Caenorhabditis elegans p97 controls germline-specific sex determination by controlling the TRA-1 level in a CUL-2-dependent manner.

p97 (CDC-48 in Caenorhabditis elegans) is a ubiquitin-selective AAA (ATPases associated with diverse cellular activities) chaperone and its key function is to disassemble protein complexes. p97 functions in diverse cellular processes including endoplasmic reticulum (ER)-associated degradation, membrane fusion, and meiotic and mitotic progression. However, its cellular functions in development have not yet been clarified. Here, we present data that p97 is involved in the switch from spermatogenesis to oogenesis in the germline of the C. elegans hermaphrodite. We found that the cdc-48.1 deletion mutant produced less sperm than the wild type and thus showed a decreased brood size. The cdc-48.1 mutation suppressed the sperm-overproducing phenotypes of fbf-1 and fem-3(gf) mutants. In addition, the p97/CDC-48–UFD-1–NPL-4 complex interacted with the E3 ubiquitin ligase CUL-2 complex via NPL-4 binding to Elongin C. Furthermore, TRA-1A, which is the terminal effector of the sex determination pathway and is regulated by CUL-2-mediated proteolysis, accumulated in the cdc-48.1 mutant. Proteasome activity was also required for the brood size determination and sperm-oocyte switch. Our results demonstrate that the C. elegans p97/CDC-48–UFD-1–NPL-4 complex controls the sperm-oocyte switch by regulating CUL-2-mediated TRA-1A proteasome degradation.


Introduction
Caenorhabditis elegans occurs naturally in two sexes, selffertilizing hermaphrodites and males. Males have a male soma and a germline that produces sperm continuously from the fourth larval stage (L4) throughout adulthood. By contrast, the hermaphrodite soma is female; its germline produces an average of 300 sperm during the L4 stage and then produces oocytes continuously throughout adulthood. The switching mechanism from spermatogenesis to oogenesis (i.e. the sex determination mechanism) involves many negative regulatory factors and is highly complex (Ellis and Schedl, 2006;Kimble and Crittenden, 2007). A regulatory hierarchy controls the activity of the master sexual regulator TRA-1 (Hodgkin, 1987;Zarkower and Hodgkin, 1992) (Fig. 3A). In the hermaphrodite germline, tra-1 would initially be switched off (spermatogenesis) but, later in development, would be switched on (oogenesis). tra-1 encodes two protein: TRA-1A (1109 amino acids, 135 kDa), which contains five zinc fingers and TRA-1B (287 amino acids, 37 kDa), which contains only the first two zinc fingers (Zarkower and Hodgkin, 1992;Zarkower and Hodgkin, 1993). TRA-1A belongs to the GLI family of transcription factors including mammalian proteins GLI and GLI3 and the Drosophila protein Cubitus interruptus (Ci). TRA-1A, but not TRA-1B, binds DNA in vitro and functions as a transcription regulator for egl-1, mab-3, and fog-3 gene expression (Zarkower and Hodgkin, 1993;Conradt and Horvitz, 1999;Chen and Ellis, 2000;Yi et al., 2000).
Three fem genes located immediately upstream of TRA-1 in the genetic sex determination pathway promote the male fate by inhibiting TRA-1. FEM-1 contains Ankyrin repeats but its biochemical activity remains unknown (Spence et al., 1990); FEM-2 is a Type 2C protein Ser/Thr phosphatase (Pilgrim et al., 1995;Chin-Sang and Spence, 1996); and FEM-3 is a novel protein with no obvious motifs (Ahringer et al., 1992). Based on the fact that the loss of any of the FEM proteins causes TRA-1A accumulation, it was proposed that the role of FEM proteins in sex determination might be to stimulate TRA-1A degradation (Schvarzstein and Spence, 2006). Recently, it has been demonstrated that TRA-1 is regulated by ubiquitin-proteasome degradation mediated by a CUL-2 complex containing FEM proteins as a substrate recognition receptor (Starostina et al., 2007). CUL-2 belongs to the cullin family of E3 ubiquitin ligases and CUL-2 complexes consist of the following: CUL-2, which forms a rigid scaffold; the RBX-1 RING-H2 finger subunit that binds to the C-terminus of CUL-2; the adaptor protein ELC-1/Elongin-C, which binds to the N-terminus of CUL-2 and to the ubiquitin-like protein ELB-1/Elongin-B; and a variable substrate-recognition receptor that binds to ELC-1/Elongin-C (Kipreos, 2005;Petroski and Deshaies, 2005).
