|Home | About | Journals | Submit | Contact Us | Français|
Caenorhabditis elegans and C. briggsae are both self-fertile hermaphroditic nematodes that evolved independently from male/female ancestors. In C. elegans, FEM-1, FEM-2, and FEM-3 specify male fates by promoting proteolysis of the male-repressing transcription factor, TRA-1. Phenotypes of tra-1 and fem mutants are consistent with this simple linear model in the soma, but not in the germ line. While both XX and XO tra-1(lf) mutants have functional male somas, they produce both sperm and oocytes. Further, all three tra-1; fem double mutants retain the expected male soma, but make only oocytes (the germline fem phenotype). Thus, a poorly characterized tra-1 activity is important for sustained male spermatogenesis, and the fem genes affect germline sexual fate independently of their role in regulating TRA-1. C. briggsae tra-1 mutants are phenotypically identical to their C. elegans counterparts, while the fem mutants differ in the germline: XX and XO C. elegans fem mutants are true females, but in C. briggsae they are self-fertile hermaphrodites. To further explore how C. briggsae hermaphrodites regulate germline sex, we analyzed Cb-tra-1/Cb-fem interactions. Cb-tra-1 is fully epistatic to Cb-fem-2 in the germ line, unlike the orthologous C. elegans combination. In contrast, Cb-fem-3 shifts the Cb-tra-1(lf) germline phenotype to that of a nearly normal hermaphrodite in the context of a male somatic gonad. This suggests that Cb-fem-3 is epistatic to Cb-tra-1(lf) (as in C. elegans), and that the normal control of C. briggsae XX spermatogenesis targets Cb-tra-1-independent factors downstream of Cb-fem-3. The effect of Cb-fem-3(lf) on Cb-tra-1(lf) is not mediated by change in the expression of Cb-fog-3, a likely direct germline target of Cb-tra-1. As Cb-fem-2 and Cb-fem-3 have identical single mutant phenotypes, Cb-tra-1 provides a sensitized background that reveals differences in how they promote male germline development. These results represent another way in which C. briggsae germline sex determination is incongruent with that of the outwardly similar C. elegans.
The nematode genus Caenorhabditis is emerging as a powerful system for exploring the intersection of development, genetics, and evolution (Haag and Pilgrim 2005; Kammenga et al. 2008). The most striking variable among Caenorhabditis is reproductive mode. While C. elegans and C. briggsae produce males and hermaphrodites, the remaining described species produce males and true females. Hermaphrodites are anatomically female, but produce a limited amount of amoeboid sperm (200-300) during the final larval stage (in C. elegans) or in early adulthood (C. briggsae) before switching to oocyte production for their remaining life span. Phylogenetic studies suggest hermaphroditism has evolved from male/female species numerous times in rhabditid nematodes (Cho et al. 2004; Kiontke et al. 2004; Kiontke and Sudhaus 2006), and recent developmental genetic comparisons have further supported the convergence scenario for C. elegans and C. briggsae (Hill et al. 2006; Nayak et al. 2005).
The primary determinant of sex in C. elegans is the X:A ratio (Madl and Herman 1979; Nigon 1951). A high ratio (XX) represses xol-1 transcription, resulting in hermaphrodite development. A low ratio (XO) allows high xol-1 expression, resulting in male development (Rhind et al. 1995). xol-1 activity controls sexual fate via a negative regulatory pathway (Figure 1, top panel), ultimately resulting in the activation (in hermaphrodites) or repression (in males) of the terminal global transcriptional regulator, TRA-1. At the center of this pathway is a signal transduction cascade consisting of HER-1 (a secreted signaling protein; Perry et al. 1993), TRA-2 (a transmembrane receptor for HER-1; Hamaoka et al. 2004; Kuwabara and Kimble 1995), and TRA-3 (a protease required for TRA-2 function; Barnes and Hodgkin 1996; Sokol and Kuwabara 2000), as well as the three cytoplasmic FEM proteins (Zarkower 2006). The three tra genes promote female fate; her-1 and the fem genes promote male fate. The above pathway is used for both somatic and germline sex determination. However, in hermaphrodites additional genes are required for the modulation of activity of core pathway genes to produce both sperm and oocytes (Figure 1, middle panel). In C. elegans, tra-2 and fem-3 have been most strongly implicated as the targets of this modulation (Kimble and Crittenden 2007). Consistent with this, mutations in any of the three fem genes converts XX hermaphrodites into true females (Hodgkin 1986).
