Independent recruitments of a translational regulator in the evolution of self-fertile nematodes

Abstract

Pleiotropic developmental regulators have been repeatedly linked to the evolution of anatomical novelties. Known mechanisms include cis-regulatory DNA changes that alter regulator transcription patterns or modify target-gene linkages. Here, we examine the role of another form of regulation, translational control, in the repeated evolution of self-fertile hermaphroditism in Caenorhabditis nematodes. Caenorhabditis elegans hermaphrodites initiate spermatogenesis in an otherwise female body through translational repression of the gene tra-2. This repression is mediated by GLD-1, an RNA-binding protein also required for oocyte meiosis and differentiation. By contrast, we show that in the convergently hermaphroditic Caenorhabditis briggsae, GLD-1 acts to promote oogenesis. The opposite functions of gld-1 in these species are not gene-intrinsic, but instead result from the unique contexts for its action that evolved in each. In C. elegans, GLD-1 became essential for promoting XX spermatogenesis via changes in the tra-2 mRNA and evolution of the species-specific protein FOG-2. C. briggsae GLD-1 became an essential repressor of sperm-promoting genes, including Cbr-puf-8, and did not evolve a strong association with tra-2. Despite its variable roles in sex determination, the function of gld-1 in female meiotic progression is ancient and conserved. This conserved role may explain why gld-1 is repeatedly recruited to regulate hermaphroditism. We conclude that, as with transcription factors, spatially localized translational regulators play important roles in the evolution of anatomical novelties.

Many important adaptations involve localized modifications of development. Because the genes that regulate development often function in multiple times and places, mutations in cis-regulatory elements that locally alter their expression are expected to offer a simple route to tissue-specific changes in function (1, 2). This circumvention of pleiotropy by changes in gene regulation, recently dubbed the Stern-Carroll Rule (3), has been borne out in both animals and plants (e.g., refs. 4–10). In each case, transcriptional enhancers appear to have been the target of selection.

In Caenorhabditis nematodes, self-fertile hermaphrodites evolved independently from females in Caenorhabditis elegans and Caenorhabditis briggsae (11–14). Selfing is an important reproductive adaptation that profoundly affects the efficacy of natural selection (15), population genetic variation (16–18), and genome content (19). However, the limited XX spermatogenesis that underlies hermaphroditism represents a developmental novelty worthy of study in its own right. In particular, the prominence of posttranscriptional gene regulation in the germ line (20, 21) suggests self-fertility may evolve by mechanisms that are distinct from those described in the soma.

Here we compare the role of GLD-1, a regulator of translation (22), in C. elegans and C. briggsae germ-line sex determination. GLD-1 is an RNA-binding protein of the STAR (for signal transduction and activation of RNA metabolism) family. STAR proteins are implicated in diverse cellular processes, including cell division, gametogenesis, apoptosis, and embryonic and larval development, and are found across the Metazoa (e.g., refs. 23–27). C. elegans GLD-1 is a germ-line–specific, pleiotropic translational repressor (22, 28, 29) required for the mitosis/meiosis decision of germ-line stem cells, meiotic progression of oocyte-fated cells, and specification of C. elegans hermaphrodite sperm in an otherwise female body (30, 31).

In this study we show that gld-1 has been recruited to regulate hermaphrodite development in C. briggsae. However, it acts to promote oogenesis, rather than spermatogenesis as in C. elegans. These alternative roles are the result of differences in the cis-regulatory RNA of a conserved sex-determination gene, tra-2, and in the downstream function of a conserved target, puf-8. Our results provide insights into how pleiotropic translational regulators, as with transcription factors, are redeployed during phenotypic evolution.

Results

Characterization of Cbr-gld-1 Mutations.

In a screen for recessive mutations that cause germ-line–specific sexual transformation in C. briggsae hermaphrodites, the alleles nm41 and nm64 manifested excess sperm and ectopic proliferation of germ cells (Fig. 1 and Fig. S1). The overproduction of sperm resembled the phenotype reported for RNA interference (RNAi) knockdown of Cbr-gld-1 (14). Mutants differed, however, in that simultaneous feminization was not required for the frequent formation of tumors. Both nm41 and nm64 fail to complement, genetically map to Cbr-gld-1, and are associated with a premature stop codon in its ORF (Fig. 1A). Neither allele produced detectable GLD-1 protein (Fig. 1C). However, they differed in expression of the oocyte marker RME-2 (Fig. 1C) (32) and in the frequencies of their mutant phenotypes (Fig. 1B), and thus one or both alleles might retain residual function.

