Impaired meiotic DNA-damage repair and lack of crossing-over during spermatogenesis in BRCA1 full-length isoform deficient mice

Mammalian spermatogenesis proceeds through well-defined phases of mitotic and meiotic divisions as spermatogonia develop into spermatocytes, round spermatids and, finally, spermatozoa. In mice, all cell division in the testis is mitotic division before postnatal day 7 (P7). Starting from P8,spermatogonia enter the meiotic prophase of the meiotic pathway and homologous chromosomes undergo a programmed sequence of compaction, synapsis,recombination and segregation that leads to a reduction of genetic material from diploid to haploid (reviewed by Cohen and Pollard, 2001; Tarsounas and Moens, 2001; Zickler and Kleckner, 1999). A unique feature of meiosis I prophase chromatin is a high frequency of DNA double-strand breaks (DSBs), which are required for recombination between homologous chromosomes (reviewed byCohen and Pollard, 2001;Roeder, 1997). Repair of DSBs is necessary to maintain genome integrity and limit the mutational potential of these breaks. Loss-of-function mutations of many factors involved in DNA damage repair result in defective spermatogenesis (reviewed byCohen and Pollard, 2001;Dasika et al., 1999;Tarsounas and Moens,2001).

Germline mutations of BRCA1 are found in approximately half of familial breast cancer cases and the majority of kindreds with combined familial breast and ovary cancer (Alberg and Helzlsouer, 1997; Brody and Biesecker, 1998; Paterson,1998). BRCA1 contains 24 exons that encode proteins of 1863 and 1812 amino acids in human and mouse, respectively(Lane et al., 1995;Miki et al., 1994). It has been shown that BRCA1 contains multiple functional domains that interact with numerous molecules including products of tumor suppressor genes, oncogenes,DNA damage-repair proteins, cell-cycle regulators, ubiquitin hydrolases, and transcriptional activators and repressors (reviewed byDeng and Brodie, 2000). These observations indicate that BRCA1 plays important roles in multiple biological processes and pathways. Mutations of mouse Brca1 using various gene-targeting constructs that introduce null, hypomorphic and tissue-specific mutations result in embryonic lethality, cellular growth defects, increased apoptosis, premature aging and/or tumorigenesis(Bachelier et al., 2003;Cao et al., 2003;Gowen et al., 1996;Hakem et al., 1996;Hohenstein et al., 2001;Liu et al., 1996;Ludwig et al., 1997;Ludwig et al., 2001;Shen et al., 1998;Xu et al., 1999a;Xu et al., 1999b;Xu et al., 2001).

Although pleiotropic effects are associated with Brca1 mutations,it is generally believed that a major function of BRCA1 is to maintain genome integrity. For example, tumorigenesis following loss of BRCA1 function is a consequence of genome instability (Deng,2001). Brca1-mutant embryos and embryonic fibroblast cells display hypersensitivity to γ-irradiation, demonstrating chromosome structural and numeric alterations, loss of G2/M cell-cycle checkpoint and centrosome amplification(Aprelikova et al., 2001;Shen et al., 1998;Xu et al., 1999a). Studies of somatic cell lines carrying mutations in either human BRCA1 or mouseBrca1 reveal essential functions of this gene in multiple DNA damage-repair pathways. This includes nuclear excision repair(Hartman and Ford, 2002),transcription-coupled repair of oxidative DNA damage(Abbott et al., 1999;Gowen et al., 1998),homologous recombinational repair(Moynahan et al., 1999;Moynahan et al., 2001;Snouwaert et al., 1999) and nonhomologous end-joining (Baldeyron et al., 2002; Zhong et al.,2002a; Zhong et al.,2002b). In conclusion, genetic instability associated with BRCA1 deficiency is caused, at least in part, by impaired DNA-damage repair.

Significantly, mice homozygous for a hypomorphic mutation of Brca1survive to adulthood in a p53 (Trp53 — Mouse Genome Informatics) null background at low frequency(Cressman et al., 1999). However, the two Brca1 and p53 double homozygous mutant(Brca1^-/-p53^-/-) male mice examined for this report were infertile because of azoospermia. Moreover, the number of apoptotic cells in the testes was increased, indicating that the germ cell death observed in Brca1^-/-p53^-/- mice was p53-independent. However, the reason why Brca1 deficiency causes failure of spermatogenesis was not examined because of the difficulty in producingBrca1^-/-p53^-/- mice.