TRA-1 is a Gli/Ci family zinc-finger transcription factor (Zarkower and Hodgkin 1992). The tra-1 gene encodes two transcripts, one producing the protein TRA-1A (1109 amino acids) and another the smaller TRA-1B (287 amino acids). TRA-1A contains five C2H2 zinc fingers and binds DNA, while TRA-1B has only the first two zinc fingers, does not bind DNA, and has no known function (Zarkower and Hodgkin 1992; Zarkower and Hodgkin 1993). Most TRA-1A is proteolytically processed in hermaphrodites to a truncated form known as TRA-1100, which accumulates to high levels and is presumably responsible for tra-1’s male-repressing activity (Schvarzstein and Spence 2006). Starostina et al. (2007) have shown that TRA-1100 levels are directly regulated by the FEM proteins through their participation in a CUL-2-based ubiquitin ligase complex. FEM-1 (an ankyrin-repeat protein; Spence et al. 1990) is the substrate-recognition subunit, and FEM-2 (a protein phosphatase; Chin-Sang and Spence 1996; Pilgrim et al. 1995) and FEM-3 (a novel, rapidly evolving protein; Haag et al. 2002; Rosenquist and Kimble 1988) act as cofactors. The activity of the FEM complex is inhibited by direct interaction between FEM-3 and cytoplasmic C-terminus of TRA-2 (Mehra et al. 1999), and this inhibition is in turn thought to be abrogated in males by binding of HER-1 to TRA-2 (Hamaoka et al. 2004; Hunter and Wood 1992).
TRA-1 is known to negatively regulate four male-specific genes, egl-1, mab-3, dmd-3, and fog-3 (Chen and Ellis 2000; Conradt and Horvitz 1999; Mason et al. 2008; Yi et al. 2000), suggesting repression of male fates is crucial to the control of sex determination. Consistent with this, strong loss-of-function alleles of tra-1 convert XX hermaphrodites into well formed males that are capable of mating and siring progeny (Hodgkin 1987; Hodgkin and Brenner 1977). However, both XX and XO tra-1 males cannot consistently sustain spermatogenesis, and often make oocytes of variable quality in a male somatic gonad (Hodgkin 1987; Schedl et al. 1989). This suggests that a poorly characterized activity of tra-1 is necessary for the promotion of male fates in the germ line. In addition, fem; tra-1 double mutants display a male soma (like tra-1 alone), but a completely feminized germline (the phenotype of the fem genes; Hodgkin 1986). This implies that a tra-1-independent role exists for the fem genes in the control of germline sex (Ellis and Schedl 2007).
C. briggsae tra-1 mutants have both the complete XX somatic masculinization and incomplete XO germline feminization of their C. elegans counterparts (Kelleher et al. 2008), indicating that the potentially complex regulation of tra-1 is conserved. In contrast, XX C. briggsae fem mutants are normal hermaphrodites, and XO Cb-fem-2 and Cb-fem-3 mutants are completely sex-reversed to hermaphrodites, rather than to females as in C. elegans (Hill et al. 2006). The modulations in germline sex determination used by C. briggsae hermaphrodites are thus distinct from those of C. elegans, and are likely to act downstream of the Cb-fem genes (Figure 1, bottom panel). One goal in this study was to assess whether this different regulation of hermaphrodite spermatogenesis seen in C. briggsae would be reflected in the genetic interaction between Cb-tra-1 and Cb-fems. We also sought further evidence for the existence of a tra-1-independent role of the fem genes that may be a general feature in Caenorhabditis.