Fig. 1.

C. briggsae and C. elegans gld-1 mutations produce opposite sexual transformations of the hermaphrodite germ line. (A) Structure of Cbr-gld-1, with exon-intron boundaries, conserved coding domains, and mutant lesions indicated. (B) Phenotypic distributions of Cbr-gld-1 alleles measured in XX animals on day 1 of adulthood by DIC microscopy of Hoechst-33258–stained gonads. n = 200 gonad arms examined for each genotype. ectopic prolif., Mitotic proliferation of germ cells proximal to the distal stem-cell niche (Fig. S1). (C) Immunoblots of C. elegans and C. briggsae wild-type L4 and adult hermaphrodites and mutant adults of indicated genotype. q485 is a null allele of Ce-gld-1 (30); q268 and q93oz50 have premature stop lesions affecting the equivalent codons altered in nm64 and nm41, respectively [but note that q93oz50 also harbors a downstream mutation (30)]. RME-2 and MSP antibodies mark oocytes and spermatocytes, respectively; tubulin is a loading control. (D–F) C. briggsae gld-1(nm68) XX extruded gonad, stained with Hoechst-33258 (D), anti-MSP (E), and anti–RME-2 (F). (G–I) C. elegans gld-1(q485) XX extruded gonad, stained as in D to F. spermatocytes (sc), sperm (sp), ectopic proliferation (ep) Experimental manipulations in the two species were performed simultaneously and identically; gonads are representative of each. (Magnification, 140×.)

To confirm the Cbr-gld-1 loss-of-function phenotype, we isolated the deletion mutant nm68, which eliminates conserved sequences important for RNA binding and homodimerization (Fig. 1A) (33). As with nm41 and nm64, no GLD-1 protein is detectable in nm68 homozygotes. We infer that nm68 likely represents a null allele. Like nm41, nm68 causes some ectopic germ-cell proliferation and a high frequency of excess spermatogenesis, with no evidence of oocyte-fated cells. XO Cbr-gld-1 mutants are normal, as judged by differential interference microscopy (DIC) microscopy and mating assays, and Cbr-GLD-1 expression in males is low relative to hermaphrodites (Fig. S2). This finding indicates that, as in C. elegans (30), Cbr-gld-1 has at most a nonessential role in the C. briggsae XO male germ line.

Chromosome staining and immunohistochemistry with an anti-phosphohistone H3 antibody (a mitotic marker) in XX Cbr-gld-1 mutant germ lines indicates that, as for C. elegans gld-1 (30), ectopic germ-cell proliferation results from a failure of germ cells to complete the meiotic program (Fig. 1 B and D–F, and Fig. S1). Unlike C. elegans mutants, however, isogenic Cbr-gld-1 mutant gonads vary substantially in the extent and location of ectopic proliferation and gametogenesis along the proximal-distal axis (Fig. 1D and Fig. S1). Cbr-gld-1 germ cells that ectopically proliferate fail to express detectable amounts of the sperm marker major sperm protein (MSP) or RME-2 (Fig. 1 D–F), consistent with an un- or de-differentiated cell state. Because XX nm68 mutants develop tumors, but Cbr-gld-1(RNAi), with moderate concentrations of dsRNA, can fully masculinize without tumors (14) (Table S1), a low level of Cbr-gld-1 activity may be required for reliable XX sperm development. Other interpretations are possible, however, including a cryptic female or intersexual origin for Cbr-gld-1 mutant germ-line tumors.

To further investigate Cbr-gld-1 germ-line tumor formation, we examined interactions with other sex-determination genes. Mutations in Cbr-tra-2 and Cbr-tra-1 that masculinize the XX germ line and soma (34) suppress Cbr-gld-1(nm68) tumors, and the germ line remains masculinized (Table S1). When only the hermaphrodite germ line is feminized with Cbr-fog-3(RNAi) (35), Cbr-gld-1(RNAi) produces germ-line tumors (14) (Table S1). Furthermore, Cbr-gld-1(RNAi) Cbr- fog-3(RNAi) double-knockdown in C. briggsae wild-type XO males also results in completely penetrant tumor formation in a male somatic gonad (Table S1). Thus, Cbr-gld-1(lf)-mediated tumors form in oocyte-fated germ lines, regardless of somatic sex or karyotype, as in C. elegans (31). Cbr-tra-1 mutants also produce robust oocytes as they age (34, 36). Cbr-gld-1; Cbr-tra-1 mutants produce neither tumors nor differentiated oocytes, but do often produce sperm normally (Table S1).

