobpA cassette in the opposite transcriptional orientation. Gene targeting in AB1 embryonic stem (ES) cells and microinjection of targeted clones into C57BL/6J blastocysts were performed as described previously (11). Targeted recombinants were verified by Southern blot analysis using 5′ and 3′ external probes and by PCR analysis. The sequences of PCR primers were as follows: S1, 5′-CATAGAAAATTCCCCTTGGACTATG; S2, 5′-GCCAATGCATCCGAGGGGCTG; N1, 5′-AGCAAGGGGGAGGATTGGGAAGACA.

Generation of Snm1^−/− Trp53^−/− mice.

Snm1^−/− mice were crossed to Trp53^−/− mice (129/C57BL; kindly provided by Guillermina Lozano) to generate Snm1^+/− Trp53^+/− mice, which were then bred to generate all cohort mice. Genotyping for Trp53 was performed by PCR as described previously (12). The mice were monitored for 1 year after birth without any treatment.

Mouse handling.

All mice were maintained according to NIH guidelines and an approved Animal Care and Use Committee protocol. They were maintained in a conventional specific-pathogen-free facility. The mice were regularly monitored and were sacrificed when they were moribund or when they showed signs of chronic progressive diseases.

Pathology.

A complete necropsy was performed on mice that were found shortly after death, that were sacrificed because they had a tumor burden that was 10% of their body weight, that were moribund, or that displayed poor body condition. Also, randomly selected age-matched wild-type, heterozygous, and homozygous mutant mice were subjected to necropsy for phenotypic comparisons. All organs were collected and fixed with 10% phosphate-buffered formalin and stored until they were processed for histology. A portion of each organ was embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Tissues were analyzed by light microscopy.

Lymphocyte analyses and CSR assays.

Single-cell suspensions were prepared from thymus, spleen, bone marrow, and lymph nodes of 5- to 12-week-old mice according to a standard protocol. The cells were stained with various antibodies conjugated with fluorescein isothiocyanate (FITC) (CD8, CD43, immunoglobulin G1 [IgG1], and IgG2b), phycoerythrin (PE) (CD4 and B220), and allophycocyanin (CD3 and IgM). All antibodies were purchased from PharMingen. Fluorescence-activated cell sorter (FACS) analyses were performed using a FACSCalibur, and the results were analyzed by Cellquest (BD Biosciences). For class switch recombination (CSR), mouse sera were collected from wild-type and mutant mice, and levels of IgL, IgM, and IgG subclasses and IgA were detected by flow cytometry in a FITC and R-PE-based bead assay (SouthernBiotech). In addition, splenic B cells were also collected from wild-type and mutant mice and cultured with appropriate stimulants and cytokines to assay for class switch recombination (18).

Genotoxicity survival assays.

Snm1^+/+ and Snm1^−/− mouse embryonic fibroblasts (MEFs) were derived from 13.5-day poistcoitus embryos and cultured as described previously (1). For colonogenic survival studies, MEFs were exposed to different doses of MMC for 1 h, washed with 1× phosphate-buffered saline three times, and subsequently cultured in regular media. MEFs were also exposed to various doses of IR. For both experiments, colonies were counted 9 days later after being stained with trypan blue.

For whole-animal experiments, 9-week-old Snm1^+/+ and Snm1^−/− mice were exposed to 7.5 Gy IR. For studies with taxol (Mead Jonson), 15-week-old Snm1^+/+ and Snm1^−/− mice were injected with the drug at 15 mg/kg of body weight. For both studies, the mice were monitored every day until they died or exhibited morbidity.

Proliferation assay.

To monitor proliferation, cells were maintained on a defined 3-day passage schedule by plating 3 × 10⁵ cells (3T3 protocol) in 60-mm dishes as described previously (30). The cells were counted at each passage, and the total number was calculated prior to replating. Growth rates at passages 2, 5, 10, and 15 were determined by plating replicate cultures of 2.5 × 10⁴ cells in 35-mm dishes; duplicate cultures were harvested every day thereafter, and the cells were counted.

