Report

Involvement of Mammalian Mus81 in Genome Integrity and Tumor Suppression

See allHide authors and affiliations

Science  18 Jun 2004:
Vol. 304, Issue 5678, pp. 1822-1826
DOI: 10.1126/science.1094557

Abstract

Mus81-Eme1 endonuclease has been implicated in the rescue of stalled replication forks and the resolution of meiotic recombination intermediates in yeast. We used gene targeting to study the physiological requirements of Mus81 in mammals. Mus81–/– mice are viable and fertile, which indicates that mammalian Mus81 is not essential for recombination processes associated with meiosis. Mus81-deficient mice and cells were hypersensitive to the DNA cross-linking agent mitomycin C but not to γ-irradiation. Remarkably, both homozygous Mus81–/– and heterozygous Mus81+/– mice exhibited a similar susceptibility to spontaneous chromosomal damage and a profound and equivalent predisposition to lymphomas and other cancers. These studies demonstrate a critical role for the proper biallelic expression of the mammalian Mus81 in the maintenance of genomic integrity and tumor suppression.

Homology-directed DNA repair is a major pathway that facilitates the accurate removal of chromosomal damage resulting from exogenous stimuli, stalled replication forks (RFs), or genetically programmed processes (13). Homologous recombination during mitosis and meiosis is believed to use a four-stranded DNA structural intermediate known as the Holliday junction (HJ) (4, 5). HJs are also thought to be intermediates in the repair of stalled RFs. HJ processing in mammalian cells has recently been linked to RAD51C and XRCC3 (6).

Studies have demonstrated a role for the DNA endonuclease Mus81-Eme1/Mms4 in the processing of branched DNA structures associated with stalled RFs and HJ processing (719). Yeast mus81 mutants are sensitive to agents that collapse RFs but not those that cause double-strand breaks (79). As in yeast, mammalian Eme1/Mms4 and Mus81 constitute a structure-specific endonuclease (16, 18, 19). Schizosaccharomyces pombe and human Mus81 complexes cleave HJs in vitro (10, 12), with yeast Mus81-Eme1 showing a preference for nicked HJs (15, 17). In contrast, other studies demonstrated a preference for 3′-flap and branched fork DNA structures, compared with HJs, suggesting a role for Mus81-Eme1/Mms4 in cleaving stalled RFs to initiate recombination recovery (9, 13, 14, 16, 1820). A role for Mus81/Mms4 has also been reported in the processing of recombination intermediates that form downstream of collapsed RFs (21, 22).

In order to elucidate in vivo requirements for mammalian Mus81, we created mice with a targeted disruption of Mus81 (fig. S1). Western analyses using two different antibodies raised against the N terminus of Mus81 revealed a 61-kD product that was absent in Mus81–/– mice and failed to detect a putative 11-kD N-terminal peptide encoded by the mutant allele (fig. S1). Serial twofold dilution analysis of lysates revealed a reduction of ∼50% in the amount of Mus81 protein in heterozygote, compared with wild-type, testes. Mus81+/– and Mus81–/– mice were born at expected Mendelian frequencies and were morphologically indistinguishable from wild-type littermates, which indicates that Mus81 is dispensable for embryonic development and postnatal viability. Mus81 is also not required for mammalian meiotic recombination, as Mus81–/– males and females were fertile and exhibited no apparent defect in gametogenesis (fig. S2). Gene targeting in embryonic stem (ES) cells, a process mediated by homologous recombination, was unaffected by Mus81 deficiency (fig. S3). Furthermore, all genotypes exhibited normal lymphocyte ontogeny, indicating that Mus81 is dispensable for B and T cell development, as well as activation-driven proliferation (figs. S4 and S5, table S1). Taken together, these findings indicate a nonessential role for mammalian Mus81 in developmental pathways that require the formation or processing of double-strand breaks.

Mus81-deficient ES cells were found to be hypersensitive to mitomycin C (MMC) (Fig. 1A) but not γ-irradiation (Fig. 1B). Mus81+/– ES cells were not hypersensitive to MMC, nor were Eme1+/– ES cells (19). Mus81-deficient mice were more susceptible to MMC-induced lethality at doses of 15 mg per kilogram of body weight (P = 0.0065) and 20 mg/kg (P = 0.0004) by log-rank analysis (Fig. 1, C to E). Furthermore, similar to Eme1 mutant cells, Mus81-deficient ES cells displayed elevated levels of MMC-induced sister chromatid exchange (fig. S6). Wild-type, Mus81+/–, and Mus81–/– mice exhibited an equivalent sensitivity to 7 Gy (23); however, Mus81–/– mice were slightly more sensitive to 9 Gy γ-irradiation (P = 0.0144) (Fig. 1F). The hypersensitivity of Mus81–/– ES cells to MMC is reminiscent of Eme1–/– ES cells (19) and strongly suggests a role for the Mus81-Eme1 endonuclease in the repair of MMC-induced DNA damage. Unlike γ-irradiation–induced DNA damage, the critical MMC-induced cytotoxic lesions are DNA interstrand cross-links; it has been proposed that they are repaired through a combination of excision repair and homologous recombination (24, 25).