FEM-1 and FEM-3 proteins are targeted by the F-box protein SEL-10 and are proteolytically regulated (Jager et al., 2004). In addition, fem-3 is regulated post-transcriptionally by FBF-1 and FBF-2 (fem-3 mRNA binding factors), which are nearly identical regulators of the PUF (Pumilio and FBF) protein family (Zhang et al., 1997). Since fbf-1 fbf-2 double mutants fail to switch from spermatogenesis to oogenesis, it has been proposed that FBF-1 and FBF-2 function redundantly to promote the sperm-oocyte switch (Zhang et al., 1997). As described above, many factors involved in this switch have been identified, and their genetic interactions have been well documented. However, the molecular mechanisms of regulation of their amounts and/or activities remain poorly understood.
p97 ATPase -also called valosin-containing protein (VCP), Cdc48p in yeast, or CDC-48 in C. elegans -is one of the bestcharacterized examples among AAA (ATPases associated with diverse cellular activities) family proteins (Patel and Latterich, 1998;Ogura and Wilkinson, 2001). It has been suggested that p97 is a ubiquitin-selective chaperone and that its key function is to disassemble protein complexes (Rape et al., 2001). p97 has important roles in ER-associated degradation (ERAD), membrane fusion, activation of transcription factors, spindle disassembly, DNA replication and other protein quality-control activities (Wang et al., 2004;Uchiyama and Kondo, 2005;Ye, 2006). Functional diversity of p97 is mainly determined by differential binding of distinct cofactors or adaptor proteins (Ye, 2006;Schuberth and Buchberger, 2008). It has also been demonstrated that the interaction between p97 and the E3 ubiquitin ligases gp78 and Hrd1 has an important role in ERAD (Zhong et al., 2004;Lilley and Ploegh, 2005;Ye et al., 2005). p97 is encoded by a single gene in humans, mice and yeast, and is highly conserved. By contrast, C. elegans possesses two genescdc-48.1(C06A1.1) and cdc-48.2(C41C4.8) -encoding p97 with 88% identity over the entire protein (Yamanaka et al., 2004). We have previously reported that both homologues are essential in C. elegans and that their cellular functions are redundant (Yamanaka et al., 2004;Sasagawa et al., 2007b;Sasagawa et al., 2007c). Here, we report that the cdc-48.1 deletion decreases the brood size to half, which results from decreased sperm production. Epistasis experiments, proteinprotein interaction assays, and TRA-1A quantification revealed that CDC-48.1 interacts with CUL-2 complexes and is involved in the regulation of TRA-1A levels. We propose that C. elegans p97/CDC-48 has an important role in the switching processes from spermatogenesis to oogenesis.

Results
Reduced brood size in cdc-48.1 mutant worms We have previously reported that C. elegans possesses two p97 homologues, namely, CDC-48.1 and CDC-48.2, which are essential for embryogenesis, meiosis, ERAD and germline development (Yamanaka et al., 2004;Sasagawa et al., 2007b;Sasagawa et al., 2007c). The protein levels and expression patterns of these two p97 homologues are differentially regulated (Yamauchi et al., 2006). To elucidate the cellular functions of the two p97 homologues in more detail, we characterized the deletion mutants cdc-48.1(tm544) and cdc-48.2(tm659), which have 668 bp and 639 bp deletions, respectively (Sasagawa et al., 2007c). Although the cdc-48.1(tm544) mutant grew slightly slower than the wild type and the cdc-48.2(tm659) mutant showed only a small delay, both mutant worms grew to adult without embryonic or larval lethal phenotypes. Morphology, motility, defecation cycles and sexual behavior of hermaphrodites and males were not affected by the deletion mutations (data not shown).
Interestingly, however, we found that the cdc-48.1(tm544) mutant produced less progeny than the wild type, which produces approximately 300 progeny on average (Fig. 1).  worms were synchronized and incubated from early L4 stage. At 24, 36, 48 and 60 hour time points, P0 worms were transferred to new plates, and eggs laid from these worms during these periods were counted. The total number of fertilized eggs laid from cdc-48.1(tm544) worms was 47% of that from wild-type and cdc-48.2(tm659) worms (Fig. 1B). Importantly, during the first 24 hours, cdc-48.1(tm544) worms produced almost the same number of fertilized eggs as wild-type and cdc-48.2(tm659) worms; however, in the later periods they produced progressively fewer fertilized eggs and finally did not produce any eggs at all after 48 hours (Fig. 1B). By contrast, wild-type and cdc-48.2(tm659) worms continued to produce fertilized eggs up to 72 hours. It should be mentioned that germline and germ cells, which were dissected and stained with DAPI, of cdc-48.1(tm544) worms were indistinguishable from those of wild-type worms (see DAPI-stained images in Fig. 1D and Fig. 3B). Furthermore, the cdc-48.1(tm544) mutant laid a similar number of eggs as the wild type overall; however, half were non-fertilized. These results suggest that the cdc-48.1(tm544) mutation does not affect germline formation and oocyte production in hermaphrodites.