RESULTS

Snm1 mice exhibit decreased survival due to tumorigenesis and infections.

To investigate the function of Snm1, we generated mice with a homozygous deletion by targeted disruption in mouse ES cells. The targeting strategy was planned to delete exons 2 through 7, which would eliminate over 75% of the Snm1 coding region (Fig. 1A). The human and mouse Snm1 genes possess an unusually long 5′ untranslated region, which contains an internal ribosome entry site that regulates Snm1 expression as a function of the cell cycle by up regulation during mitosis (32). The start of translation occurs approximately one-third of the distance from the 3′ end of exon 1. Three targeted ES cell lines were identified, and two of these were used for blastocyst injection and generation of gene-targeted mice. The genotype analyses of one of these strains are shown (Fig. 1B and C). The Snm1^−/− mice were born at Mendelian ratios, were viable and fertile, and did not immediately exhibit an evident phenotype. The mice were observed for a period of 24 months, and both the Snm1^−/− and the Snm1^+/− mice showed decreased survival compared to wild-type littermates, as indicated by a Kaplan-Meier analysis (Fig. 2A). The differences were statistically significant for both the heterozygous and homozygous mutant mice (P < 0.001). The decreased survival was due principally to two identified factors: increased susceptibility to bacterial (Staphylococcus aureus) infections and accelerated tumorigenesis. When the animals were analyzed by gender, males showed a reduction in survival compared to females (Fig. 2B), and this difference was due to a greater incidence of infection in the males. Only two sites of infection were noted, namely, the preputial gland and the mandibular gland, with the majority of infections in the male occurring at the former site (Fig. 3). A much lower level of infections was observed in females, and only in the mandibular gland. To determine the relative contributions of infections and tumorigenesis to the mortality of Snm1 mice, we examined these factors independently. This analysis (Fig. 4A) showed that tumorigenesis was significantly different in both homozygous (P < 0.001) and heterogenous (P < 0.001) females and in homozygous males (P < 0.001) compared to wild-type animals of the same gender (Table 1). Homozygous females were also more prone to tumorigenesis than homozygous males (P = 0.008). Infections (Fig. 4B) were significantly more frequent only in the homozygous males compared to the wild-type males (P < 0.001) (Table 2).

FIG. 1.

Disruption of Snm1 in the mouse by gene targeting and genotype of knockout mice. (A) Schematic representations of mouse Snm1 locus (top) and the Snm1 targeting vector (middle). Solid boxes denote exons, with 5′ ends to the left. D (DraI), B (BstZ171), E (EcoRI), and S (SstI), restriction enzyme sites used for analyses of the targeted deletion. Prob, probe. (B) Southern blot analysis of tail DNA from F₂ animals. The sizes of SstI fragments are indicated. Genomic DNA digested with SstI and hybridized with the 5′ probe yielded fragments of 4.3 and 3 kb in the wild-type and targeted locus, respectively (left). Hybridization with the 3′ probe yielded fragments of 12 and 6.4 kb in the wild-type and targeted locus, respectively (right). (C) Multiplex PCR genotype analysis with primer pair S1/S2 (lanes 1 and 2), which results in a 602-bp product in the wild-type allele, and with primer pair S1/N1 (lanes 2 and 3), which resulted in a 240-bp product in the targeted allele.

FIG. 2.

Survival of Snm1^−/− mice. (A) Increased mortality in Snm1 mice shown by Kaplan-Meier analysis. Snm1^−/−, Snm1^+/−, and Snm1^+/+ animals were observed for 24 months. (B) Kaplan-Meier analysis showing survival by gender and genotype of Snm1^−/− male, Snm1^+/− male, Snm1^+/+ male, Snm1^−/− female, Snm1^+/− female, and Snm1^+/+ female cohort mice versus age in months.

FIG. 3.

Incidence of infections in Snm1 mice. (A) Frequencies of bacterial infections in preputial and mandibular glands of Snm1 mice up to 24 months of age. (B) Examples (arrows) of infected organs, mandibular gland (upper) and preputial gland (lower), obtained from Snm1^−/− mice.

