Results

Trip13 Is a Widely Expressed Mammalian Ortholog of PCH2 with Unusual Phylogenetic Relationships
The mammalian ortholog of PCH2, Trip13 (thyroid hormone receptor interacting protein 13), encodes a protein with extensive amino acid homology in regions alignable to the yeast and worm orthologs (Figure S1) [12]). Interestingly, phylogenetic analysis of TRIP13/Pch2p shows that the mammalian protein clusters more closely to plants than it does to the evolutionarily more closely related worms and flies (Figure 1A; see Discussion). Semi-quantitative reverse-transcriptase PCR (RT-PCR) analysis showed Trip13 mRNA to be expressed in a variety of embryonic and adult tissues, including testis (Figure 1B), consistent with mouse and human EST data summarized in Unigene (http://www.ncbi.nlm.nih.gov/UniGene). It is also highly expressed in human and mouse oocytes [17].
Figure 1  The Mouse PCH2 Ortholog TRIP13 and Expression in Wild Type and Mutant
(A) Phylogenetic tree of presumed PCH2/TRIP13 orthologs. The database sequence accession number of each protein is presented in Table S1. Numbers shown are bootstrap values (see Materials and Methods). The clustering was the same regardless of whether the whole entire AA sequence or trimmed sequences (where regions showing little conservation were removed) were used for the analysis. Major eukaryotic groups are indicated in color, with deuterostomia in blue, plants in green, protostomia in purple, and fungi in maroon.
(B) Amplification products of cDNA from the following tissues: 1, heart; 2, brain; 3, spleen; 4, lung; 5, liver; 6, skeletal muscle; 7, kidney; 8, testis; 9, E7 embryo; 10, E11 embryo; 11, E15 embryo; and 12, E17 embryo.
(C) Intron–exon structure of TRIP13 and insertion site of gene-trap vector. See Materials and Methods for details on how the precise insertion site was identified.
(D) RT-PCR of Trip13 and a control gene Med31 from testis RNA. The Trip13 primers are situated in the first and last exons (see Materials and Methods).
(E) Western blot analysis of testis protein with anti-TRIP13 antibody. The blot was later probed with anti-alpha tubulin actin as a loading control. The expected TRIP13 protein is ∼48 KDa.
(F) Localization of TRIP13 in testes. Wild-type (top) and mutant (bottom) testis sections were probed with chicken anti-TRIP13, and detected with HRP-conjugated anti-chicken IgG (brown/red staining). Expression in WT was most prominent in the nuclei of Type B spermatogonia (Sg), leptotene spermatocytes (LS), and early pachytene spermatocytes (PS), but not late pachytene spermatocytes (LP). No nuclear staining was seen in mutant testis sections, although reddish cytoplasmic background is present. Identification of cell types was judged in part by estimating the epithelial stage of the tubules as described [67].
(G) TRIP13 localization in surface-spread spermatocytes. Preparations were immunolabeled with anti-SYCP3 (S) and TRIP13 (T). Both individual and merged images are shown for leptotene (Lep), zygotene (Zyg), and pachytene (Pac) spermatocytes. Nuclear staining was absent in the mutant.

Generation of Trip13 Mutant Mice
To explore the function of TRIP13 in mammals, we generated mice with a gene trap-disrupted allele, Trip13RRB047 (Figure 1C; abbreviated as Trip13Gt). Heterozygotes were normal in all respects, but homozygotes were present at ∼2/3 the expected ratio from intercrosses between heterozygotes (91 Trip13 +/+, 183 Trip13Gt/+, and 61 Trip13Gt/Gt). Since >90% of prewean mice that died were mutant homozygotes, this discrepancy is apparently due to a partially penetrant lethality. Most surviving Trip13Gt/Gt animals were grossly normal. However, homozygotes that were semi-congenic (N4) on the C57BL/6J strain were often markedly smaller and/or had kinked or shorter tails (Figure 2A and 2B).
Figure 2  Developmental Phenotypes of Trip13 Mutant Mice
(A) Shown are 21-d-old littermates. Note the shortened tail in the mutant, but overall similar body size.
(B) Shown are 23-d-old littermates. The mutant is smaller in this case, but the tail is not as truncated as the mouse in (A).