Cb-tra-1(nm2) was the allele employed to generate Cb-tra-1; Cb-fem double mutants for this analysis. This is a strong loss-of-function allele caused by a nonsense mutation at codon 512. This eliminates approximately 50% of the full-length TRA-1A protein, but retains the DNA-binding zinc finger domain. The mutants exhibit robust mating behavior, and young XX nm2 males can sire cross-progeny, although at much lower levels than wild-type males (Kelleher et al. 2008). The comparable C. elegans TRA-1 mutation, e1781, has a nonsense mutation in the same region (Zarkower and Hodgkin 1992) and also has a fully transformed soma with an intersexual germ line (Hodgkin 1987). However, a true null allele of Ce-tra-1, e1099, has additional defects in somatic gonad development (Hodgkin 1987; Mathies et al. 2004), suggesting that tra-1(e1781) and Cb-tra-1(nm2) may both retain residual function in gonad development. We also present some further analyses of germline phenotypes of the Cb-tra-1(nm30) allele, a mutation in the 5′splice site of Intron 2 characterized by an incomplete male tail and well formed oocytes within a male somatic gonad (Kelleher et al 2008).
An obvious potential complication to analysis of Cb-tra-1; Cb-fem double mutants (and in contrast to C. elegans) is the lack of distinct phenotypes between the single mutants. The single Cb-fem mutants have a normal hermaphroditic germline, producing both sperm and oocytes in a double-armed gonad (Hill et al. 2006). Cb-tra-1 (nm2) single mutants have an intersexual germline that also produces both sperm and oocytes, albeit in the context of a male somatic gonad (Kelleher et al. 2008). It was therefore crucial to define germ line phenotypes quantitatively. Because the three fem mutants from a given species produce identical germline phenotypes, we predicted that any effects Cb-fem-2 and Cb-fem-3 had on the Cb-tra-1 phenotype would be the same. Surprisingly, they are not.
Mutant alleles used in this study include C. elegans tra-1(e1099) III and tra-1(e1781) III, and C. briggsae Cb-tra-1(nm2) III, Cb-tra-1(nm30) III, Cb-let(nm28) III, Cb-fem-2(nm27) III, and Cb-fem-3(nm63) IV. All genetic experiments were conducted at 20°C using standard C. elegans media (Wood 1988), with the exception of increasing the agar for NGM plates to 2.2% to discourage burrowing.
To produce Cb-fem; Cb-tra-1 (nm2) double mutants, the first step was to cross Cb-tra-1(nm2)/Cb-let(nm28) hermaphrodites with AF16 (wild-type) hermaphrodites. Males from this cross, half of which carry the nm2 allele, were mated to homozygous Cb-fem-2 or Cb-fem-3 mothers. Hermaphrodite progeny were singled as virgins, allowed to lay most of their progeny, and then genotyped to confirm Cb-fem heterozygosity as described in Hill et al. (2006). Mothers that were also Cbr-tra-1(nm2)/+ produced approximately one-quarter Tra pseudomale progeny and were retained. Hermaphrodite offspring from plates founded by double heterozygous mothers were again singled, allowed to lay progeny, and then genotyped by PCR to find animals that were both homozygous for the Cb-fem deletion allele and segregating Tra animals Cb-tra-1(nm2)/+. For Cb-fem-2 this required a recombination event, but as it is roughly 30 cM from Cb-tra-1 this was not an impediment. The resulting strains, Cb-fem-2(nm27); Cb-tra-1(nm2)/+ and Cb-fem-3(nm63); Cb-tra-1(nm2)/+, were maintained during the experiments by occasional sib selection.
For differential interference contrast (DIC) microscopy, worms were immobilized with 50mM sodium azide in M9 salts (Wood 1988) and mounted on 2% agar pads. For Hoechst 33258 staining, worms were rinsed three times with M9 salts with centrifugation for three minutes at 3400 rpm between each rinse. 400 μl of -20°C methanol was added to the worms, which were then incubated at -20°C for a minimum of 10 min. After 3 more rinses in M9, worms were incubated for 45 min. at room temperature in the dark with 200 μl of 7.5 μM Hoechst 33258 in M9. Following 3 further rinses in M9 salts, 30 μl of worm suspension was mixed with 10 μl of vectashield (Vector Laboratories) and mounted onto 3-4 agar pads (2%) for fluorescence microscopy. All images were captured with a Zeiss Axiocam digital camera and Open Lab software (Improvision). Phenotypic categories were based on those used for C. elegans tra-1 by Schedl et al (1989).