FIG. 4.

Effects of infections and tumorigenesis on the mortality of Snm1 mice. The number of mice within each genotype that were followed for tumor development and infections and the mean risk values are given in Tables 1 and 2, respectively. (A) Kaplan-Meier tumor-free-survival curve by gender and genotype of Snm1 cohort mice versus age in months. (B) Kaplan-Meier infection-free-survival curve by gender and genotype of Snm1 cohort mice versus age in months.

TABLE 1.

Tumor susceptibility of mice at 24 months

Genotype, sex	n	Riska (%)	P valueb
SNM1+/+, male	64	22.2 ± 6.2
SNM1+/−, male	39	27.9 ± 9.0	0.52
SNM1−/−, male	85	100	<0.001
SNM1+/+, female	60	20.1 ± 5.7
SNM1+/−, female	33	61.2 ± 9.9	<0.001
SNM1−/−, female	78	100	<0.001

^aRisk ± standard error.

^bLog rank test for equality of tumor development compared with wild type.

TABLE 2.

Risk of infection for mice at 24 months

Genotype, sex	n	Riska (%)	P valueb
SNM1+/+, male	64	12.4 ± 4.8
SNM1+/−, male	39	21.3 ± 7.9	0.12
SNM1−/−, male	85	80.0 ± 5.6	<0.001
SNM1+/+, female	60	0
SNM1+/−, female	33	23.6 ± 9.6	0.32
SNM1−/−, female	78	20.0 ± 7.0	0.12

^aRisk ± standard error.

^bLog rank test for equality of infection compared with wild-type male.

The types of cancers found in Snm1 mice were primarily hematologic tumors, with fewer occurrences of adenomas and sarcomas (Table 3). Without any treatment, we monitored heterozygous and homozygous Snm1-deficient mice and normal littermates for spontaneous tumor development until 20 months of age. The most common tumors arising spontaneously in these animals were lymphomas. No tumors were observed in wild-type mice (28 animals) until the age of 17 months. From 17 to 20 months of age, four wild-type mice (14.3%) developed lymphoma (systemic or mesenteric lymph node) and one mouse (3.6%) developed a histiocytic sarcoma. Heterozygous Snm1 mice started to develop tumors from the age of 16 months. Out of 16 mice, 4 heterozygous mice (25%) developed lymphoma (systemic or lung, mandibular, or adrenal lymph node) and 1 mouse (6.3%) developed histiocytic sarcoma. In our study population of homozygous animals, both females (66.66%) and males (28.57%) developed a similar spectrum of tumor types. The first noted occurrence of tumors in the males was at 15 months of age, whereas tumors first appeared in the females at around 11 months of age. Tumor occurrence in female homozygous mice increased significantly at 15 months of age (Fig. 4A). Out of 61 homozygous mice, 24 (39.3%) had lymphoma, with the highest occurrences in the mandibular lymph node, mesenteric lymph node, spleen, liver, or lung or systemically and lesser occurrences in adrenal gland, thymus, kidney, ovaries, mesoderm, bone marrow, or pancreas. Two of 24 mice with lymphoma also developed lung adenoma, and 1 of these 24 animals also developed ovarian cyst adenoma. Additionally, two mice (3.3%) had histiocytic sarcoma, three mice (5.0%) had sarcoma (liver), six mice (9.8%) had adenoma (lung and hardarian gland), six mice (9.8%) had cyst adenoma (ovary, uterus, and papillary hardarian gland), and one mouse had malignant teratoma (ovary). All tumors were analyzed in detail by histopathology (Fig. 5). We conclude that Snm1 mice are susceptible to accelerated tumorigenesis and that the incidence is greater in females than in males.

TABLE 3.

Tumor spectrum in Snm1 mice^a

Genotype	Tumor type	Incidence (%)
Snm1−/− (n = 61)	Lymphoma	39.3
	Adenoma	9.8
	Cyst adenoma	9.8
	Sarcoma	5.0
	Histiocytic sarcoma	3.3
	Teratoma	1.6
Snm1+/− (n = 16)	Lymphoma	25
	Histiocytic sarcoma	6.3
Snm1+/+ (n = 28)	Lymphoma	14.3
	Histiocytic sarcoma	3.6

^aMonitored until 20 months of age.