(C) Wild-type (WT) and homozygous Trip13 mutant (MUT) testes.
(D) and (E) are cross sections through 17.5-d-old heterozygous (“WT”) and homozygous mutant Trip13 testes, respectively. Whereas the tubules in WT show coordinated spermatogenesis with pachytene spermatocytes present in all tubules (proximal to the lumen), developmental progression in the mutant is not synchronized between tubules. Some tubules have no pachytene spermatocyes (asterisks), while in others, development is somewhat disorganized (#). RT-PCR analysis of Trip13Gt expression (Figure 1D) revealed a low level of normally spliced transcripts in testes of homozygotes that is presumably a consequence of incomplete usage of the gene trap's splice acceptor. Western blot analysis, using a polyclonal antibody raised against a peptide encoded by exon 3, revealed multiple species in wild-type and heterozygous testes, one of which corresponds to the expected size of 48 kDa (Figure 1E). This and three other species were undetectable in homozygous mutant testes, but a reduced amount of an intense ∼38 kDa smaller band was present. It is not clear if this corresponds to TRIP13. The greatly decreased Trip13 mRNA and predicted correct-length protein in mutants indicate that the Trip13RRB047 allele is severely hypomorphic.
To determine the germ cell types in which TRIP13 is expressed, and to assess possible expression in the mutant by means other than Western analysis, testis sections were immunolabeled for TRIP13 using a polyclonal chicken antipeptide antibody (see Materials and Methods). The most intensely labeled cells in control testes were Type B spermatogonia and leptotene spermatocytes (Figure 1F). Zygotene/pachytene spermatocytes stained less strongly, and there was no detectable staining in late pachytene spermatocytes. TRIP13 appeared to be nuclear localized. There was no such staining of nuclei in mutant seminiferous tubules (Figure 1F). To further assess the nuclear localization, TRIP13 was used to probe meiotic chromosomes prepared by surface spreading of spermatocyte nuclei. In wild type, there was diffuse nuclear staining, and no evidence of concentration on SC cores (marked by the axial element protein SYCP3) at any meiotic substage (Figure 1G). TRIP13 signal was noticeably absent in mutant meiotic nuclei.

Infertility Due to Meiotic Disruption in TRIP13-Deficient Meiocytes
Homozygotes of both sexes had small gonads (Figure 2C; see below) and were invariably sterile. Ovaries of adult Trip13Gt/Gt females were severely dysmorphic and had few or no follicles (Figure 3A and 3B). The majority of oocyte loss occurred in late embryogenesis or early in postnatal development, since 2 d postpartum ovaries were markedly smaller than those of control littermates, and were lacking oocytes or newly forming follicles (Figure 3C and 3D). Thus, oocytes failed to progress to the dictyate (resting) phase. Since we observed oocytes with pachytene stage chromosomes in 17.5 d Trip13Gt/Gt embryonic ovaries (unpublished data), this indicates that oocytes were eliminated somewhere between pachynema and dictyate.
Figure 3  Histology of Mutant Gonads
All are hematoxylin/eosin-stained paraffin sections. Testes are from 6-wk-old males, except as indicated below.
(A) Wild-type 25-d-old ovary.
(B) Trip13Gt /Gt 25-d-old ovary, showing dysgenesis from an absence of oocytes.
(C) Trip13Gt/+ 2-d-old control ovary. Arrows point to oocytes in newly forming follicles.
(D) Trip13Gt/Gt 2-d-old ovary, dysgenic due to lack of oocytes. Magnification is the same as its littermate in “C.”
(E) Wild-type testis.
(F) Trip13Gt/Gt testis with uniform pachytene arrest.
(G) Trip13Gt/Gt 3-mo-old testis with some postmeiotic spermatids (arrows).
(H) Spo11−/- testis. A tubule with spermatocytes at leptotene/zygotene transition is labeled ZP, and tubules with apoptotic spermatocytes are marked with an asterisk. The specimen was taken from a littermate of that in (I).
(I) Spo11 −/− Trip13Gt/Gt testis. Labeling is the same as in (H). The inset contains a tubule with leptotene-zygotene spermatocytes.