To demonstrate that the reduced brood size is caused by the depletion of CDC-48.1, we first prepared transgenic worms expressing a FLAG::CDC-48.1 fusion protein driven by the authentic cdc-48.1 promoter, and then transferred the cdc-48.1(tm544) mutation into the transgenic strain, yielding XA7203  1C). Furthermore, when cdc-48.2(RNAi) was applied to the XA7203 worms, the embryonic lethal phenotype caused by cdc-48.1(tm544);cdc-48.2(RNAi) was completely rescued by the FLAG::CDC-48.1 fusion protein. These results unambiguously demonstrate that the FLAG::CDC-48.1 fusion protein is fully functional and that the decreased brood size is caused by the cdc-48.1(tm544) mutation.
C. elegans hermaphrodites, in which sperm have been exhausted, can still produce progeny, if they are crossed with males, and sperm are supplied. To determine whether the decreased brood size in the cdc-48.1(tm544) mutant is due to an abnormality in oocytes or sperm, we performed mating experiments with normal males and mutant hermaphrodites. When cdc-48.1(tm544) mutant worms were co-cultivated with wild-type males, they produced a similar number of progeny as the wild type did, and almost all eggs laid were fertilized (wild type, 229.8±50.0, n=6; cdc-48.1(tm544), 210.8±41.6, n=6). As described above, in the cdc-48.1(tm544) mutant, an embryonic-lethal phenotype and developmental defects were not observed except for the slight reduction in growth rate. Taken together, these results clearly indicate that oocytes of the cdc-48.1(tm544) mutant do not have a functional defect.

cdc-48.1 mutant makes fewer sperm
Our results raised the possibility that the cdc-48.1(tm544) mutant produces fewer sperm or functionally defective sperm. C. elegans hermaphrodites produce sperm only in the early L4 stage and store them in the spermathecae to be used later to fertilize oocytes produced in the same germline after spermatogenesis is completed (Kimble and Crittenden, 2007). Therefore, to clarify whether the cdc-48.1(tm544) mutant produces fewer sperm or not, we observed the spermathecae of worms from the synchronized culture at the same intervals as in Fig. 1B. As shown in Fig. 1D, the cdc-48.1(tm544) mutant had a significant number of sperm until 24 hours from the early L4 stage, but sperm were almost exhausted by 48 hours. In addition, oocytes became stacked in the proximal gonad in cdc-48.1(tm544) worms at 48 hours ( Fig. 1D), which is a characteristic phenotype of spermless worms. By contrast, wildtype and cdc-48.2(tm659) mutants had sperm even at 48 hours. We then counted the number of sperm in DAPI-stained young adults that contained oocytes, but no embryos. Single arms of gonads from wild-type, cdc-48.1(tm544) and cdc-48.2(tm659) worms contained an average number of sperm of 151 (range, 120-199; n=7), 96 (75-113; n=14) and 148 (98-186; n=6), respectively. These results clearly indicate that the cdc-48.1(tm544) mutant produces fewer sperm. To further confirm this, we resolved total lysates of worms on 2D PAGE gels and identified the spot corresponding to the major sperm protein (MSP), a sperm-specific protein, by MALDI-TOF analysis. As clearly shown in Fig. 1E, adult worms of cdc-48.1(tm544) contained Results are mean ± s.d. of at least ten animals. (D) Deletion of CDC-48.1 causes early exhaustion of sperm in hermaphrodites. Typical images are shown (frequency: 10/10). Wild-type, cdc-48.1(tm544) and cdc-48.2(tm659) worms were fixed and stained with DAPI after incubation for 24 or 48 hours from early L4 stage (upper rows). Each germline containing spermathecae was dissected on a poly-L-lysine-coated glass slide and stained with DAPI (middle rows, DIC; lower rows, DAPI). White and yellow boxed areas represent hermaphrodite spermathecae containing sperm or no sperm, respectively. Scale bars: 50 μm. (E) Top panel shows 2D gel electrophoresis of the total protein extract of wild-type worm 48 hours after early L4 stage. Boxed regions of wild-type, cdc-48.1(tm544) and cdc-48.2(tm659) extracts are shown enlarged in the bottom panels. Arrowheads indicate major sperm protein (MSP). only a trace amount of MSP compared with wild-type and cdc-48.2(tm659) worms. This is fully consistent with the results that sperm in cdc-48.1(tm544) worms were exhausted earlier, as shown in Fig. 1D. These results clearly demonstrate that the cdc-48.1(tm544) mutant produces fewer sperm than wild-type worms.