The Brca1 gene is expressed in meiotic germ cells. Expression levels increase in the late pachytene and diplotene stages of meiosis I(Zabludoff et al., 1996). We hypothesized that the defect in spermatogenesis observed in Brca1mutant mice was secondary to inadequate DNA repair during meiosis. A mouse model that carried an in-frame deletion of Brca1 exon 11(Brca1^Δ11/Δ11) plus functional deletions of either one or both p53 alleles was utilized to test this hypothesis. Although Brca1^Δ11/Δ11 mice carrying two wild type p53 alleles die in late gestation, they can survive to adulthood if one or both p53 alleles(Brca1^Δ11/Δ11p53^+/- orBrca1^Δ11/Δ11p53^-/-) are mutated(Xu et al., 2001). In this study, these mice were used to demonstrate that deletion of Brca1exon 11 disrupts spermatogenesis but not oogenesis. Disruption of spermatogenesis was accompanied by both p53-independent and p53-dependent apoptosis, associated with altered RNA expression levels of a number of genes that are involved in DNA-damage repair and mislocalization of Rad51 and Mlh1. These defects are accompanied by prolonged localization of γH2AX, a DNA damage sensor, in nuclear foci, suggesting that the meiotic chromosomes fromBrca1 mutant mice accumulate unrepaired DNA damage.

Brca1^Δ11/Δ11p53^+/+,Brca1^Δ11/Δ11p53^+/-,Brca1^Δ11/Δ11p53^-/- mice and control(p53^+/+, p53^+/- andp53^-/-) mice were maintained in a mixed genetic background of 129/FVB/Black Swiss (Xu et al.,2001). BrdU was injected 2 hours prior to euthanasia. Testes were fixed in either Bouin's (for immunofluorescence studies) or 10% formalin (for BrdU labeling) for either 4 hours or overnight. Tissues were dehydrated in increasing ethanol concentrations (70%, 95%, 100%) for 30 minutes each. They then were treated 2 times with xylene (10 minutes each time), 1 time with a xylene/wax mixture at 50°C (1 hour), 3 times with wax alone (1 hour each time), followed by embedding in paraffin and sectioning at a thickness of 4-6μm. Sections either were stained with Hematoxylin and Eosin (H&E) or processed for immunohistochemical analysis using antibodies to germ cell nuclear antigen (GCNA) (1:50, a kind gift from Dr George Enders, University of Kansas Medical Center) or to antiphospho-histone H3 (Upstate, Biotechnology,Lake Placid, NY). Slides were examined and photographed under a light microscope.

Chromosome spreads were prepared from testes of P10-28 males and ovaries of embryonic day (E) 16.5-17.5 female embryos as described(Romanienko and Camerini-Otero,2000). The slides bearing chromosome spreads were either used fresh or within two months of freezing at -20°C (slides thawed at room temperature for ∼5 minutes). After washing with phosphate-buffered saline(PBS) three times for 5 minutes each, slides were processed for immunofluorescent staining using standard procedures. The first antibodies used were: RAD51 (SC-8349, Santa Cruz Biotechnology, Inc. Santa Cruz, CA at 1:100 dilution), DMC1 (SC-8973, Santa Cruz Biotechnology at 1:100 dilution),MLH1 (a mouse monoclonal antibody from Pharmingen, San Diego, CA at 1:100 dilution), rabbit anti-phospho-histone H2AX (a mouse monoclonal antibody from Upstate Biotechnology, Lake Placid, NY at 1:500 dilution) and rabbit anti-SCP3 antibody (Dobson et al.,1994). Alex-red 560 and green 488-conjugated secondary antibodies(Molecular Probes, Eugene, OR) were used at 1:500 dilution.

Western blots were carried out using standard procedures using the following antibodies: goat anti-DMC1 at 1:1000 dilution (Santa Cruz Biotechnology); rabbit anti-RAD51 at 1:1000 dilution (Santa Cruz Biotechnology); mouse anti-MLH1 at 1:1000 dilution; and mouse anti-Pms2 at 1:1000 dilution (Pharmingen, San Diego, CA).

Total RNA was extracted from testes at different developmental stages,including P14, P16, P21 and P22. Reverse-transcription reactions were carried out using a 1st strand cDNA-synthesis kit (Roche, Indianapolis, IN). The cDNA samples were stored at -20°C. One μg of RNA from each sample was used as template for each reaction and 1 μl of cDNA from each sample was used for PCR reaction. The optimal number of cycles for amplification was determined according to the cycle number that yielded the strongest band in the linear range. The ranges of cycles varied from 25-28, depending on the specific RNA target and primer set. The samples were heated to 94°C for 2 minutes and then run through 25-28 cycles of 94°C for 30 s, 60°C for 30 s and 72°C for 1 minute, followed by 72°C for 10 minutes and then 4°C. Primers used in this study and the DNA lengths amplified are shown below.