(J) Mei1 −/− Trip13Gt/+ testis. The specimen was taken from a littermate of that in (K).
(K) Mei1 −/− Trip13Gt/Gt testis.
(L) Rec8Mei8/Rec8Mei8 Trip13Gt/+ testis. The Rec8Mei8 allele was described [39]. The specimen was taken from a littermate of that in (M).
(M) Rec8Mei8/Rec8Mei8 Trip13Gt/Gt testis.
(N) Dmc1 −/− Trip13Gt/Gt testis.
(O) Spo11 −/− Trip13Gt/+ 25-d-old ovary. The specimen was taken from a littermate of that in (P).
(P) Spo11 −/− Trip13Gt/Gt 25-d-old ovary.
(Q) Mei1 −/− Trip13Gt/+ 25-d-old ovary. The specimen was taken from a littermate of that in (R).
(R) Mei1 −/− Trip13Gt/Gt 25-d-old ovary.
(S) Rec8Mei8/Rec8Mei8 Trip13Gt/+ 25-d-old ovary. The specimen was taken from a littermate of that in (T).
(T) Rec8Mei8/Rec8Mei8 Trip13Gt/Gt 25-d-old ovary. Histological sections of mutant testes revealed a lack of postmeiotic cell types that are characteristic of wild-type seminiferous tubules (Figure 3E). The most developmentally advanced seminiferous tubules contained adluminal spermatocytes with condensed chromatin characteristic of pachynema (Figure 3F). The absence of coordinated spermatogenic progression beyond this stage is indicative of a pachytene arrest. This was revealed more clearly by chromosome analysis (see below). Some sections of adult seminiferous tubules contained postmeiotic spermatids (Figure 3G), although we saw no motile epididymal sperm. These drastic meiotic defects stand in contrast to yeast and C. elegans, in which deletion of Pch2 alone has minor effects on spore/gamete development [2,8].

TRIP13-Deficient Meiocytes Undergo Homologous Chromosome Synapsis Despite the Presence of Unrepaired DSBs in Pachynema
To better characterize the degree of meiotic progression in Trip13Gt/Gt spermatocytes, we immunostained chromosome spreads for SYCP3 and SYCP1, components of the axial/lateral elements and transverse filaments, respectively, of the synaptonemal complex (SC). Pachytene spermatocyte nuclei from postpubertal mutant testes could assemble normal SC cores and exhibited full synapsis of chromosomes as judged by colabeling of SYCP1 and SYCP3 along the full lengths of all autosomes (Figure 4A). Additionally, the X and Y chromosomes were normally synapsed at their pseudoautosomal region. More prepubertal (17.5 d postpartum) mutant spermatocytes contained asynaptic or terminally asynapsed chromosomes than age-matched controls (62.5% versus 25%, respectively; Figure 4B). We attribute this to a delay in the first wave of postnatal spermatogenesis (Figure 2D and 2E), likely related to systemic developmental retardation (Figure 2A and 2B). Nevertheless, since Trip13Gt/Gt spermatocytes progress to pachynema with no gross SC abnormalities, and oocytes were eliminated soon after birth (a characteristic of DNA repair mutants [13]), this suggested that unrepaired DSBs are responsible for eventual meiotic arrest and elimination.
Figure 4  Immunohistochemical Analysis of Pachytene Spermatocyte Chromosomes
Surface-spread chromosomes were immunolabeled with the indicated antibodies and fluorophores. As indicated in the upper right of each panel, cells were from wild type (WT, either +/+ or Trip13Gt/+) or Trip13Gt/Gt (Mut). There were no differences seen between heterozygotes and +/+ spermatocytes.
(A) A mutant pachytene nucleus with full synapsis. Areas of SYCP1/SYCP3 colabeling are yellow.
(B–E) Spermatocytes nucleus from 17.5 d postpartum mutant. Asynapsed chromosomes or regions of chromosomes are indicated by white and yellow arrows, respectively. Unlike the normal distribution in wild-type pachytene spermatocytes (C), BLM foci are present on synapsed pachytene chromosomes in the mutant (D). RAD51 foci, which are abundant earlier in prophase, disappear from autosomes in wild-type pachytene nuclei (E) and the bulk of staining is over the XY body (arrow).