We observed the expression pattern of FLAG::CDC-48.1 in the early L4 stage, during which sperm are produced. FLAG::CDC-48.1 was expressed from the bend to the proximal gonad ( Fig. 1A; Fig. 2A), although it was predominantly expressed at the pachytene stage and in spermatocytes (Fig. 2F). No FLAG signal was observed in control XA7202, which expressed no FLAG-tagged protein (Fig.  2B). The expression pattern of MSP, which is a spermatogenesis marker, was similar to that of FLAG::CDC-48.1 (Fig. 2C, compare with 2A). These results suggest that CDC-48.1 is required for the commitment to sperm production and are consistent with the abovementioned results that the cdc-48.1(tm544) mutant produces fewer sperm. However, MSP was also localized in sperm, which have no signal of phosphorylation of histone H3 at Ser10 (Fig. 2E), whereas FLAG::CDC-48.1 was not localized in sperm (Fig. 2D). These results also indicate that CDC-48.1 is not involved in sperm maintenance. It should also be noted that FLAG::CDC-48.1 was expressed in the distal tip cell in the mitotic region (Fig. 2G, asterisk).
Next, we determined whether cdc-48.1(tm544) mutant males have defects in sperm formation or not. Males of the cdc-48.1(tm544) mutant were generated at a similar frequency to those of the wild type and cdc-48.2(tm659) mutant. Gonad structure of mutant males stained with DAPI appeared to be unaffected by the mutation and a significant number of sperm were observed, which were indistinguishable from wild-type worms (data not shown), suggesting that the processes of sperm formation in males are not affected by the cdc-48.1(tm544) mutation. Furthermore, mutant males were able to mate normally with hermaphrodites and the mutation was indeed transferred, implying that the sexual behavior of the cdc-48.1(tm544) mutant and the motility and fertility of their sperm are not defective. Given that oocytes and sperm themselves are not functionally impaired by the cdc-48.1(tm544) mutation and that the sperm of cdc-48.1(tm544) worms are exhausted earlier than those of wild-type worms, it appears likely that the switch from spermatogenesis to oogenesis (the sperm-oocyte switch) in the germline of cdc-48.1(tm544) hermaphrodites takes place at an earlier stage than in the wild type.
To confirm this idea, we analyzed expression of RME-2, which is an oogenesis marker (Grant and Hirsh, 1999), for 12 hours after L4 entry. We counted the numbers of RME-2-positive gonads and MSP-positive gonads at each time point (Fig. 3A). Typical gonads at each time point are shown in Fig. 3B. At 0 hour from L4 entry, no worm expressed RME-2 in gonads, but all worms expressed MSP as expected. In the wild type at 7 hours from L4 entry, only 6.3% of gonads faintly expressed RME-2 (type I expression). By contrast, in cdc-48.1(tm544) at the same time, 27% of gonads faintly expressed RME-2 (type I) and 48.6% of gonads expressed RME-2 in masses (type II expression). At 12 hours from L4 entry, all worms expressed RME-2 in gonads. These results strongly indicate that the sperm-oocyte switch in cdc-48.1(tm544) hermaphrodites takes place at an earlier stage than in the wild type.

The sperm-oocyte switch
To investigate whether CDC-48.1 is involved in the known spermoocyte switching process (Fig. 4A) or in a novel process, we Journal of Cell Science 122 (20) , anti-FLAG antibody (green) and anti-histone H3 phospho-Ser10 antibody (red). (C, E) Fixed gonads stained with DAPI (blue), anti-MSP antibody (green) and anti-histone H3 phospho-Ser 10 antibody (red). Phosphorylation of histone H3 at Ser10 was used as an M-phase specific marker, and is detected in spermatocytes undergoing mitosis and mitotic nuclei (A-G), but not in sperm (D,E). MSP localizes to spermatogenic germ cells and sperm. FLAG::CDC-48.1 does not localize to sperm, which have no signal for phosphorylation of histone H3 at Ser10. Asterisks and arrowheads indicate distal tip cell and proximal gonad, respectively. Open arrowheads indicate sperm. Arrows indicate zygotene nuclei. Lines 'a' and 'b' show pachytene nuclei and spermatocytes, respectively. Scale bars: 50 μm.
performed genetic interaction analysis. FBF-1 (fem-3 mRNA binding factor) binds to the 3Ј-untranslated region of fem-3 mRNA and represses fem-3 translation (Zhang et al., 1997). The fbf-1(ok91) mutant reportedly produces more sperm than the wild type (Crittenden et al., 2002). We first confirmed that the fbf-1(ok91) mutant had a larger brood size as reported (Fig. 4B). When the cdc-48.1(tm544) mutation was introduced into the fbf-1(ok91) mutant, the larger brood size was clearly suppressed, suggesting that cdc-48.1 genetically interacts with fbf-1. Next, we investigated the genetic interaction between cdc-48.1(tm544) and fem-3. The fem-3(q20gf) hermaphrodites have been shown to produce only sperm at 25°C, but produce oocytes and excess sperm compared with the wild type at 15°C (Barton et al., 1987). The fem-3(q20gf) mutant was confirmed to have an extremely larger brood size at 15°C (Fig.  4C). As shown in Fig. 4C, the cdc-48.1(tm544) mutation again suppressed the larger brood size as a result of the fem-3(q20gf) mutation. These results suggest that CDC-48.1 has a role downstream or parallel to FEM-3 and affects the sperm-oocyte switch.