• TFIIH amplification (680 bp):

  • 5′-cagaaaatcagtccagaggg-3′,5′-ccctcgtctaatggcttatc-3′

• TFIID amplification (360bp):

  • 5′-gttgaaacagacatggtttgc-3′,5′-cagccatgctaagaaagctac-3′

• RuvB amplification (540 bp):

  • 5′-gacacaccattcacagccatc-3′,5′-gatctgttctcggacttctg-3′

• GAPDH amplification (300 bp):

  • 5′-acagccgcatcttcttgtgc-3′,5′-tttgatgttagaggggtctgc-3′

Testis sections were analyzed for apoptosis using the ApoTag kit as recommended by the manufacturer (Intergen Company, Purchase, NY). To evaluate cellular proliferation rates, BrdU incorporation was measured using a cellular proliferation kit following the manufacturer's directions (Amersham Bioscience, England).

Mice at 1 month of age were irradiated at 80 Gy (1Gy minutes^-1,Gammacell 40) and were euthanized 10 minutes, 30 minutes, 2 hours and 4 hours after irradiation. Testes were removed and processed for chromosome spreads and immunofluorescence staining as described above.

Student's T test was used to compare differences of pachytene and diplotene spermatocytes, rates of apoptosis, numbers of foci for Mlh1 and Rad51, and numbers of sex bodies for γH2AX between Brca1 mutant and control mice at ages specified in the text. P≤0.05 was considered statistically significant.

Mature (≥6 weeks of age)Brca1^Δ11/Δ11p53^+/- male mice(n=10) contained no spermatozoa in their testes(Fig. 1B, and data not shown). To identify the earliest stages at which defective spermatogenesis occurred,testes from P10-28Brca1^Δ11/Δ11p53^+/- mutant andp53^+/- control mice were isolated and compared. At P10,testes of both Brca1 mutant and control mice were similar in size(Fig. 1C). Histological analyses demonstrated no obvious differences in seminiferous tubule size and the numbers of spermatogonial cells (Fig. 1D,E). However, starting from P16, testes fromBrca1^Δ11/Δ11p53^+/- mutant mice were significantly smaller than controls (Fig. 1F,I,L). In control mice, spermatocytes began to enter the diplotene stage at P16 and spermatocyte size increased (arrow,Fig. 1G), but no diplotene spermatocytes were detected in the testes from Brca1 mutant mice(Fig. 1H). At P21,spermatocytes from control mice had finished meiosis and formed spermatids(arrow, Fig. 1J). No spermatids were found in the testes from Brca1 mutant mice(Fig. 1K), instead, cells with dark Hematoxylin-stained nuclei were found (arrow,Fig. 1K). This phenotype was apparent in the testes of Brca1 mutant males from P28 to P90(Fig. 1B,N and data not shown). These results indicate that spermatogenesis, including the first wave of spermatogenesis, did not progress beyond meiosis I in the Brca1mutant mice.

Molecular markers were used to define the stages of meiosis I that were defective in the Brca1 mutant male mice. GCNA is expressed specifically in premeiotic cells up to the pachytene stage of meiosis I(Enders and May, 1994). The protein is degraded and no longer detected after the pachytene stage. At P10,the pattern of GCNA staining was similar in Brca1 mutants and control mice (Fig. 2A,B). At P16, some GCNA-negative spermatocytes were found in the testes from control mice,whereas others still stained for GCNA (Fig. 2C). The change in the staining pattern indicated that some, but not all, spermatocytes had passed through the pachytene stage. By contrast,all spermatocytes from Brca1 mutant males demonstrated GCNA staining at day P16 (Fig. 2D). At P21,only one layer of GCNA-positive spermatogonia was observed on the tubular basement membrane in control mice, the majority of cells were GCNA negative(Fig. 2E). By contrast, all cells remained GCNA positive in Brca1 mutant mice at P21(Fig. 1F). The differences in staining patterns between the control and Brca1 mutant mice indicated that spermatogenesis in Brca1 mutant mice did not pass through the pachytene stage of meiosis I.

To determine if mitotic division in the testes was affected by mutation ofBrca1, BrdU-labeling experiments were performed to identify rapidly dividing spermatogonial cells. No significant differences in rates of DNA synthesis between Brca1 mutant and control mice were found(Fig. 2G,H). Similarly, there were no apparent differences in the staining patterns of an antibody to phosphorylated histone H3, which binds specifically to the chromatin of cells in the M phase of mitotic division (Chadee et al., 1995)(Fig. 2I,J). These experiments demonstrated that mitotic division was not impaired in the testes ofBrca1 mutant mice.