Worms of the appropriate genotype and within their first day of adulthood were picked in groups of 5 into 10 μl drops of nuclease-free water in the caps of 0.6 ml microcentrifuge tubes and then spun into the bottom of the tubes after gently capping them. This was repeated twice for hermaphrodite worms or thrice for males and Tra animals (i.e. 10 or 15 animals per tube). Worms were then mixed with 150 μl Tri Reagent (Molecular Research Center) and frozen at −80° C for at least overnight. After thawing, the worms were lysed with four cycles of grinding and pelleting with disposable plastic pestles. Total RNA was isolated from lysates according to manufacturer’s instructions, with the addition of 2 μl of Polyacryl carrier (Molecular Research Center). RNA pellets were resuspended in the water and oligo-dT primer components of the Transcriptor cDNA synthesis kit (Roche) and reverse-transcribed according to manufacturer’s instructions.
2μl of each cDNA preparation was used as template for quantitative PCR using the Light Cycler 480 SYBR Green reagent (Roche) and the Roche Light Cycler 480 machine. A 232 nt Cb-fog-3 cDNA fragment was amplified with the oligonucleotide primers EH36 (5′ GGATGTTGGCTTGAACGTGAAC 3′) and EH39 (5′ ATACTGATTCACACTATCAGCC 3′). EH36 is wholly contained within Exon 7, while EH39 anneals to the Exon 4/5 junction. For the actin internal standard, a 222 nt actin fragment was amplified using the primers EH37 (5′ TACCTCATGAAGATCCTCACCG 3′) and EH38 (5′ CATACCCAAGAAGGATGGCTGG 3′). Because little is known about sex-specific regulation of actin family members in C. briggsae, EH37 and EH38 were designed using an alignment of all C. briggsae actin gene predictions in WormBase (http://www.wormbase.org) so as to anneal to essentially all family members without amplicon length variation (ESH, unpublished data). Threshold cycle number was determined using the second derivative maximum method. Amplification efficiencies for actin and Cb-fog-3 were determined directly from multiple amplification curves (Ramakers et al. 2003) to be 0.57-0.58. Threshold cycle numbers and efficiencies were then used to produce normalized linear expression values (Livak and Schmittgen 2001).
To establish the range of phenotypes, we first scored over 200 Cbr-tra-1(nm2) mutant animals by DIC microscopy. As noted by Kelleher et al. (2008), most XX pseudomales produced only sperm into the first day of adulthood, while older animals switched to an abnormal oogenesis. However, the age at which oogenesis begins varies from as early as the L4 (last larval) stage (Figure 2A) to not at all (not shown). In addition to oocyte production, some mutants exhibited poorly organized germ lines, often consisting of a granular, apparently acellular cytoplasm. This material was found most often between regions of sperm and oocytes (Figure 2B), although in some older adults it extended throughout most of the proximal gonad arm.
Staining of mutants with Hoechst 33258 revealed that both well-defined oocytes and the apparently acellular granular material often harbored swollen, endoduplicated nuclei (Figure 2C, D). This endomitotic oocyte (Emo) phenotype has been shown in C. elegans to be due to inappropriate activation of oocyte mitotic cell cycle in the absence of ovulation (Greenstein 2005). A second Cb-tra-1 allele, nm30, is also frequently Emo (Table 1). While oocytes have long been recognized in the gonads of C. elegans tra-1 mutants (Hodgkin 1987; Schedl et al. 1989), they have not been reported to be Emo. We stained the null allele of C. elegans tra-1, e1099, as well an nm2-like allele, e1781, and found that in both cases they can also be Emo (Figure 2E, F; Table 1). e1781 animals are frequently Emo, but e1099 mutants only rarely so, presumably because the latter only rarely produces oocytes to begin with (Hodgkin 1987; Schedl et al. 1989).