FIG. 5.

Histologic analysis of tumors found in Snm1^−/− mice. (a to c) Examples of spontaneous tumors from different organs of Snm1^−/− mice: (a) liver, (b) adrenal gland, and (c) uterus and ovary. (d to f) Representative histology of lymphomas of Snm1^−/− mice: (d) mandibular lymph node lymphoma, ×100 magnification; (e) spleen lymphoma, ×50 magnification; and (f) adrenal gland lymphoma, ×200 magnification. (g to i) Representative histology of adenomas of Snm1^−/− mice: (g) lung adenoma, ×20 magnification; (h) papillary hardarian gland adenoma, ×50 magnification; and (i) ovary cyst adenoma, ×200 magnification. All tumors were obtained from 15-month- to 20-month-old Snm1^−/− mice. The tissues were paraffin embedded and stained with hematoxylin and eosin stain.

Snm1 mice have normal T- and B-cell development and immunoglobulin class switch recombination.

As noted above (Fig. 3), the frequency of infections, particularly in the Snm1^−/− male mice, was extremely high, with almost 80% of these animals exhibiting the condition between 10 and 24 months of age. The vast majority of these infections occurred in the preputial gland, indicating that this organ in the male was particularly susceptible. To assess for a possible immunological dysfunction in Snm1^−/− mice, we analyzed the levels of T and B lymphocytes. As shown (see Fig. S1 in the supplemental material), T and B lymphocytes were grossly normal in Snm1^−/− mice compared to wild-type littermates. We found that total thymocyte numbers in Snm1^−/− mice were very similar to those in wild-type mice and that CD4/CD8 double- or single-positive populations were comparable to those found in wild-type animals. Both CD4⁺ and CD8⁺ T cells are normal in peripheral blood (see Fig. S1A in the supplemental material). Similar results were obtained in the B-lymphocyte compartment. We checked bone marrow, spleen, and lymph nodes and found no differences in FACS profiles between Snm1^−/− and wild-type mice (see Fig. S1B in the supplemental material). Based on these observations, we conclude that Snm1, unlike its family member Artemis, is not required for V(D)J recombination.

To determine if Snm1^−/− B cells are able to undergo CSR, we first examined the level of serum immunoglobulin isotypes of Snm1^−/− and wild-type mice by a flow cytometry assay using FITC- and PE-based beads. Our results showed that all isotypes examined were present in both male and female Snm1^−/− mice (data not shown), indicating that Snm1 is not required for CSR. To confirm that Snm1^−/− B cells are able to undergo class switch recombination, we stimulated in vitro cultures of splenocytes from mutant and wild-type mice (18). The results demonstrated that splenocytes from Snm1^−/− mice underwent CSR normally after lipopolysaccharide or interleukin-4/CD40 stimulation (data not shown). Therefore, both in vivo and in vitro analyses indicated that Snm1 mutant mice are able to undergo class switching, and we conclude that Snm1 is not essential for CSR.

Snm1 deficiency accelerates tumorigenesis in Trp53 null mice.

A number of studies have shown cooperation between a potential tumor suppressor gene and Trp53 (3, 7, 24, 34). Thus, to further assess the tumor suppression activity of Snm1, we derived Snm1^+/− Trp53^+/− mice and then bred progeny. All genotypes were born at Mendelian frequency and appeared normal at birth. However, we observed a statistically significant decrease in the survival of Snm1^−/− Trp53^−/− mice compared to Snm1^+/+ Trp53^−/− littermates (P = 0.05) (Fig. 6). The time to 50% survival was approximately 4 months for the former mice compared to 5.5 months for the latter mice. Histological analysis indicated that 100% of the Snm1^−/− Trp^−/− mice had developed thymic lymphomas, and a majority of these animals had also developed nonthymic lymphomas (Table 4). Sarcomas were not observed in any of the 14 Snm1^−/− Trp^−/− mice, whereas sarcomas were observed in Snm1^+/− Trp^−/− mice.