TRA-1A accumulation in cdc-48.1 mutant worms The zinc finger transcription factor TRA-1A is the terminal effector of the sex-determination pathway and is negatively regulated by FEM proteins (Chen and Ellis, 2000;Conradt and Horvitz, 1999;Schvarzstein and Spence, 2006;Yi et al., 2000;Zarkower and Hodgkin, 1992;Zarkower and Hodgkin, 1993). Fig. 3. The sperm-oocyte switch in cdc-48.1(tm544) hermaphrodites takes place at an earlier stage than in the wild type. Worms were first synchronized at the L3 stage, and incubated for several hours until they grew up to the L4 stage. Worms were analyzed from L4 entry for 12 hours. Gonads were dissected, fixed and stained with DAPI (blue), anti-MSP antibody (green) and anti-RME-2 antibody (red) at 0, 7 and 12 hours. RME-2 was used as an oogenesis marker (Grant and Hirsh, 1999). The numbers of RME-2-positive gonads and MSP-positive gonads were counted at each time point (A). Typical gonads at each time point are shown in (B). At 0 hour from L4 entry, no worm expressed RME-2 in its gonads. In the wild type at 7 hours from L4 entry, only 6.3% of gonads faintly expressed RME-2 (type I expression). In cdc-48.1(tm544) at 7 hours from L4 stage, 27% of gonads faintly expressed RME-2 (type I), and 48.6% of gonads highly expressed RME-2 (type II expression). At 12 hours from L4 entry, all worms expressed RME-2 in their gonads. Arrowheads indicate RME-2 signal. Scale bars: 50 μm.  (Ellis and Schedl, 2006). Arrow indicates activation and bars indicate inhibition. An indication of the level of expression is given for sperm and oocytes. (B,C) N2 wild-type and mutant worms were synchronized and incubated from early L4 stage. F1 eggs laid from these worms were pooled and counted. Experiments for fbf-1(ok91) (B) and fem-3(q20gf) (C) were carried out at 20°C and 15°C, respectively. Results are mean ± s.d. from at least five animals.
Recently, it has been shown that TRA-1A is regulated by degradation mediated by a CUL-2-based E3 ubiquitin ligase complex containing FEM-1, FEM-2 and FEM-3 as a substraterecognition receptor and that TRA-1A accumulates in the feminized mutants, fem-1, fem-2 and fem-3, compared with the wild type (Starostina et al., 2007). Given that cdc-48.1 genetically interacts with fem-3 as described above, TRA-1A is expected to accumulate in the cdc-48.1(tm544) mutant. To test this idea, we first prepared an anti-TRA-1A antibody. This antibody recognized a 118 kDa protein, which was markedly decreased in the tra-1A(RNAi) worms (Fig. 5A), suggesting that the 118 kDa protein is TRA-1A. TRA-1A was increased 2.9-fold in the fem-3 loss-offunction mutant worms (Fig. 5B), which is consistent with a previous report (Starostina et al., 2007). By quantifying TRA-1A, we found that the TRA-1A level in the cdc-48.1(tm544) mutant was 1.5-fold that of the wild-type (Fig. 5C), suggesting that p97 regulates TRA-1A degradation.
It is known that p97 and the Cul2 complex are well conserved in eukaryotes Ye, 2006;Schuberth and Buchberger, 2008). It is also interesting to mention that human FEM1B, a homologue of C. elegans FEM-1, has been shown to be a component of the Cul2 complex, although its function has not Journal of Cell Science 122 (20) Anti-actin antibody was used as a loading control. White arrowhead indicates actin. (B) Extracts of N2 wild-type and fem-3(e2006lf) young adult worms incubated at 25°C were subjected to western blotting. TRA-1A and actin were quantified. TRA-1A accumulation in fem-3(e2006lf) was 2.9-fold greater than that of the wild type. (C) Extracts of N2 wild-type, cdc-48.1(tm544) and cdc-48.2(tm659) young adult worms incubated at 25°C were subjected to western blotting. TRA-1A and actin were quantified. TRA-1A accumulation in cdc-48.1(tm544) was 1.5 times greater than in the wild-type. Fig. 6. Interaction between NPL-4.1 and Elongin C. (A) Yeast two-hybrid assay with BD-NPL-4.1 and AD-ELC-1. Mutant ECL-1 (mECL-1) carries L47D-L49D-Y88D-Y91D mutations. Transformants were tested for growth on SD-LEU/TRP and SD-LEU/TRP/HIS/ADE plates at 30°C for 3 days. (B) GST-NPL-4.1 and FLAG-ELC-1 were prepared and mixtures were incubated and pulled down with glutathione-Sepharose, and analyzed with anti-GST and anti-FLAG antibodies. (C) Yeast two-hybrid assay with human Npl4-BD and human Elongin-C-AD. Transformants were tested for growth on SD-LEU/TRP and SD-LEU/TRP/HIS/ADE plates at 30°C for 3 days. p97 regulates sperm-oocyte switch yet been clarified . To elucidate whether the interaction between the p97-Ufd1-Npl4 complex and Cullin 2-type E3 ubiquitin ligase is specific for C. elegans or common in eukaryotes, we carried out a yeast two-hybrid assay using human Npl4 and Elongin C. As shown in Fig. 6C, Npl4 clearly interacted with Elongin C, suggesting that the interaction between p97-Ufd1-Npl4 and the Cul2 complex through Npl4 binding to Elongin C is well conserved in eukaryotes.