(F) RAD51 persists on the synapsed mutant chromosomes (arrows).
(G) H2AX phosphorylation is restricted to the XY body in WT.
(H) In addition to a large area of γH2AX staining (arrow) over the XY body, there is extensive autosomal H2AX phosphorylation (arrows).
(I, J) Note that in wild-type pachytene spermatocytes, TOPBP1 is present only over the XY body (yellow arrow). In the mutant (J), an arrow denotes one area of intensive staining that may be over the sex chromosomes, but many other chromosome cores are positively stained.
(K, L) RPA persists along synapsed cores in the mutant, not WT.
(M, N) Arrows indicate examples of MLH3 foci on SCs.
(O) In WT late pachytene spermatocytes, RAD51 is present only at background levels.
(P) As in (F), extensive RAD51 staining delineates SCs in mutant pachytene nuclei (indicated by white arcs). MLH1 foci colocalize with these tracts (arrows) at the typical 1–2 foci per chromosome as in (M). To elucidate the cause of meiotic arrest, we analyzed meiotic chromosomes with a variety of markers that are diagnostic of recombination and synapsis. Recombination in Trip13Gt/Gt spermatocytes appeared to initiate normally as judged by the presence of γH2AX in leptonema (Figure S2A and S2B), which reflects the presence of meiotically induced DSBs [18]. RAD51 and/or DMC1, components of early recombination nodules (ERNs), was also present as abundant foci in Trip13Gt/Gt zygotene spermatocytes (unpublished data; the anti-RAD51 antibody cross-reacts with DMC1), indicating that recombinational repair of DSBs is initiated. The cohesin complex, which is essential for completion and/or maintenance of synaptic associations, appeared to assemble normally as judged by immunolabeling for the meiosis-specific cohesins STAG3 (Figure S2C and S2D) and REC8 (unpublished data). Because yeast PCH2 localizes to telomeres in a Sir3p-dependent manner, we tested for possible telomere defects by immunolabeling for TRF2, a component of a protein complex that plays an essential role in telomere protection [19]. It was localized to telomeres of both fully synapsed and telomerically asynaptic mutant chromosomes (Figure S2E and S2F).
Defects in DSB repair became apparent in pachynema upon probing of mutant spermatocyte nuclei with antibodies against molecules involved in various stages of recombination. In >99% of Trip13Gt/Gt chromosome spreads, BLM helicase (Figure 4C and 4D), RAD51/DMC1 (Figure 4E and 4F), γH2AX (Figure 4G and 4H), and TOPBP1 (Figure 4I and 4J) all persisted abnormally on synapsed chromosomes. For RAD51/DMC1, mutant pachytene spermatocytes contained 138 ± 6 foci (compared to 11 ± 3 foci in wild type, most of which were on the XY body), down from 218 ± 13 in zygonema (compared to 220 ± 13 foci in wild type). TOPBP1 is a DNA damage–checkpoint protein involved in ATM protein–dependent activation of ATR protein [20,21]. It binds sites of DSBs and unsynapsed regions of meiotic chromosomes [22,23]. BLM has been reported to colocalize with markers (RPA and MSH4) of recombination at sites distinct from those that become resolved as crossovers (CO) [24]. We therefore assessed the distribution of RPA, the ssDNA binding protein, which is normally present at focal sites of synapsing meiotic chromosomes before disappearing in mid-pachynema [25]. It is thought to bind D-loops of recombination intermediates [26]. RPA also persisted on pachytene mutant chromosomes (Figure 4K and 4L). These data indicate that unrepaired DSBs, or unresolved recombination intermediates, remain in pachynema and activate a DNA damage checkpoint system.
It should be noted that chromosomes affected by meiotic sex chromosome inactivation (MSCI) and meiotic silencing of unpaired chromatin (MSUC) are heavily stained by antibodies for several DSB repair-associated molecules, including γH2AX. H2AX phosphorylation due to MSCI and MSUC is conducted by ATR, not ATM [27–29]. Since mutant chromosomes are fully synapsed, and MSUC is known to occur only as a result of asynapsis, the decoration of Trip13Gt/Gt chromosomes with DNA repair markers is probably attributable to incomplete DNA repair rather than transcriptional silencing.