FIG. 6.

Homozygosity for Snm1 decreases the survival of Trp53^−/− mice. Kaplan-Meier analysis showing the survival of Snm1^−/− Trp53^−/−, Snm1^+/+ Trp53^−/−, and Snm1^−/− Trp53^+/− cohort mice versus age in months.

TABLE 4.

Tumor types in Trp531Snm1 mice

Tumor type	No. (%) in Trp53−/− mice
	Snm1−/− (n = 14)	Snm1+/− (n = 15)	Snm1+/+ (n = 8)
Lymphoma
Thymic	14 (100)	8 (53.3)	3 (37.5)
Nonthymic	9 (64.3)	7 (46.7)	4 (50)
Sarcoma
Soft tissue	0 (0)	3 (18.8)	3 (37.5)
Osteosarcoma	0 (0)	2 (12.5)	1 (12.5)

Snm1 mice are hypersensitive to taxol.

As indicated in the introduction, Snm1^−/− MEFs are highly sensitive to spindle poisons, such as nocodazole and taxol (1). To determine the effects of a spindle poison on whole animals, we injected wild-type and homozygous mice with a single dose of taxol (15 mg/kg of body weight) and subsequently monitored the animals for morbidity. By 64 days, all of the Snm1^−/− mice had died compared to approximately 50% of the wild-type mice (P = 0.0282) (Fig. 7). These findings are consistent with our previous observations of hypersensitivity to spindle poisons in Snm1^−/− MEFs. We have also found that human Snm1 localizes to sites of DSBs after exposure of cells to IR (23); however, Dronkert et al. (9) have reported that Snm1^−/− ES cells do not exhibit hypersensitivity to IR. To determine the IR sensitivity of whole animals, we irradiated wild-type and homozygous mice with 7.5 Gy; however, no statistically significant difference in survival between the two genotypes was observed (results not shown).

FIG. 7.

Snm1 homozygous mice are hypersensitive to the spindle poison taxol. Kaplan-Meier analysis showing the survival of Snm1^−/− and Snm1^+/+ mice versus age after injection of taxol at 15 mg/kg of body weight.

Snm1 MEFs exhibit a proliferation defect in culture.

To examine the effects of Snm1 deficiency at the cellular level, embryonic fibroblasts (MEFs) were prepared from 13.5-day embryos and examined by the 3T3 proliferation protocol (30). As shown in Fig. S2 in the supplemental material, MEFs derived from homozygous mice exhibited a profound proliferation defect at all passages examined compared to wild-type MEFs. These results indicate that under tissue culture conditions Snm1 has a role in normal cell growth. Upon spontaneous immortalization, both mutant and wild-type MEFs grew at approximately equal rates. Finally, as indicated above, a previous report showed that Snm1^−/− ES cells exhibit an approximately twofold hypersensitivity to MMC and no demonstrable sensitivity to IR compared to wild-type cells (9). We performed similar experiments with the Snm1 MEFs and essentially confirmed these findings, i.e., no hypersensitivity to IR and a small hypersensitivity to MMC (see Fig. S3 in the supplemental material).

DISCUSSION

Our studies of Snm1 knockout mice have revealed an involvement of this gene in the suppression of spontaneous tumorigenesis and in mediating resistance to infection. The latter function was found to be statistically significant only in male animals, principally due to the location of the infections in the preputial gland. This gland occurs in both males and females but is anatomically quite distinct between the two genders. The basis for the presumed immunodeficiency remains unresolved, as analyses of T- and B-cell development did not indicate any defects in these processes. Since Snm1 is a member of the gene family that includes Artemis, an involvement in immune function seemed a possibility. Artemis deficiency results in a SCID syndrome in which mature T and B cells fail to develop due to impaired V(D)J rearrangements at T-cell receptor and immunoglobulin genes (19, 25). Our analyses of Snm1^−/− mice clearly indicated that Snm1 is not involved in V(D)J recombination. Additionally, we showed previously that human Snm1 interacts with 53BP1 (23), and Trp53bp1 mutant mice have been shown to be critical for immunoglobulin class switch recombination (18). Our results demonstrated that Snm1 is not essential for CSR, since both male and female Snm1-deficient mice are able to undergo CSR. It is also possible that the Snm1 deficiency may lead to a defect in innate immunity against certain bacterial infections. While the nature of the immune dysfunction remains unsettled, it is clear that it contributes significantly to the mortality of the male homozygous mutant mice.