A major goal of this study was to determine whether loss of Cb-fem-2 or Cb-fem-3 could feminize (or otherwise modify) the variably intersexual germline phenotype of Cb-tra-1. Due to the similarities between Cb-tra-1 and Cb-fem-2/3 mutants, we scored many animals to produce phenotypic distributions that could potentially reveal quantitative shifts. Because of its ability to reveal small numbers of sperm and endomitotic oocytes that are difficult to see by transmitted light microscopy, we used Hoechst 33258 staining for this purpose. That this method is generally accurate is supported by the similarity of our analysis of two C. elegans tra-1 alleles to that of Schedl et al. (1989; Figure 3).
By both qualitative and quantitative criteria, Cb-tra-1(nm2); Cb-fem-2(nm27) mutants were indistinguishable from Cb-tra-1(nm2) alone. They manifested both the temporally variable switch to oogenesis at a similar frequency (Figure 3) and produced endomitotic oocytes (Figure 4A, B). In contrast, Cb-tra-1(nm2); Cb-fem-3(nm63) mutants differed from both Cb-tra-1(nm2) and Cb-tra-1(nm2); Cb-fem-2(nm27) both qualitatively and quantitatively. First, oocytes were never made by L4 Cb-tra-1(nm2); Cb-fem-3(nm63) mutants (not shown), yet appeared in over 90% of young adults (Figure 2). In contrast, over half of both Cb-tra-1(nm2) and Cb-tra-1(nm2); Cb-fem-2(nm27) young adults made only sperm (Figure 3). When oocytes were made in Cb-tra-1(nm2); Cb-fem-3(nm63) animals, they were noticeably more robust than in Cb-tra-1(nm2) (Figure 4C), and frequently became endomitotic (Figure 4E, F).
In addition to the differences in gamete development described above, 8% of Cb-tra-1(nm2); Cb-fem-3(nm63) animals scored with DIC microscopy had unexpected somatic gonad abnormalities, in many cases including the presence of both a robust, testis-like posterior arm and a smaller anterior arm (Figure 4D). These abnormalities were less apparent in Hoechst 33258-stained specimens (Figure 3), but were not seen in Cb-tra-1(nm2) or Cb-tra-1(nm2); Cb-fem-2(nm27) animals even when scored by DIC microscopy (data not shown).
The only known direct target of TRA-1 in the germ line in C. elegans is fog-3 (Chen and Ellis 2000). As fog-3 is conserved in C. briggsae (Chen et al. 2001), we sought to test whether the feminization of the Cb-tra-1(nm2) germ line by loss of Cb-fem-3 activity might be due to reduced fog-3 expression. As obtaining large numbers of double homozygotes is not feasible, we turned to quantitative PCR of cDNA prepared from small pools of animals. As shown in Figure 5, wild-type males, Cb-tra-1(nm2) pseudomales, and Cb-tra-1(nm2); Cb-fem-2/3 double mutants produce large amounts of fog-3 transcript, while wild-type hermaphrodites and Cb-fem-2/3 single mutants produce very little. Expression levels between these two classes are significantly different from each other (e.g. 2-tailed T-test for difference between Cb-tra-1(nm2); Cb-fem-3(nm63) and Cb-fem-3(nm63), p=0.038), but not within them. We conclude that Cb-fog-3 transcript levels in double mutants are not distinguishable from each other nor from Cb-tra-1(nm2) alone.
C. elegans hermaphrodites lacking function of factors involved in muscle physiology, cell cycle regulation, and signaling have been reported to be Emo (Aono et al. 2004; Hajnal and Berset 2002; Inoue et al. 2006; Iwasaki et al. 1996; Kostic et al. 2003; Kuwabara et al. 2000; Ono and Ono 2004; Wissmann et al. 1999). Ablations of somatic gonad cells that block ovulation also produce endomitotic oocytes (McCarter et al. 1997). Although tra-1 oocytes have not previously been described as endomitotic, the data presented here clearly show that tra-1(lf) germ cells from two different species frequently are.