The presence of prominent dark Hematoxylin-stained spermatocyte nuclei in the testes of Brca1 mutant mice(Fig. 1K) suggested that theBrca1 mutation caused increased rates of spermatocyte apoptosis. Tissue sections from testes ofBrca1^Δ11/Δ11p53^+/- andp53^+/- control mice at P10, P16 and P21 were analyzed by TUNEL assay to quantify these differences. At P10, <2% of seminiferous tubules in control and mutant testes contained apoptotic cells. This did not change in testes from control mice at P16 and P21. However, in Brca1mutant mice the number of seminiferous tubules that contained apoptotic spermatocytes increased significantly at P16 and P21 (22% and 30%,respectively) (Fig. 3A). To directly compare rates of cellular apoptosis, the percentage of apoptotic cells were determined in TUNEL-positive seminiferous tubules that were selected randomly from testes of ten control and ten Brca1 mutant mice. At P10, in testes from both control and Brca1 mutant mice fewer than 4% of cells were apoptotic (Fig. 3B). At P16 and P21 the percentage of apoptotic cells did not change in testes from control mice, but increased significantly to 40% at P16 and 50% at P21 (both P≤0.0001 compared to control mice) in testes from Brca1 mutant mice (Fig. 3B). Thus, the premature termination of spermatogenesis at the pachytene stage of meiosis I in Brca1 mutant mice was accompanied by significantly increased rates of apoptosis.

Because activation of p53 induces apoptosis in a number of tissue/organ systems in Brca1^Δ11/Δ11 mutant mice(Xu et al., 1999b;Xu et al., 2001), we next determined whether the increased rate of apoptosis in the Brca1mutant testes was p53 dependent. Because only a few (1-2%) ofBrca1^Δ11/Δ11p53^+/+ male mice survive to adulthood (Xu et al.,2001), rates of apoptosis were compared in testes isolated fromBrca1^Δ11/Δ11p53^+/+,Brca1^Δ11/Δ11p53^+/- andBrca1^Δ11/Δ11p53^-/- mice. TUNEL-positive cells were detected in testes of all three genotypes,indicating that the loss of p53 did not prevent apoptosis. In testes isolated from Brca1^Δ11/Δ11p53^+/+ mice at P19(Fig. 3C,D), the absolute numbers of apoptotic cells in each tubule was less than those found in testes from Brca1^Δ11/Δ11p53^+/- andBrca1^Δ11/Δ11p53^-/- mice. However,all the tubules were nearly empty, suggesting either that more extensive cell death happened earlier in spermatogenesis or that germ cells did not robustly colonize the gonads during early development. A direct comparison betweenBrca1^Δ11/Δ11p53^+/-(Fig. 3E,F) andBrca1^Δ11/Δ11p53^-/-(Fig. 3G,H) testes demonstrated significant lower rates of apoptosis in p53^-/- mice than in p53^+/- mice. In ten, randomly selected TUNEL-positive seminiferous tubules fromBrca1^Δ11/Δ11p53^-/- mice, 12% of cells were apoptotic at P16 and 22% at P21 compared to 40% at P16 and 50% at P21 in Brca1^Δ11/Δ11p53^+/- testes(Fig. 3B). Histological sections demonstrated a graded increase in the total number of spermatocytes when the number of wild type p53 alleles decreased from 2(p53^+/+) to 0 (p53^-/-) that correlated with the decreased number of apoptotic cells in testes isolated fromBrca1^Δ11/Δ11p53^+/+,Brca1^Δ11/Δ11p53^+/- andBrca1^Δ11/Δ11p53^-/- mice. These results indicated that Brca1 mutation triggered apoptosis through p53-dependent and p53-independent mechanisms.

To determine whether or not the homologous synapsis process was impaired byBrca1 mutation, testicular chromosome spreads were stained with an antibody to synaptonemal complex protein 3 (SCP3). SCP3 is one of the components of the axial and lateral element of the synaptonemal complex. It appears in leptotene stage spermatocytes and disappears in late meiotic cells(Dobson et al., 1994). Analysis of >2000 spermatocytes fromBrca1^Δ11/Δ11p53^+/- and control mice at P10-P21 demonstrated that the chromosomes from spermatocytes ofBrca1 mutant mice behaved normally until the pachytene stage(Fig. 4A-D). Only 3% of the cells showed delayed pairing of one or two chromosomes in Brca1mutant mice (data not shown). Chromosomal spreads prepared from P16Brca1 mutant and control mice demonstrated that approximately one third of spermatocytes in both strains completed synapses of homologous chromosomes (Fig. 4G). These observations were consistent with our earlier histological findings demonstrating that mutant spermatocytes developed relatively normally until the pachytene stage.