Snm1 is the third member of the mammalian SNM1 gene family to be implicated as a tumor suppressor. In a genomewide screen of high-risk pedigrees, mutations in ELAC2/HPC2 were found to segregate with prostate cancer (29). ELAC2 is a gene of unknown function whose risk genotypes have been estimated to cause approximately 2% of prostate cancer in the general population (5); however, other studies have not found such a definitive role for ELAC2 in this disease (27). Disruption of Artemis in the mouse confirmed the SCID phenotype but did not indicate any accelerated tumorigenesis in the knockout animals (25). However, when these mice were crossed with Trp53-deficient animals, a marked and early-onset predisposition to progenitor B-cell lymphomas was observed, indicating that Artemis is a tumor suppressor (24). Artemis-deficient cells exhibit genomic instability, which may be due to a defect in the NHEJ pathway of DSB repair or in cell cycle regulation (15, 22, 33). A previous study of a disruption of Snm1 in the mouse did not report any evidence of accelerated tumorigenesis in these animals (9). While other explanations are possible, the most likely reason for the discrepancy is that the targeted gene strategy used in that study resulted in disruption of only a portion of intron 3 and 25 amino acids of exon 4. In addition, PCR analysis indicated that transcripts produced by the use of an alternative splice site in exon 5 restored the reading frame. Thus, it is possible that the previously reported Snm1 knockout represents a hypomorphic rather than a null allele.

Our previous studies have defined a role for Snm1 in an early mitotic checkpoint that arrests cells in response to spindle poisons, such as nocodazole and taxol (1). This checkpoint occurs in prophase before the onset of chromosome condensation. The phenotype of Snm1-deficient cells is highly comparable to the phenotype of Chfr-deficient cells in that both appear defective in a prophase checkpoint in response to spindle poisons (28). Chfr was found to be mutated in about half of human cancer cell lines examined (28), and furthermore, in a recent report a knockout of Chfr in the mouse indicated that it had a strong tumor suppression function (31). Both Snm1- and Chfr-deficient cells have a marked hypersensitivity to spindle poisons, consistent with a defect in the prophase checkpoint. Our findings reported here confirm these results by demonstrating that Snm1 homozygous animals exhibit increased mortality compared to wild-type mice upon injection of taxol. Taken together, these findings indicate that the prophase checkpoint pathway plays an important role in tumor suppression.

Yeast Snm1 mutants are highly and singularly sensitive to interstrand cross-linking agents (10, 26), however, a specific role for mammalian Snm1 in DNA cross-link repair seems unlikely. Mouse ES cells disrupted in Snm1 showed slight hypersensitivity to MMC, but not to the cross-linking agent melphalan or psoralen plus UVA (9), and our findings indicated only a slight hypersensitivity to MMC in MEFs. One suggested possibility to account for this result is that there is redundancy among the SNM1 family members in DNA repair pathways. However, outside of the SNM1 domain, the mammalian family members show no sequence conservation. Additionally, Artemis-deficient cells exhibit no sensitivity to cross-linking agents (19, 25), and small interfering RNA depletion of Snm1B revealed only minor sensitivity to cisplatin or MMC (8). These results suggest that while SNM1 genes have a role in the stress response to genotoxic agents, they are unlikely to have a direct role in the repair of DNA interstrand cross-links.

In summary, our findings provide the first genetic model demonstrating the role of Snm1 as a tumor suppressor gene in mice. We have also demonstrated that Snm1 has a role in immunity, although the nature of this function and why it is particularly important in the male preputial gland remain to be resolved.