Our observations are consistent with mutant animals being increasingly likely to be Emo as they age. Endomitotic cells are also typically proximal (“downstream”) to oocytes that are in diakinesis of meiosis I, the stage at which normal oocytes arrest prior to fertilization. It is thus likely that the Tra Emo phenotype results from activation of the cell cycle in essentially mature oocytes in the absence of fertilization and ovulation, just as it does in Emo hermaphrodites. Oocyte activation in C. elegans hermaphrodites is normally mediated by the sperm-derived MSP signal (Miller et al. 2001; Miller et al. 2003). As tra-1(lf) mutant males almost always produce at least a few sperm, these sperm may signal to oocytes much as they would in a hermaphrodite gonad. Activation is presumably not accompanied by fertilization, however, as embryos with eggshells are never seen. As tra-1(lf) sperm are capable of cross-fertilization with hermaphrodites, this may be due to a general inability of fertilization to occur in a male somatic gonad, perhaps more specifically to failure of sperm that are otherwise competent to undergo activation inside the testis. This activation is accomplished in males by passage through the vas deferens during ejaculation, and in C. elegans hermaphrodites by contact with the spermatheca (L’Hernault 1997; Ward and Carrel 1979). Neither of these mechanisms would be available to tra-1(lf) spermatids with respect to their ability to fertilize adjacent oocytes in the same gonad.
In our previous analyses of Cb-fem-2 and Cb-fem-3 mutants, we saw no differences in their single mutant germline phenotypes (Hill et al. 2006). Given the common participation of their C. elegans orthologs in a complex that acts to regulate TRA-1 proteolysis (Starostina et al. 2007; Tan et al. 2001), this was not surprising. However, all three C. elegans fem mutations are epistatic to tra-1 in the germ line, indicating that, at least in that species, there must be some additional male-promoting role for the fem genes other than their regulation of TRA-1 levels. We therefore sought to determine whether this was a general feature of Caenorhabditis sex determination by examining the equivalent mutants in C. briggsae. The Cb-fem-3(nm63) mutation eliminates the L4 oogenesis seen in Cb-tra-1(nm2) mutants, produces near-complete production of sperm and eggs early in adulthood, and qualitatively enhances oocyte development relative to tra-1(nm2). This represents the closest thing to a well-organized hermaphrodite germ line as a Tra male could possibly make. We therefore suggest that Cb-fem-3 is epistatic to Cb-tra-1(nm2) in the germ line, similar to what is observed in C. elegans (but with a different phenotype). Whether this is due to direct effects of the loss of Cb-fem-3 in germ cell nuclei, or indirect effects mediated by the surrounding somatic gonad is unclear. That most Cb-fem-3; Cb-tra-1 animals have normal male gonads suggests Cb-fem-3 epistasis is a germ cell property, but the low level of two-armed gonads seen is consistent with somatic influences.
Surprisingly, there is no indication that Cb-fem-2(nm27) has the same “hermaphroditizing” effect on Cb-tra-1(nm2) observed for Cb-fem-3(nm63). By both qualitative and quantitative criteria, there are no differences between the single and double mutants. This represents a departure from the uniform behavior of the fem genes in both single and double mutant analyses in C. elegans, and from single mutant phenotypes in C. briggsae. There are two potential reasons for the difference. First, the nm2 allele is not likely to represent a complete functional null. It may therefore be that the two Cb-fem mutations have different effects on residual TRA-1 product, such that their loss has different consequences. Alternatively, given the essentially perfect somatic masculinization seen in Cb-tra-1(nm2), it is possible that any residual TRA-1 product it produces is insufficient to impact sex determination (though it may effect somatic gonad development; Kelleher et al. 2008; Mathies et al. 2004). In this scenario, the feminization seen when Cb-tra-1(nm2) is combined with Cb-fem-3(nm63) must result from a tra-1-independent role of Cb-fem-3 in promoting male germline fates. If we interpret the low level of two-armed gonads seen in Cb-tra-1(nm2); Cb-fem-3(nm63) as feminization, this effect extends (albeit weakly) to the soma as well.