Consistent with the presence of rare (<1%) Trip13Gt/Gt pachytene spermatocytes devoid of persistent DNA repair markers, and testis histology showing some degree of postmeiotic progression (Figure 3G), we observed both diplotene nuclei that lacked autosomal RAD51/DMC1 and γH2AX (Figure S3A–S3D), and also metaphase I spreads with 20 bivalents (Figure S3E–S3F). Since Trip13Gt may not be a complete null, these diplotene and metaphase I spermatocytes might arise by virtue of having sufficient wild-type TRIP13.

CO-Associated Markers Appear Normally in the Absence of TRIP13
The persistence of BLM on Trip13Gt/Gt spermatocyte chromosomes suggests that at least a subset of the unrepaired DSBs correspond to sites of defective NCO recombinational repair. To assess whether CO recombination occurs in the mutant, we examined the distribution of the mismatch repair proteins MLH1 and MLH3, which are normally detectable as foci in mid-late pachynema and mark the locations of chiasmata [30,31]. Remarkably, MLH1/3 foci were formed; we observed 1–2 foci/chromosome as in wild type and at typical overall levels (MLH3 = 23 ± 2, N = 10; [30,32]) on mid-late pachytene chromosomes (Figure 4M and 4N; MLH1 not shown). Since <1% of Trip13Gt/Gt pachytene nuclei had normal repair (as judged by absence of persistent DSB repair markers; see above), but most of the pachytene nuclei had MLH1/3 foci, it was unlikely that the MLH1/3 foci formed only on chromosomes with fully repaired DSBs. To test this directly, we conducted double staining for MLH1 and RAD51/DMC1. MLH1 foci were present on chromosomes that also contained numerous RAD51/DMC1 foci (Figure 4O and 4P).
To assess whether these MLH1/3 foci in Trip13Gt/Gt pachytene spermatocytes represent CO events completed to a point where they could maintain interhomolog attachments though the end of prophase I, we treated testicular cells from 17.5–20.5-d-old control (+/+), Trip13Gt/Gt, and Dmc1 −/− mice with the protein phosphatase inhibitor okadaic acid (OA), a chemical that induces degradation of the SC, chromosome condensation, and premature progression to metaphase I [33]. Fifteen metaphase spreads were identified for each genotype. Whereas all of the Dmc1 −/− spreads had ∼35 or more condensed chromosomes, all of the +/+ and Trip13Gt/Gt spreads had 20–25, suggesting that the MLH1/3 foci in Trip13Gt/Gt pachytene spermatocytes represent sites of completed, or near-completed, COs. Because the preparations were made from whole testes, it is possible that the univalent-containing metaphases from Dmc1 −/− mice were from spermatogonia, not spermatocytes.

TRIP13 Deficiency Does Not Alleviate Meiotic Arrest Phenotypes of Mutants Defective in Synapsis
To determine if TRIP13 deficiency prevents apoptosis triggered by asynapsis as in C. elegans, we analyzed mice that were doubly mutant for Spo11 and Trip13. SPO11 is a transesterase that is essential for the creation of genetically programmed DSB during leptonema of many organisms, including mice [18]. In C. elegans, spo-11 mutant gametes have extensive asynapsis, which triggers PCH-2 dependent apoptosis in pachynema [2]. In mice, Spo11 −/− spermatocytes are severely defective in homologous chromosome synapsis [34,35], and arrest with chromosomes in a state characteristic of the zygotene/pachytene transition (Figure 3H). Spermatocytes in Trip13Gt/Gt Spo11 −/− testes progressed maximally to that point before undergoing death (Figs 3I), well before the spindle checkpoint that eliminates achiasmate spermatocytes [36]. There was no evidence of metaphase I spermatocytes or postmeiotic spermatids in these testes, unlike those seen in Trip13 single mutants (Figure 3G). In contrast to the complete synapsis in Trip13Gt/Gt pachytene spermatocytes (Figure 5A), in which SPO11 is available in leptonema to initiate (via DSB induction, Figure S2A and S2B) a recombination-driven homolog search, chromosome synapsis in doubly mutant spermatocytes was highly disrupted as in Spo11 single mutants (Figure 5B and 5C). Identical studies were performed with mice deficient for Mei1, a vertebrate-specific gene also required for DSB formation and chromosome synapsis [37], with similar results (Figure 3J and 3K; immunocytology not shown).