Phylogenetic Survey of gld-1 Function.

Orthologs of gld-1 exist across Caenorhabditis (Fig. S3), and their highly XX-biased expression is conserved (Fig. S2) (29). We knocked down gld-1 in females of Caenorhabditis japonica, Caenorhabditis brenneri, Caenorhabditis remanei, and C. briggsae/C. sp. 9 F1 hybrids (37), and in C. briggsae hermaphrodites (Fig. 2). This process eliminated differentiated oocytes in all cases, but only caused germ-line masculinization in C. briggsae. In gonochoristic gld-1(RNAi), the female germ line largely fills with ectopically proliferating germ cells that fail to express detectable amounts of RME-2 or MSP (Fig. 2), but males suffer no observable defects and are fertile. We conclude that the XX female ancestors of C. briggsae relied on gld-1 for oocyte meiosis and differentiation, but not for repression of the sperm fate. Thus, Cbr-gld-1 was recruited into germ-line sex determination during (or possibly subsequent to) the evolution of self-fertility.

Fig. 2.

Cbr-gld-1 sperm repression is of recent origin, but its role in meiotic progression is ancient. (Top) Gonads of wild-type C. sp. 9 (representative of gonochoristic species). (Left) Unmated XX female, showing oocytes (o) of one ovarian arm. (Right) XO male with spermatids (sp) and primary spermatocytes (sc) in the single-armed testis. (Middle) gld-1(RNAi) phenotypes in gonochoristic Caenorhabditis. ep, ectopic proliferation; sp, sperm. DIC micrographs of gonads of adult XX progeny of injected mothers are shown (absence of sperm in gonochoristic species was confirmed by staining with Hoechst-33258 dye). (Lower Right) Immunoblots of proteins from untreated (lanes 2, 4, 6, 8, 10) or gld-1 loss-of-function (lanes 1, 3, 5, 7, 9) XX adults of C. japonica (lanes 1 and 2), C. elegans (lanes 3 and 4), C. remanei (lanes 5 and 6), C. briggsae-C. sp. 9 F1 hybrids (lanes 7 and 8), and C. briggsae (lanes 9 and 10). Loss-of-function for C. elegans was via gld-1(q485), for C. briggsae was via Cbr-gld-1(nm68), and via species-specific gld-1(RNAi) treatments for all others. For C. briggsae-C. sp. 9 hybrids, the C. briggsae sequence was used. (Bottom) Phylogenetic interpretation (tree compiled from refs. 37 and 62), with inferred origins of self-fertility (gray) and of Cbr-gld-1’s role as a sperm repressor (black) indicated.

Context-Dependent Role of gld-1 in Hermaphrodite Sex Determination.

gld-1 is pleiotropic (31), has hundreds of target mRNAs (38), and is itself both positively and negatively regulated at the mRNA (39–41) and protein (42, 43) levels. The opposite sex-determination phenotypes for gld-1 in C. briggsae and C. elegans could be intrinsic to gld-1 itself. Alternatively, the strong conservation of GLD-1 sequence (Fig. S4) and expression pattern (14) suggested that factors with which GLD-1 interacts may be responsible. To test these alternatives, we introduced a Cbr-gld-1 transgene into C. elegans gld-1(q485)-null mutants. Two transgenic lines expressing Cbr-GLD-1 (Fig. S5) restored both robust XX spermatogenesis and normal oogenesis to gld-1(q485) homozygotes (Fig. 3). Thus, the opposite roles of gld-1 in C. elegans and C. briggsae germ-line sex determination are because of species-specific context.

Fig. 3.

Cbr-gld-1 can fully substitute for C. elegans gld-1. (A) Germ-line phenotype of adult XX C. elegans gld-1(q485) mutants showing extensive ectopic proliferation of oocyte-fated cells that have exited meiosis and reentered mitosis. ep, Ectopic proliferation. The distal tip of the anterior gonad is marked with an asterisk. v, vulva. (B) Ce-gld-1(q485) homozygote bearing an HA epitope-tagged wild-type Cbr-gld-1 transgene with a normal rachis (r), oocytes (o), sperm (sp), and abundant selfed embryos (e). The distal tip of the anterior gonad is marked with an asterisk. v, vulva. Micrograph is representative of the animals with this genotype. (Magnification, 100×.)

Cbr-puf-8 Is an Oogenesis-Promoting Target of Cbr-GLD-1.