Fig. 1.

Mus81-deficient ES cells and mice display differential sensitivity to MMC but not to γ-irradiation. Sensitivity of Mus81-deficient ES cells to MMC (A) and γ-irradiation (B) is shown. (C to E) Survival of mouse cohorts injected intraperitoneally with 10 mg/kg MMC [(C), n = 7, 7, and 10 for wild-type, Mus81+/–, and Mus81–/–, respectively], 15 mg/kg MMC [(D), n = 10 for all three cohorts], and 20 mg/kg MMC [(E), n = 7, 10, and 11 for wild-type, Mus81+/–, and Mus81–/–, respectively]. (F) Survival of mouse cohort irradiated with 9 Gy (n = 7, 5, and 5 for wild-type, Mus81+/–, and Mus81–/–, respectively). Data points from wild-type and Mus81+/– are superimposed in panel (C).

Metaphase spreads from activated T cells derived from wild-type, Mus81+/–, and Mus81–/– siblings were scored for chromosomal aberrations representing all detectable chromosomal abnormalities. Surprisingly, Mus81 was found to be haploinsufficient for genomic instability as both Mus81+/– and Mus81–/– T cells displayed a mild propensity for aneuploidy and chromosomal damage in the absence of exogenous clastogenic stimuli, compared with wild-type T cells (Fig. 2 and table S2). In contrast to wild-type controls, in which we observed no defect in 190 metaphases studied, 6.7 ± 2.2% of Mus81–/– (n = 183) and 7.2 ± 3.1% (n = 181) of Mus81+/– T cells were aneuploid. Unlike wild-type activated T cells, in which no chromosomal defects were observed, 8.4 ± 4.3% of Mus81+/– and 6.6 ± 1.9% of Mus81–/– T cells exhibited chromosomal abnormalities including chromosomal breaks, fragments, fusions, triradials, and quadriradials. As Mus81 deficiency results in hypersensitivity to MMC, we examined metaphases from activated T cells after exposure to 40 ng/ml of MMC. MMC treatment induced aneuploidy in 11.2 ± 1.5% (n = 179) of wild-type, 13.0 ± 3.6% (n = 186) of Mus81+/–, and 16.2 ± 2.6% (n = 185) of Mus81–/– T cells; chromosomal abnormalities after MMC treatment were observed in 10.7 ± 0.9%, 23.2 ± 2.5%, and 24.4 ± 3.6% of wild-type, Mus81+/–, and Mus81–/– T cells, respectively (Fig. 2 and table S3). The equivalent MMC-induced increase of aberrant metaphases between all three genotypes may reflect the increased death of heavily damaged Mus81+/– and Mus81–/– cells. Taken together, these findings suggest a role for mammalian Mus81 as a genomic caretaker, as well as in DNA repair processes that are used after MMC exposure.

Fig. 2.

Chromosomal aberrations in wild-type, Mus81+/–, and Mus81–/– T cells. Representative metaphases from wild-type (A), Mus81+/– (B and C), and Mus81–/– (D to F) are shown, depicting normal metaphase chromosomes (A); labels are chromosome breaks (br), chromosome fusion (f), triradial (tri), double-minute (dm), and dicentric chromosome (dc). (G) Frequency of spontaneous and MMC-induced aneuploidy, chromosomal aberrations, and chromosomal damage (breaks, triradials, and fusions).

We monitored the health and survival of wild-type (n = 19), Mus81+/– (n = 73), and Mus81–/– (n = 15) mice for a period of 1 year. In contrast to 95% of wild-type mice, only 27% of homozygotes and 50% of heterozygotes remained healthy and survived through the first year of life (Fig. 3A). All sick animals examined to date were found to be afflicted with tumors, especially non-Hodgkin's lymphoma in heterozygote (incidence 0.97) and homozygote (incidence 1.00) mice that typically accumulated in the cervical or axial lymph nodes of afflicted mice (Fig. 4 and table S4). To ascertain the cellular origin of the lymphomas, tumor sections were stained with antibody against CD45R/B220 (a B cell–specific antibody), and a CD3 antibody that is T cell–specific. Lymphomas from 5 out of 12 heterozygotes and 3 of 3 homozygotes were classified as B cell lymphomas, whereas 7 of 12 lymphomas from heterozygotes were of T cell origin (table S4 and fig. S7). Seven of 33 lymphomas from Mus81+/– mice and 2 of 12 lymphomas from Mus81–/– mice demonstrated metastatic potential, as detected by infiltration to other nonlymphoid organs, including the lungs, liver, and kidney. Cytogenetic analysis of three of these tumors revealed a high frequency of aneuploid cells without the presence of detectable chromosomal translocations (table S5). The different spectrum of genomic abnormalities in the Mus81 tumors compared with nontransformed activated peripheral lymphocytes may reflect (i) the elimination of cells bearing certain types of genomic aberrations, (ii) selective expansion of aneuploid cells, or (iii) differences in proliferation between normal and malignant Mus81 cells.