The quantitative RT-PCR results shown in Figure 4 suggest that the putative Cb-tra-1- independent promotion of sperm fate by Cb-fem-3 does not work through transcriptional regulation of Cb-fog-3, the ortholog of the sole known germline target of C. elegans tra-1. Chen and Ellis (2000) examined the effects of tra-1 and fem mutations on fog-3 transcript levels in C. elegans. They found that XX tra-1 mutants have elevated fog-3 mRNA levels, fem mutants have reduced levels, and tra-1; fem double mutants have levels of fog-3 mRNA comparable to (or slightly higher than) wild-type hermaphrodites, even though no sperm are made. We observed no downshift of Cb-fog-3 mRNA levels when Cb-fem-3 is combined with Cb-tra-1, even though strong feminization is observed. It thus appears that in both species cells that are fated to be oocytes can still express fog-3 if they lack fem-3.
Whether the phenotypes of the double mutants analyzed here differ because of differential regulation of residual Cb-tra-1 function in nm2 mutants or to a tra-1-independent role for fem-3 alone, another mystery is what molecular mechanism would enable them to have distinct phenotypes. One possibility relates to intrinsic conditional expression of phenotype and maternal effects. In C. elegans, null alleles of the three fem genes manifest distinct degrees of temperature sensitivity and maternal rescue. Full expression of feminization by fem-2 mutants requires that the homozygous mutant both lack maternal product and be grown at elevated temperature, while fem-3 is zygotic and largely temperature-independent (Hodgkin 1986). Differences in their phenotype could, in principle, be due to different levels of maternal rescue or temperature dependence. However, as we analyzed stocks that had been homozygous for both Cb-fem mutations for multiple generation prior to analysis, and since earlier work found no evidence of maternal rescue or temperature sensitivity (Hill et al. 2006), this is not likely. A more likely possibility is that Cb-fem-3 has a biochemical activity that does not rely upon the presence of the other Cb-fem genes, while Cb-fem-2 does not. This putative activity may nevertheless normally occur in the context of the FEM complex, but it also possible that it participates in one or more as-yet-uncharacterized alternative interactions.
That Cb-fem-3 would have a particularly important role relative to other Cb-fem genes has some precedent in the C. elegans literature. For example, overexpression of fem-3 is sufficient for masculinization of XX animals, while that of fem-1 and fem-2 are not (Lee and Portman 2007; Mehra et al. 1999). Also, Schedl et al. (1989) reported that fem-3(gf) masculinized the germline of tra-1(lf) mutants, which would again indicate that fem-3 has a tra-1-independent effect on sexual fate. An important subject for future research will therefore be the discovery of this mechanism.
Cb-tra-2; Cb-fem double mutants are self-fertile hermaphrodites (Hill et al. 2006), and thus, unlike in C. elegans, neither Cb-tra-2 nor Cb-fem-3 are required for the normal physiological control of hermaphrodite spermatogenesis. This was interpreted by Hill et al. (2006) as evidence that the control of XX spermatogenesis acts further downstream in the sex determination pathway (e.g. Cb-tra-1, Cb-fog-3, Cb-fog-1, and perhaps unknown factors). In this study, simultaneous loss of Cb-fem-3 and Cb-tra-1 also produced a temporally regulated bisexual gonad (although in a male soma). The absence of Cb-fem-3 activity thus appears to be sufficient to trigger the hermaphrodite germline program, even in the absence (or near-absence) of Cb-tra-1. We also note that the ability of a well-formed male somatic gonad to house a vigorous hermaphrodite germ line in C. briggsae argues against the importance for soma-germ line interactions in determining sexual fate that has been reported in C. elegans (McCarter et al. 1997). Overall, the available data suggest that tra-1-independent mechanisms of sex determination are important in the germ line of both C. elegans and C. briggsae, and that in C. briggsae (but not C. elegans) these mechanisms were the locus of regulatory changes that allowed the evolution of hermaphroditism.
We thank Masato Yoshizawa and Karen Carleton for advice on qRT-PCR, and Harold Smith and Steve Mount for useful discussions. This research was supported by grant 5R01GM079414 to ESH from the National Institute for General Medical Sciences of the NIH.