Proteolytic activity is probably involved in determination of brood size
It has been reported that proteasomal ubiquitin receptor RPN-10 controls sex determination in C. elegans and that double knockdown of rpn-10 and ufd-2 overcomes the germlinemusculinizing effect of fem-3(gf) (Shimada et al., 2006). As mentioned above, we found that the cdc-48.1(tm544) mutation overcame the sperm-overproducing phenotype of fem-3(gf) (Fig.  4C). Therefore, we investigated the effect of rpn-10 mutation on brood size. Approximately 60% of rpn-10(tm1180) worms showed a sterile phenotype and the remaining 40% had a markedly decreased brood size (Fig. 7A), which is consistent with a previous report (Shimada et al., 2006). Given that rpn-10 encodes a subunit of the 26S proteasome, proteolytic activity is probably involved in the process of determining brood size. When the rpn-10(tm1180) mutation was introduced into the cdc-48.1(tm544) and cdc-48.2(tm659) mutants, almost all worms became sterile (Fig.  7A). As shown in Fig. 7B, although development and formation of the germline of these double mutant worms appear to be normal (compare Fig. 7B, panels h and n, with panel a), there were no sperm in the spermathecae, no embryos in the uterus, and stacked oocytes, which is a characteristic phenotype of spermless worms (Fig. 7B). However, when these mutant worms were crossed with wild-type males, they then produced progeny (Fig. 7B, compare panels m and s with l and r). Furthermore, the gonad structure of double mutant [cdc-48.2(tm659);rpn-10(tm1180)] males stained with DAPI appeared to be unaffected and a significant number of sperm were observed, making them indistinguishable from wildtype worms (Fig. 7B, compare panels and g). These results indicate  1(tm544) and cdc-48.2(tm659) mutants. The total number of progeny was counted in at least 15 worms of each genotype. Means ± s.d are shown. Sterility of the F1 generation was also counted. cdc-48.1(tm544);rpn-10(tm1180) and cdc-48.2(tm659);rpn-10(tm1180) double deletion mutants were maintained as heterozygotes and segregating double homozygotes were used. (B) Germline feminization phenotype was observed in cdc-48.1(tm544);rpn-10(tm1180) and cdc-48.2(tm659);rpn-10(tm1180) double mutants. a-g, N2 wild-type; h-m, cdc-48.1(tm544);rpn-10(tm1180); and n-t, cdc-48.2(tm659);rpn-10(tm1180). Chromosomal DNA was stained with DAPI after whole-mount fixation (a,b,g,h,i,n,o,t). Germline development was normal in cdc-48.1(tm544);rpn-10(tm1180) (h) and cdc-48.2(tm659);rpn-10(tm1180) (n) double mutants, similarly to the wild type (a). Arrowheads indicate normally elongated gonad arms. The part of the gonad containing the spermathecae shown in a, h and n was enlarged and is shown in b, i and o, respectively. Each germline containing a spermatheca was dissected on a poly-L-lysine-coated slide and stained with DAPI (c,j,p; d,k,q; DAPI). White and yellow boxed regions represent hermaphrodite spermathecae containing sperm and no sperm, respectively. Worms were observed without fixation before (e,l,r) and after (f,m,s) mating. Arrows and asterisks indicate vulva and stacked oocytes, respectively. Males of wild type (g) and cdc-48.2(tm659);rpn-10(tm1180) (t) were fixed and stained with DAPI. Boxed regions indicate spermathecae. Scale bars: 50 μm. that oocyte development was not impaired, but sperm were not produced by the double mutant hermaphrodites of cdc-48.1(tm544);rpn-10(tm1180) and cdc-48.2(tm659);rpn-10(tm1180). This implies that these double mutants show a gonad feminization phenotype. Taking these results into account, TRA-1A is proteolytically regulated in concert with CDC-48-UFD-1-NPL-4.1 and the CUL-2-based ubiquitin ligase complex and controls the sperm-oocyte switch.