During normal spermatogenesis, diplotene stage spermatocytes begin to appear at P16 and represent 20-25% of the total spermatocyte population at P21(Fig. 4E). A hallmark of chromosome behavior at these stages is the extensive separation of homologous chromosomes except for the crossing-over sites that are held by chiasmata(arrows, Fig. 4E). By contrast,few homologous chromosomes demonstrated separation in the spermatocytes fromBrca1 mutant mice and those few spermatocytes that were not already eliminated through apoptosis showed fragmented cores/synaptonemal complexs(Fig. 4F). In fact, <1% of spermatocytes from Brca1 mutant mice exhibited even partial separation of their chromosomes (Fig. 4G). These observations demonstrated that spermatocytes fromBrca1 mutant mice were arrested at the late pachytene stage and failed to enter the diplotene stage.

Many DNA damage-repair proteins, including MLH1, RAD51 and DMC1, are involved in the crossing-over process. To investigate the molecular mechanisms underlying the failure of crossing-over formation in spermatocytes fromBrca1 mutant mice, the expression patterns and sites of cellular localization of Mlh1, Rad51 and Dmc1 were determined.

In control mice, Mlh1 first appeared as distinct spots along the chromosomes of late zygotene spermatocytes and was maintained until the mid-pachytene stage in spermatocytes (Fig. 5A, n=267). By contrast, no Mlh1 foci were observed on the majority (156 out of 162) of chromosome spreads prepared from late zygotene to mid-pachytene stage spermatocytes of Brca1 mutant mice. The remaining six spermatocytes showed very weak staining(Fig. 5B). Western blotting was performed to determine whether the abnormal distribution of Mlh1 on the chromosomes from Brca1 mutant mice was due to reduced protein concentrations of either Mlh1 or its binding partner Pms2. There were no obvious differences in the concentrations of these proteins betweenBrca1 mutant and control mice at any of the stages examined(Fig. 5D). In conclusion,although the distribution of Mlh1 on chromosomes was abnormal inBrca1 mutant mice, overall expression levels were normal. BecauseBrca1 mutant females were fertile, localization of Mlh1 was also checked in female mice. Nearly all of the chromosomes isolated from 200 late zygotene to mid-pachytene stage oocytes taken from E16.5-17.5Brca1^Δ11/Δ11p53^+/- female embryos had Mlh1 foci. There were no obvious differences in either the localization pattern or intensity of the foci between Brca1 mutant(Fig. 5C) and control (not shown) oocytes.

Rad51 and Dmc1 are homologs of Escherichia coli RecA. They bind to DSBs and are involved in meiotic recombination repair(Masson and West, 2001). Because BRCA1 colocalizes with RAD51 on meiotic chromosomes and DMC1 colocalizes with RAD51 (Masson and West,2001; Scully et al.,1997), we tested whether Brca1 mutation affected localization of these proteins. In control spermatocytes, Rad51 foci started to appear on the chromosomes of leptotene-stage spermatocytes(Fig. 6A, n=176) and disappeared on chromosomes from mid-pachytene-stage spermatocytes (data not shown). Analyses of 211 spermatocytes from three Brca1 mutant mice demonstrated significantly diminished numbers of Rad51 foci on all chromosomes(Fig. 6B). Western blots were performed to see if the failure to detect Rad51 on chromosomes was caused by reduced Rad51 protein concentrations. This analysis revealed comparable levels of Rad51 in Brca1 mutant and control animals(Fig. 6E), suggesting that the absence of full-length Brca1 isoform affected the localization of Rad51 but not the abundance. There were no apparent differences in Dmc1 localization and concentration between Brca1 mutant and control mice on chromosomes isolated from >300 spermatocytes (Fig. 6C-E).

H2AX is a member of the mammalian histone H2A family (reviewed byRedon et al., 2002). It is phosphorylated and relocated to DSBs within minutes of irradiation, which is consistent with a crucial role for this gene in DSB repair(Paull et al., 2000). H2AX-deficient mice display defective spermatogenesis and diminished numbers of Brca1 chromosomal foci (Celeste et al.,2002). To determine if Brca1 mutation alters H2AX phosphorylation and localization, chromosome spreads from P18, P21 and P28 testes of Brca1 mutant (n=6) and control (n=6) mice were prepared and subjected to double staining with an antibody to SCP3 and an antibody specific for phosphorylated H2AX (γH2AX). In control mice, we observed changes in γH2AX staining at different stages of spermatogenesis. It appeared first as diffused staining in leptotene-stage spermatocytes (Fig. 7A) but evolved into distinctly stained foci on chromosomes from late leptotene and early zygotene-stage spermatocytes (Fig. 7B). Beginning at the late zygotene stage, staining was restricted to the sex (XY) body and became gradually brighter and more condensed (arrow,Fig. 7C) untill γH2AX localization was restricted to the sex body in almost all the cells examined(98%, 288 out of 304) throughout the pachytene stage(Fig. 7D). In Brca1mutant mice, the pattern of γH2AX localization was similar to control mice at the leptotene and early zygotene stages(Fig. 7F,G). However, at the late zygotene and early pachytene stages, >90% (388 out of 427) of spermatocytes failed to show localization of γH2AX to the XY body. Instead, it remained as multiple distinct foci(Fig. 7H,I). The intensity of staining in the 10% of cells that demonstrated γH2AX localization in the XY body was decreased significantly (data not shown). The differences in the percentages of spermatocytes that showed γH2AX staining in the presence and absence of Brca1 mutation was highly significant(P≤0.0001). This indicates that the localization pattern ofγH2AX was disrupted by Brca1 mutation, similar to the disruptions of the Mlh1 and Rad51 patterns of localization.