As GLD-1 represses mRNA translation (22, 25), we asked whether the Cbr-gld-1 excess sperm phenotype could be explained by an association between Cbr-GLD-1 and sperm-promoting mRNAs. To address this question, GLD-1-associated mRNA was immunoprecipitated from wild-type worms (Fig. S6A). An initial survey using quantitative RT-PCR (qRT-PCR) failed to implicate any known C. briggsae sperm-promoting mRNAs (Fig. S6B). To identify new sperm-promoting targets, we queried a whole-genome microarray with Cbr-GLD-1–associated mRNA and performed RNAi knockdown of 125 putative target genes. Only knockdown of the Puf family RNA-binding protein gene, Cbr-puf-8, produced the expected feminized mutant phenotype (Table 1), but the effect was weak.

Table 1.

Genetic interactions between Cbr-puf-8 and other Puf family genes

	Germ-line phenotype*†
Targets of RNAi	Sperm only	Sperm + oocytes	Oocytes only	Reduced germ line, no gametes
Cbr-gld-1‡	240	0	0	0
Cbr-puf-8	0	172	9	4
Cbr-puf-1.2	0	294	6	0
Cbr-gld-1 + Cbr-puf-8	0	0	4	122
Cbr-puf-8 + Cbr-puf-1.1	0	77	0	0
Cbr-puf-8 + Cbr-puf-1.2	0	8	144	0
Cbr-puf-8 + Cbr-puf-1.2 (XO progeny)	22	0	0	0
Cbr-puf-8 + Cbr-puf-2	0	128	6	1
Cbr-gld-1 + Cbr-puf-8 + Cbr-puf-1.2	0	3	43	109

*All phenotypes were scored by DIC microscopy in the progeny of injected wild-type hermaphrodites within the first two days of adulthood. All progeny XX unless otherwise noted.

^†Only sex determination or germ cell proliferation phenotypes are given. Other phenotypes include aberrant or delayed gametogenesis, (largely proximal) ectopic germ-cell proliferation, and degenerate proximal-most oocytes.

^‡Cbr-gld-1(RNAi) was used here because it rarely induces ectopic proliferation and because there is no convenient phenotypic marker for Cbr-gld-1.

C. elegans puf-8 also acts in germ-line sex determination, doing so redundantly with another Puf gene, fbf-1 (44) to inhibit the sperm fate. Although opposite in phenotype, we surmised that, as in C. elegans, other Puf family members might act redundantly with Cbr-puf-8 to regulate germ-line sex. Knockdown of Cbr-puf-8 in combination with each of the other Puf genes positively enriched in the microarray analysis (CBG09894, CBG13175, and CBG01774) produced no sex-determination disruptions, nor did selected double and triple knockdowns. We next examined three other C. briggsae Puf genes related to fbf-1/2 (45), Cbr-puf-1.1 (CBG02701), Cbr-puf-1.2 (CBG13460), and Cbr-puf-2 (CBG02702). Knockdown of Cbr-puf-1.2 is weakly feminized on its own, but strongly enhanced the feminization of Cbr-puf-8(RNAi) (Fig. 4A and Table 1). This feminization is not observed in XO male siblings (Table 1).

Fig. 4.

Cbr-puf-8 is a sperm-promoting target of Cbr-GLD-1. (A) Cbr-puf-8(RNAi);Cbr-puf-1.2(RNAi) feminizes the hermaphrodite germ line. Of the animals that produce gametes, XX adults have only oocytes (o) and have an empty spermatheca (st) and uterus (u). v, vulva. (B) GLD-1 binds Cbr-puf-8 3′ UTR directly in vitro. Wedge, Cbr-GLD-1–dependent complex in gel-shift assay. –CTL RNA is derived from the Cbr-tra-2 5′ UTR. (C) Conserved association of puf-8 with GLD-1 in C. briggsae and C. elegans. qRT-PCR enrichments for GLD-1 immunoprecipitated (IP) vs. mock IP RNA preparations are given for negative control (pan-actin and nol-1), positive control (oma-1/2 and rme-2), and puf-8. Each is expressed as the average of at least three biological replicates (± SEM). (Magnification, 140×.)