Figure 5  Immunocytological Analysis of Trip13 Compound Mutants
Surface-spread chromosomes were immunolabeled with the indicated antibodies and fluorophores. Genotypes are indicated, as are those panels in which dual staining patterns are merged. Note that (H) and (I) are at lower magnification to show multiple nuclei. In yeast, deletion of PCH2 alleviates the pachytene arrest caused by asynaptic mutants zip1 and zip2 [8]. Although mouse SYCP1 might be a functional equivalent of Zip1p, because Sycp1 mutant spermatocytes arrest at approximately the same point as Trip13 mutants, there would be no opportunity to observe bypass of Sycp1 −/−. Since Zip2p is present at sites of axial associations, even in zip1 mutants, it has been suggested that Zip2p promotes initiation of chromosome synapsis [38]. These observations raise the possibility that in yeast, Pch2p responds to synapsis polymerization rather than initiation. To test this, we performed epistasis analysis with a Rec8 allele (Rec8Mei8, abbreviated as Rec8 −). Meiotic chromosomes of Rec8 mutant spermatocytes undergo apparent homolog pairing and interhomolog synaptic initiation, but are defective in DSB repair and fail to maintain interhomolog synapsis [39,40]. Rather, sister chromatids appear to synapse and are bound by SYCP1 along their axes. Rec8 mutants do not progress to diplonema or metaphase I. Double mutant analysis indicated that Rec8 is epistatic to Trip13. As in the Spo11 and Mei1 experiments, histology of testes deficient for both REC8 and TRIP13 resembled the Rec8 mutant, with no evidence of progression to metaphase I that occurs in Trip13Gt/Gt mice (Figure 3L and 3M). Immunocytological analysis of spread chromosomes showed a failure of homologous chromosome synapsis in both the Rec8 −/− and Rec8 −/− Trip13Gt/Gt spermatocytes, as previously reported for Rec8 mutants (Figure 5D and 5E) [39,40].
Although subsequent reports indicate otherwise [10,12], deletion of PCH2 in yeast was originally reported to alleviate meiotic arrest caused by deficiency for the meiosis-specific RecA homolog DMC1 [8]. To investigate this relationship in mice, we constructed animals doubly mutant for Trip13 and Dmc1. As in Dmc1 −/− mice, in which spermatocytes undergo meiotic arrest from defective DSB repair and failed chromosome synapsis [16], spermatogenesis in Dmc1 −/− Trip13Gt/Gt testes was uniformly arrested at the point where spermatocytes contained chromatin characteristic of zygonema/pachynema (Figure 3N). Immunocytological analysis indicated that both Dmc1 −/− and Dmc1 −/− Trip13Gt/Gt chromosomes had extensive asynapsis compared to Trip13Gt single mutants (Figure 5F–5H), and all had persistent RAD51/DMC1 foci and phosphorylated H2AX (γH2AX; Figure 5I–5L), confirming that Dmc1 is epistatic to Trip13. Doubly mutant females had residual ovaries, phenocopying Dmc1 −/− and Trip13Gt/Gt single mutants (unpublished data).

Meiotic Defects in Trip13Gt/Gt Oocytes Are DSB-Dependent
Epistasis analysis of females was insightful with respect to the cause of arrest in Trip13 mutants. Both Mei1 −/−/Trip13Gt/Gt and Spo11 −/−/Trip13Gt/Gt females had ovaries with numerous follicles, identical to Mei1 and Spo11 single mutants (Figure 3O–3R). Thus, Spo11 and Mei1 are epistatic to Trip13 in oogenesis, just as they are to Dmc1 [13,41]. This demonstrates that oocyte loss in Trip13Gt/Gt females is dependent on DSB formation. In conjunction with the immunohistochemical data, these data provide strong evidence that meiotic arrest in Trip13 mutant mice is due to defects in DSB repair. As expected, ovaries of Rec8 Trip13 double mutants were devoid of oocytes as were those from either single mutant (Figure 3B, 3S, and 3T).