Next we investigated possible reasons for the prolonged presence ofγH2AX foci on the chromosomes and the absence of γH2AX in the XY body. Because γH2AX detects and binds to DSBs(Mahadevaiah et al., 2001), we suspected that the prolonged presence of γH2AX foci in mutant cells could be caused by the accumulation of unrepaired DSBs. We further hypothesized that if sufficient DNA damage could be caused in control mice,they might exhibit defects similar to those observed in Brca1 mutant mice. To test this, we irradiated control and Brca1 mutant mice with 80 Gy of γ-irradiation to introduce acute, extensive DNA damage, and examined the formation of γH2AX foci at various time points post irradiation. Spermatocytes were isolated from the irradiated mice at 10 minutes, 30 minutes, 2 hours and 4 hours following irradiation. All irradiatedp53^+/- spermatocytes demonstrated prolonged γH2AX foci formation without XY body staining, similar to that found in the spermatocytes from unirradiated Brca1 mutant mice(Fig. 7D). There were no differences in the numbers of γH2AX foci in spermatocytes from irradiated Brca1 mutant mice compared to those from irradiated control mice (Fig. 7H).

Unwinding of parental DNA strands is a prerequisite for DNA replication,repair and recombination. The unwinding of duplex DNA is catalyzed by DNA helicases, which destabilize the hydrogen bounds between complementary double strands of DNA (Kanemaki et al.,1999). The absence of crossing-over inBrca1^Δ11/Δ11 p53^+/- mutant spermatocytes prompted us to measure the expression of genes that encode proteins with DNA helicase, branch migration and basic transcription activity. RNA samples from P14, P16, P21 and P22 were prepared fromp53^+/- control and Brca1^Δ11/Δ11p53^+/- mutant mice. An appropriately sized RT-PCR product for p47, which encodes a RuvB-like DNA helicase, was detected in testes from both control and Brca1 mutant mice. In RNA isolated from control testicular tissue, the RNA encoding p47 began to increase at P16 and was maintained at high levels through P22 (Fig. 8A). By contrast, the concentration of p47 RNA did not increase in testicular tissue from Brca1 mutant animals(Fig. 8A). The absence of an increase in p47 RNA during spermatogenesis in the absence of full-length Brca1 is consistent with the possibility that Brca1 could be a positive regulator of p47 expression during spermatogenesis.

Our data also revealed significant changes in the expression patterns of three additional genes that encode proteins that are involved in DSB repair. These are p62, a subunit of TFIIH, a basic transcription factor, XPB, a subunit of TFIIH with DNA-helicase activity, and TFIID, a basic transcription factor. As with p47, the concentrations of RNA encoding each of these proteins gradually increased from P14 to P22 in control animals, but not in spermatocytes of Brca1 mutant mice(Fig. 8A). Brca1 is known to regulate gene expression through both p53-dependent and p53-independent mechanisms. To determine if the changes in expression patterns of these genes were p53 dependent, patterns of gene expression were compared betweenBrca1 mutant and wild type mice in a p53^-/-background. Expression of all four genes was significantly lower inBrca1^Δ11/Δ11p53^-/- cells than inBrca1^+/+p53^-/- cells(Fig. 8B), indicating that the changes in expression patterns were independent of p53.

Because defective spermatogenesis began at P16 in Brca1 mutant animals, one possible reason for the lack of increased expression of these genes in Brca1 mutant mice could be the lack of spermatocytes at later ages (such as diplonema). If this were the case, lower expression levels would not necessarily be associated with absence of full-length Brca1. To test if lower expression levels directly correlated with absence of full-length Brca1, gene expression was examined in Brca1 mutant embryos isolated from E9.5, E10.5 and E16.5 (Fig. 8C). These analyses demonstrated a direct correlation between the absence of full-length Brca1 and lower expression levels of p47, p62, XPB and TFIID, which is consistent with the notion that Brca1 is a positive regulator of these genes.