Fig. 3.

Mus81 and haploinsufficiency for tumor suppression. (A) Kaplan-Meier analysis representing the percent survival of wild-type (n = 19), Mus81+/– (n = 73), and Mus81–/– (n = 15) cohort mice versus age in days. (B) Representative Western analysis of Mus81 expression in protein extracts from tumors obtained from Mus81+/– and Mus81–/– mice. Lane C is wild-type testes extract loaded as a positive control. (C) Representative Southern analysis of Hind III–digested DNA derived from tumors obtained from Mus81+/– mice. Positions of wild-type (wt) and mutant (mut) alleles are shown.

Fig. 4.

Tumor formation in Mus81+/– and Mus81–/– mice. Non-Hodgkin's lymphoma in cervical lymph nodes (arrows) of Mus81+/– mouse (A) and healthy Mus81+/– control [(B) Harderian gland, arrowhead]. (C) Inguinal lymph nodes from mouse depicted in (A) (top) compared with (B) (bottom). Representative histology (D to F) of Mus81+/– T lymphoma [(D), hematoxylin and eosin (H&E) stain; (E), B220 stain; (F), CD3 stain] and (G to I) Mus81+/– B lymphoma [(G), H&E stain; (H), B220 stain; (I), CD3 stain]. Flow cytometry of (J) wild-type lymph nodes, (K) Mus81+/– T-lineage lymphoma, and (L) Mus81+/– B-lineage lymphoma. (M) Mus81+/– serous carcinoma of the ovary (H&E stain); (N) Mus81+/– breast carcinoma with ductal morphology (H&E stain); (O) Mus81–/– comedocarcinoma of the neck (H&E stain).

Several cohort mice developed multiple cancers, including (i) T lymphoma, carcinoma, and sarcoma; (ii) comedocarcinoma of the neck and lymphoma; (iii) breast carcinoma and lymphoma; and (iv) ovarian carcinoma and T cell lymphoma (table S4). Hence, Mus81 mutation resulted in a dramatic predisposition to lymphomas and other cancers.

Surprisingly, Mus81+/– mice showed a predisposition to cancer susceptibility as profound as that of Mus81–/– animals. We evaluated whether the wild-type Mus81 locus was lost or epigenetically suppressed or whether Mus81 might be haploinsufficient for tumor suppression. Mus81 was found to be expressed in all 11 tumors tested from heterozygotic animals (Fig. 3B). The wild-type Mus81 locus was retained in 16 tumors from Mus81+/– mice analyzed by Southern blotting (Fig. 3C). The lack of rearrangement in the majority of tumors examined, together with the expression of Mus81 protein in tumors from heterozygotes, suggests that loss of heterozygosity is not a requisite for tumor formation in our Mus81+/– mice.

Our findings indicate proper biallelic expression of Mus81 is required for genomic integrity and tumor suppression. Although classical studies have demonstrated that tumor suppressor genes lead to neoplastic transformation via inactivation of both allelic copies (26), subsequent studies have demonstrated that the inactivation of one copy of certain genes may result in cancer predisposition [reviewed in (27, 28)]. Loss of a single Mus81 allele was sufficient to result in a loss of genomic integrity in preneoplastic lymphocytes. Hence, the increased genomic instability that results from Mus81 haploinsufficiency might facilitate tumorigenesis. The amount of Mus81 protein in heterozygotes is reduced compared with that of wild-type mice (fig. S1). Hence, the amount of Mus81 protein in heterozygotes might be below a critical threshold, thereby impairing the cleavage of intermediate DNA structures that arise during DNA damage repair. Although tumor predisposition and the frequency of genomic instability appear similar in heterozygotes compared with homozygotes, Mus81–/– ES cells and mice were more susceptible to the clastogenic effects of MMC than Mus81+/– ES cells and mice, which suggests that long-term survival outcome after MMC-induced DNA damage might not be impaired in heterozygote backgrounds. Our findings implicate human Mus81 on 11q13 as a candidate tumor suppressor gene.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5678/1822/DC1

Materials and Methods

Figs. S1 to S7

Tables S1 to S5

References and Notes

References and Notes

View Abstract

Navigate This Article