Consistent with Cbr-puf-8's role as a major sex-determining target of Cbr-GLD-1, we find that Cbr-puf-8(RNAi); Cbr-puf-1.2(RNAi) fully suppresses the sperm production of Cbr-gld-1(RNAi) (Table 1). This triple knockdown also produces surprisingly normal oocytes in a minority of animals, although most had no overtly differentiated gametes. We also observed a reduced germ-line phenotype with Cbr-puf-8(RNAi), especially in combination with Cbr-gld-1(RNAi) (Table 1), which is also seen in a minority of C. elegans puf-8 mutants (46). In addition to suppressing Cbr-gld-1(lf) germ-line masculinization, the triple knockdown of Cbr-gld-1; Cbr-puf-8; puf-1.2 partially rescues this reduction in germ-cell number. Cbr-puf-8 and Cbr-puf-1.2 may therefore have antagonistic roles in germ-cell proliferation, similar to C. elegans fbf-1 and fbf-2 (40, 45).

A likely GLD-1 binding site (33, 38) is present at nucleotides 25–31 3′ of the Cbr-puf-8 stop codon. The purified STAR domain of Cbr-GLD-1 and a synthetic puf-8 3′ UTR fragment containing this region interact in vitro without other factors (Fig. 4B). Taken together, the genetic and molecular evidence are consistent with the Cbr-puf-8 mRNA being a direct target of Cbr-GLD-1, and suggest that de-repression of Cbr-puf-8 is a major contributor to the germ-line masculinization of Cbr-gld-1 mutants.

The 3′UTR of C. elegans puf-8 also contains a potential GLD-1 binding site (33), and Ce-GLD-1 and Ce-puf-8 mRNA interact in vivo (Fig. 4C). Because Ce-puf-8 promotes the oocyte fate with fbf-1 (44), this finding raised the possibility that Ce-puf-8 hyperactivity contributes to the feminization of C. elegans gld-1 mutants. However, in an epistasis test, XX puf-8 fbf-1; gld-1 triple homozygotes fail to produce differentiated gametes, and germ cells resemble those of gld-1 single mutants (Fig. S7).

Differential tra-2–GLD-1 Association in C. elegans and C. briggsae.

Germ-line hyperactivity of tra-2 is a cause of germ-line feminization in XX Ce-gld-1 mutants (22, 28). C. briggsae tra-2 plays a conserved female-promoting role in both the soma and germ line (34, 47). Cbr-tra-2 mRNA can also be repressed via its 3′ UTR in the soma (48), which in C. elegans is mediated by SUP-26 (49), and potentially the GLD-1 paralog ASD-2 (50) (Fig. S3). However, it is unclear whether Cbr-tra-2 is regulated by Cbr-GLD-1 in the germ line. We therefore asked whether reduced association of tra-2 mRNA and GLD-1 in C. briggsae vs. C. elegans might contribute to their different gld-1 phenotypes. To address this question, we immunoprecipitated GLD-associated mRNA from both species, and compared the extent of tra-2 mRNA enrichment. Although positive and negative controls were enriched to a similar extent in both species, only in C. elegans is there strong association between GLD-1 and tra-2 mRNA (Fig. 5A). Stronger association between GLD-1 and tra-2 mRNA in C. elegans may equate with stronger germ-line tra-2 repression in that species compared with C. briggsae.

Fig. 5.

tra-2 mRNA associates with GLD-1 in C. elegans but not C. briggsae. (A) Comparison of tra-2 association with GLD-1 in C. briggsae and C. elegans. qRT-PCR enrichments for GLD-1 IP vs. mock IP RNA preparations are given for negative controls, positive controls, and tra-2; each is expressed as the average of at least three biological replicates (± SEM). (B) Recent multimerization of the GLD-1–binding site in the C. elegans tra-2 3′ UTR. The experimentally determined tra-2 3′ UTR sequences for C. elegans (12) and C. briggsae (13) are shown. For C. elegans tra-2, the boldface hexanucleotides indicate exact matches to the SBE “conservative consensus” (UACU[C/A]A) that binds GLD-1 (14); for C. briggsae tra-2 they indicate the single variant motif (CACUCA). The 28-nt direct repeat element (DRE) in the Ce-tra-2 UTR is underlined, and the three perfect repeats of the SBE-bearing 10-mer are shown in blue, red, and green type. (C) Model for different contexts of GLD-1 sex-determination function in C. elegans and C. briggsae. Heavy solid bars indicate strong or genetically significant repression; dashed gray lines signify weaker or genetically insignificant repression.