This study demonstrated an essential role for full-length Brca1 in spermatogenesis. Brca1 is required either directly or indirectly for repair of the double-stranded breaks that are introduced during the normal process of spermatocyte meiosis. By contrast, Brca1 does not appear to be required for oocyte meiosis. In Brca1 mutant mice spermatogenesis did not advance beyond the pachytene stage of prophase during meiosis I and was accompanied by increased rates of apoptosis. Prophase of meiosis I involves in recombination between homologous chromosomes (reviewed byCohen and Pollard, 2001;Roeder, 1997). This process requires extensive DNA damage-repair activity because defective meiosis can result from the individual loss-of-function of several different factors involved in DNA-damage repair (reviewed byCohen and Pollard, 2001;Dasika et al., 1999;Tarsounas and Moens,2001).

The majority of homologous chromosomes synapsed normally at the pachytene stage in spermatocytes isolated from Brca1 mutant mice, indicating that Brca1 is not required for this process. However, spermatocytes from Brca1 mutant mice were arrested at the late pachytene stage and failed to enter the diplotene stage. This defect correlated with the absence of Mlh1 foci on the homologous chromosomes. Because Mlh1 foci localize to crossing-over sites on meiotic chromosomes and serve as a molecular marker for this event (Baker et al., 1996;Moens et al., 2002), it appears that full-length Brca1 is required for crossing-over. Crossing-over is a multi-step process that is initiated by a DSB in DNA, followed by exonucleolytic digestion, strand invasion, Holliday junction formation, branch migration and junction resolution, and, finally, heteroduplex DNA formation(reviewed by Cohen and Pollard,2001; Roeder,1997). It is possible that Brca1 is not required for the early steps of crossing-over because the phenotype of Brca1-deficient spermatocytes appeared at a later stage than those reported for SPO11-deficient(Romanienko and Camerini-Otero,2000), ATM-deficient (Barlow et al., 1998), DMC1-deficient(Yoshida et al., 1998) and MSH4-deficient (Kneitz et al.,2000) spermatocytes.

Our data indicate a link between BRCA1 function and the proper localization of MLH1. Previous investigations revealed that MLH1, together with either MSH4/MSH5/PMS2 or MSH4/MSH5/MLH3, are involved in both mismatch repair and crossing-over (Cohen and Pollard,2001). BRCA1 also displays an affinity for branched DNA structures and forms protein-DNA complexes cooperatively between multiple DNA strands without DNA sequence specificity (Paull et al., 2001). The results of the experiments presented here are consistent with the possibility that BRCA1 acts to recruit MLH1 to sites of recombination.

There are both similarities and differences between the phenotypes ofBrca1 and Mlh1 deficient spermatocytes(Baker et al., 1996;Eaker et al., 2002). In both strains, the majority of homologous chromosomes undergo synapsis. A 10-100 fold reduction in chiasma formation is found in Mlh1 mutant mice whereas <1% of spermatocytes in Brca1 mutant mice show any sign of chiasma formation and spermatocytes deficient in either protein have high rates of apoptosis. However, although Brca1 mutant spermatocytes died at the end of the pachytene stage, the majority of Mlh1 mutant spermatocytes die during metaphase (Eaker et al., 2002). It is interesting to note that the effect ofBrca1 mutation on formation of Mlh1 foci is male specific because Mlh1 foci are formed normally in oocytes.

In an attempt to further define the interaction between BRCA1 and MLH1, we performed co-immunoprecipitation experiments between BRCA1 and MLH1 in human spermatocytes but could not detect any direct binding between these proteins(data not shown). Thus, the interaction between BRCA1 and MLH1 may be mediated indirectly through other proteins. Consistent with this, a mass-spectrometry study showed that BRCA1 and other proteins, including MSH2, MSH6, MLH1, ATM,BLM, RAD50, MRE11, NBS and RFC, form a complex called BRCA1-assocated genome-surveillance complex (BASC) (Wang et al., 2000) that is postulated to be involved in the recognition and repair of aberrant DNA structures. Wang et al.(2000) were able to demonstrate binding of BRCA1 to MSH2 and MSH6, both factors that bind directly to MHL1.

A candidate approach was used to study the localization and expression of several genes and gene products that are involved in various aspects of DNA-damage repair. We showed that distribution of Rad51 on chromosomes was disrupted in testes from Brca1 mutant mice. This finding is significant because Rad51 is known to interact with Brca1 either directly or indirectly after treatment with DNA-damaging reagents(Huber et al., 2001;Scully et al., 1997). Therefore, the abnormal distribution of Rad51 in mutant chromosomes provides strong supporting evidence for the hypothesis that full-length BRCA1 is required for the repair of DNA during meiosis in spermatocytes. Notably, the localization of Dmc1, a closely related partner of Rad51(Masson and West, 2001;Tarsounas et al., 1999), was not altered, which indicates that loss of full-length Brca1 disrupts the function of some but not all DNA damage-repair proteins.