Discussion
In the present study, we found that the brood size was decreased in p97/CDC-48 mutants (Fig. 1). The number of sperm produced was also decreased; sperm were exhausted earlier than in the wild type; no defect was observed in oocyte production; and the spermoverproducing phenotype of fbf-1(ok91) and fem-3(q20gf) mutants was suppressed by the cdc-48.1(tm544) mutation (Figs 1 and 4). These results suggest that the decreased brood size resulted from decreased sperm production. Furthermore, the RME-2 signal in the cdc-48.1(tm544) mutant appeared earlier than in the wild type (Fig. 3). Therefore, we conclude that p97/CDC-48 is involved in the spermoocyte switching process. It is noteworthy to mention that the onset of spermatogenesis is governed by the post-transcriptional regulation of tra-2, whereas the sperm-oocyte switch is regulated by the posttranscriptional control of fem-3 (Ahringer and Kimble, 1991;Puoti et al., 2001;Shimada et al., 2006). Determination of the precise targeting process of p97/CDC-48 remains to be addressed.
We also found that CDC-48-UFD-1-NPL-4.1 interacts with the CUL-2 complex through the interaction between NPL-4.1 and ELC-1/Elongin-C (Fig. 6). It has been shown that TRA-1A is regulated by degradation mediated by a CUL-2-based ubiquitin ligase complex containing FEM-1, FEM-2 and FEM-3 as a substraterecognition receptor (Starostina et al., 2007). Here, we indeed demonstrated that TRA-1A accumulates in the cdc-48.1(tm544) mutant (Fig. 5). Taking these results together, we propose that p97, an AAA chaperone, with UFD-1 and NPL-4 has an important role in regulating the TRA-1A level, in combination with the CUL-2 complex containing FEM-1, FEM-2 and FEM-3, and is required for the appropriate operation of the sperm-oocyte switch (Fig. 8A).
AAA proteins including p97 are thought to be an unfoldase or a segregase (Patel and Latterich, 1998). Furthermore, it has been suggested that p97 is a ubiquitin-selective chaperone and that its key function is to disassemble protein complexes (Rape et al., 2001). Thus, p97 probably functions as a segregase to disassemble polyubiquitylated proteins from the CUL-2 complex in an ATPdependent manner and delivers them to the proteasome containing RPN-10 in combination with shuttle factors such as Rad23 and Dsk2 (Richly et al., 2005) (Fig. 8B). It should be emphasized that the interaction of p97 and E3 ubiquitin ligases is not only required for the degradation of ERAD substrates (Zhong et al., 2004;Lilley and Ploegh, 2005;Ye et al., 2005), but also soluble cytoplasmic substrates. Therefore, in the p97 mutants, more TRA-1A, the master sexual regulator, accumulated than in the wild type, which in turn caused the sperm-oocyte switch to occur earlier than in the wild type. It is interesting to mention that the defect in the brood size of cdc-48.1(tm544) is severer than that of cdc-48.2(tm659). We have previously reported that the amount of CDC-48.1 is double that of CDC-48.2 (Yamauchi et al., 2006). These results appear to support the above-mentioned notion that the chaperone activity of p97 controls the efficiency of CUL-2-mediated TRA-1A degradation by the proteasome. It should also be mentioned that the interaction between p97 and the Cul2 complex is not restricted to C. elegans, but might be conserved in eukaryotes (Fig. 6C). Therefore, Cul2mediated degradation of substrate proteins generally occurs with the aid of p97 chaperone activity.
It has been reported that in C. elegans, CUL-2 is required for several key processes in cell division and embryonic development, including Journal of Cell Science 122 (20)  -48.1(tm544) mutant and thus switching from spermatogenesis to oogenesis occurs earlier than in the wild type. Proteins that promote spermatogenesis are blue, and those that promote oogenesis are red. (B) Proposed model of p97 requirement for disassembly of the ubiquitylated substrate from the CUL-2 ubiquitin ligase complex. p97 might disassemble the substrate by changing its conformation using ATP-hydrolyzing energy. p97 regulates sperm-oocyte switch cyclin-dependent kinase (CDK) inhibitor CKI-1 degradation, meiotic progression, PAR-6 degradation and the establishment of anteriorposterior polarity, cyclin B1 degradation, chromatin condensation, mitotic progression, proper cytoplasmic organization and the degradation of CCCH-finger polarity proteins such as PIE-1 (Feng et al., 1999;DeRenzo et al., 2003;Liu et al., 2004;Sonneville and Gonczy, 2004;Sasagawa et al., 2005;Sasagawa et al., 2007a;Pacquelet et al., 2008). Importantly, some of the phenotypes observed in the cul-2 mutant, including defects in meiotic progression, chromosome condensation, cyclin B1 degradation, and establishment of anterior-posterior polarity, have also been observed in p97depleted worms (Sasagawa et al., 2007b) (and our unpublished results). Therefore, it can be reasonably assumed that p97 is globally involved in CUL-2-dependent protein degradation. It is important to identify substrates and substrate recognition receptors to decipher the detailed cellular functions of p97. Moreover, although the levels of TRA-2 were reported to be proteolytically regulated in an RPN-10dependent manner (Shimada et al., 2006), E3 ubiquitin ligase and the substrate recognition receptor have not yet been clarified. In addition, SEL-10 is a WD40-repeat-containing F-box protein and might serve as a substrate-recognition component that targets substrate proteins including FEM-1 and FEM-3 for ubiquitin-mediated proteolysis by the proteasome (Jager et al., 2004). It will thus also be interesting to investigate whether p97 is involved in processes performed by other E3 ubiquitin ligases, such as CUL-1 and the anaphase-promoting complex/cyclosome (APC/C), in the future.