Recent investigations have highlighted the interaction between γH2AX and BRCA1 and the importance of γH2AX in DNA-damage repair. After using a laser beam to generate a DSB, γH2AX becomes phosphorylated and moves to the site of the lesion within minutes(Paull et al., 2000). BRCA1 moves to the same site approximately 45 minutes later. This suggests thatγH2AX serves as a DNA damage sensor and that it is responsible for recruiting BRCA1 to sites of DNA damage. In fact, BRCA1 foci do not form on chromosomes from mice that lack H2AX, which results in increased genetic instability (Celeste et al.,2002).

At the late zygotene and early pachytene stages of normal spermatogenesis,γH2AX relocalizes from sites of DSBs on chromosomes to the XY body. This did not occur in the absence of full-length Brca1. Instead, γH2AX remained as multiple, distinct foci on chromosomes. Because γH2AX binds to DNA prior to binding Brca1 (Paull et al., 2000), it is likely that the binding of γH2AX to DSBs is independent of Brca1 and that the absence of full-length Brca1 did not alter this process in spermatocytes. However, loss of full-length Brca1 did block relocalization of γH2AX to the XY body. There are two possible explanations for this observation. First, recruitment of Brca1 by γH2AX may be essential for repairing DNA damage in spermatocytes. Therefore, in the absence of full-length Brca1, unrepaired DNA damage accumulates in the spermatocytes, which traps γH2AX and prevents it from moving to the XY bodies. Experimental support for this possibility comes from the observation that the phenotypes of control spermatocytes that sustained extensive DNA damage from γ-irradiation and that of Brca1 mutant spermatocytes were similar. Alternatively, Brca1 could play a role that actively recruitsγH2AX to the XY body. The absence of Brca1 consequently results in the failure of γH2AX localization to the XY body. Recently, it as been proposed that Brca1 is involved in X-chromosome inactivation(Ganesan et al., 2002). Future studies are needed to determine whether loss of localization of γH2AX to the XY body in Brca1 mutant spermatocytes affects X-inactivation.

Although BRCA1 appears to regulate the expression of many genes that are involved in multiple biological processes(Chen et al., 1999;Deng and Brodie, 2000), only three of these genes are known to play a role in DNA-damage repair. These areGadd45 (a DNA damage-repair and response gene)(Harkin et al., 1999),Ddb2 (which is defective in Xeroderma Pigmentosum group E cells and encodes the p48-damaged DNA-binding protein)(Takimoto et al., 2002) andXpc (a gene encoding Xeroderma Pigmentosum group C complementing protein) (Hartman and Ford,2002). All three genes are regulated by p53 and are involved in the nuclear excision-repair pathway(Hartman and Ford, 2002;MacLachlan et al., 2002;Takimoto et al., 2002). In this study, we found that loss of full-length Brca1 correlated with decreased expression of several genes that encode proteins involved in the repair of DSBs, including RuvB-like DNA helicase, TFIIH (p62) and TFIID. Additional investigations will determine if these genes are transcriptional targets of BRCA1 and if this regulation is direct or indirect.

The p53-dependent apoptotic pathway plays a dominant role in some but not all organs (reviewed by Burns and El-Deiry,1999). We showed that spermatocyte apoptosis triggered by loss of full-length Brca1 was mediated by p53-dependent and p53-independent pathways. The most likely proximal trigger for both p53-dependent and p53-independent apoptosis in Brca1-deficient spermatocytes is the development of genetic instability. Thus, mutant spermatocytes that escaped the first wave of p53-dependent apoptosis survived slightly longer but still failed to progress into the diplotene stage. This suggests that activation of more than one apoptotic pathway ensures that spermatogenesis cannot be completed in the presence of significant genetic instability.

In summary, this study revealed that Brca1 plays an essential role in DNA-damage repair during spermatogenesis(Fig. 9). Brca1 functions through at least two different mechanisms, the recruitment of DNA damage-repair proteins to sites of DNA damage, and the regulation of the expression of DNA damage-repair genes. Examples of the first mechanism include the diminished numbers of Rad51 and Mlh1 foci on chromosomes in the absence of full-length Brca1. Examples of the second mechanism are the altered expression levels of RuvB-like DNA helicase, XPB, p62 and TFIID in the absence of full-length Brca1. Thus, Brca1-deficient spermatocytes unavoidably accumulate unrepaired DNA damage that triggers both p53-dependent and p53-independent apoptosis, and failure of spermatogenesis.

We thank Drs L. P. K. Jones, M. Tilli, R. H. Wang, L. Cao, W. Qiao, and other members of Dr Furth's laboratory and Dr Deng's laboratory for their helpful discussions.