Construction of transgenic line expressing FLAG-tagged CDC-48.1
We constructed the plasmid to express FLAG::CDC-48.1 and wild-type UNC-119. First, the wild-type unc-119 gene was cloned into pHSG298 (TaKaRa), yielding pHUNC1. The wild-type UNC-119 was used to rescue the unc-119(ed3) mutation. The cdc-48.1 gene with a 1 kb 5Ј-upstream region and a 1 kb 3Ј-untranslated region was PCR amplified from the wild-type genomic DNA, and then cloned into pHUNC1. Finally, the DNA sequences encoding the FLAG tag were inserted immediately upstream of the translation initiation codon of cdc-48.1 to produce an N-terminal fusion protein. The plasmid was bombarded into unc-119(ed3) mutant worms using the BioRad Biolistic PDS-1000/He particle delivery system, as described previously (Praitis et al., 2001). Unc-rescued worms were obtained and FLAG::CDC-48.1-expressing transgenic lines were then screened by western blotting with anti-FLAG antibodies (Sigma, clone M2). The cdc-48.1(tm544) mutation was transferred by mating. In this way, the following strains were prepared: XA7200 (unc-119(ed3) qaIs7200 [unc-119(+)

RNAi
To construct RNAi feeding bacteria, the cDNA fragment for tra-1A (nucleotides 835-2133) was amplified using an RT-PCR kit (Qiagen) with total RNA from N2 adult hermaphrodites as a template. The following primer sets were employed: tra-1Fw+BamHI 5Ј-TAC TGA GGA TCC CGA TAC GGT TGT CGA GGC TCA ACG-3Ј (forward) and tra-1Rv+XhoI 5Ј-TAT CTC GAG TCA AAT CTC ATT ATC ACC ACC AAA ACC AAC CTC ACA C-3Ј (reverse). Note that the amplified fragment is specific for tra-1A but not for tra-1B. The cDNA fragment was cloned into the plasmid LITMUS 28 (New England BioLabs). The dsRNA was prepared in vitro using T7 RNA polymerase, and tra-1A dsRNA (1 mg/ml) was microinjected. The RNAi construct for cdc-48.2 was described previously (Sasagawa et al., 2007b). Plasmids were transformed into Escherichia coli strain HT115 (DE3). We used Optimal Protocol I for feeding RNAi methods, as described previously (Kamath et al., 2001).

Western blotting analysis
TRA-1A (residues 278-1110) protein was expressed in E. coli, purified and used to immunize rabbits to obtain the anti-TRA-1A antibody. Total lysates of 100 young adult worms were resolved on 5-20% SDS-PAGE and transferred to a nitrocellulose membrane. Signals were detected with the antibodies against TRA-1A and actin (mouse anti-actin monoclonal antibody (C4), Millipore, 1:5000 dilution), and quantified using ImageJ (National Institutes of Health, Bethesda, MD). The amount of actin was detected as a loading control.

Pull-down assay
GST-NPL-4.1 and FLAG-ELC-1 were prepared using the TNT SP6 High-Yield Protein Expression System (Promega). Mixtures of GST-NPL-4.1 and FLAG-ELC-1 were incubated and pulled down with glutathione-Sepharose (GE Healthcare Bioscience), and then subjected to western blotting with anti-GST (GE Healthcare Bioscience) and anti-FLAG (Sigma) antibodies.

Two-dimensional gel electrophoresis and protein identification by MALDI-TOF MS
Two-dimensional (2D) gel electrophoresis of total protein extracts of C. elegans and the subsequent identification of proteins by MALDI-TOF MS were performed as described previously (Yamauchi et al., 2008).
We thank Yuko Saito-Sasagawa for constructing FLAG::CDC-48.1 expressing transgenic worms, Shohei Mitani (Tokyo Women's Medical University) and the Caenorhabditis Genetic Center for C. elegans mutants, Yuji Kohara (National Institute of Genetics) for C. elegans cDNA clones, Barth Grant (Rutgers University) for antibodies against