To explain this difference in GLD-1–tra-2 interaction, we compared the tra-2 3′UTR sequences of C. elegans and C. briggsae. GLD-1 binds a short motif in its mRNA targets, the STAR protein-binding element (SBE) (33) or GLD-1-binding motif (GBM) (38). In Ce-tra-2, GLD-1 binding sites are found within larger direct repeat elements (DREs) (28), containing three SBEs/GBMs, plus a fourth candidate site more 5′ in the 3′ UTR (Fig. 5B). In contrast, the Cbr-tra-2 3′ UTR lacks DREs and possesses only a single GBM variant. As GLD-1 association with mRNA is determined by the strength and number of GBMs within UTRs (38), this suggests the differential association of tra-2 and GLD-1 in C. elegans and C. briggsae is because of differences in cis-regulatory RNA sequences in the tra-2 3′ UTR.

Discussion

The above data demonstrate that GLD-1 had an ancestral function in the regulation of female meiotic progression, and that it has been independently recruited to promote or limit hermaphrodite spermatogenesis in C. elegans and C. briggsae, respectively. We provide evidence that these alternative roles are because of both cis-regulatory changes in a key GLD-1 target mRNA (tra-2) and downstream changes that alter the output of a conserved interaction (GLD-1/puf-8). A model summarizing our interpretation of these results is presented in Fig. 5C.

Evolution of cis-Regulatory Elements in the tra-2 mRNA.

The C. elegans tra-2 DRE is not found in other sequenced Caenorhabditis genomes. Multimerized GLD-1 binding sites in the context of species-specific perfect repeats strongly indicates a recent evolutionary event. Further supporting the functional significance of increased SBE/GBM number, C. elegans tra-2 mutants possessing only one DRE exhibit dominant, hermaphrodite-specific germ-line feminization (28, 51). GLD-1 binds RNA as a dimer, but each protomer can potentially bind separate sequence elements (33). Thus, perhaps four GLD-1 dimers could be recruited to the C. elegans tra-2 3′ UTR, but the single SBE of Cbr-tra-2 suggests a maximum of one. Different stoichiometry of the GLD-1/tra-2 mRNA complex may explain some or all of the differential affinity and different gld-1 mutant phenotypes we find in vivo.

Another C. elegans-specific factor that may contribute to the Ce-GLD-1/tra-2 association is FOG-2, an F-box protein that binds GLD-1 and is specifically required for XX spermatogenesis (14). The exact role of FOG-2 in C. elegans tra-2 regulation is not known, although it is not required for GLD-1 binding of tra-2 mRNA in vitro (42). Interestingly, C. briggsae hermaphrodite spermatogenesis also requires its own species-specific F-box protein, SHE-1 (48), at elevated temperatures, but SHE-1 does not physically associate with Cbr-GLD-1.

Alternative Roles for puf-8 Orthologs in Hermaphrodite Sex Determination.

C. briggsae GLD-1's novel role in sperm repression is likely achieved, at least in part, through direct translational repression of Cbr-puf-8. Cbr-PUF-8 is itself a translational regulator, and its C. elegans ortholog is also involved in sex determination. However, as with GLD-1, its sexual roles differ, as it promotes oocyte production (with fbf-1) in C. elegans (44) but sperm production (with Cbr-puf-1.2) in C. briggsae. Because Cbr-puf-8 (alone or with Cbr-puf-1.2) is not required for male spermatogenesis (Table 1), and the female ancestors of C. briggsae did not produce sperm (by definition), its role in C. briggsae germ-line sex determination likely arose during the evolution of selfing in that lineage.

C. elegans GLD-1 and puf-8 mRNA also associate in vivo (Fig. 4C). However, puf-8 fbf-1 fails to restore spermatogenesis to mutants lacking GLD-1 (Fig. S7) or its cofactor, FOG-2 (44). This finding suggests that the Ce-GLD-1–puf-8 interaction is a minor factor regulating germ-line sex compared with Ce-GLD-1–tra-2, but pleiotropy of all three genes complicates interpretation. The roles of gld-1 and puf-8 in C. elegans germ-line stem-cell proliferation (52, 53) also suggest their interaction might be conserved primarily for this. The mRNA targets of PUF-8 and its ancestral functions in Caenorhabditis are unknown. Thus, puf-8 represents an important subject for future comparative studies.

Translational Control and the Evolution of Development.

Recent studies have stressed the importance of cis-regulatory DNA and transcription factors in the evolution of novel phenotypes (1, 54, 55). This work highlights the role of another level of regulation, translational control, in the convergent evolution of a significant adaptation, self-fertile hermaphroditism. Analogous to transcriptional evolution, evolutionarily dynamic cis-acting sequences in mRNAs (e.g., tra-2) can interact with conserved, pleiotropic translational control factors (e.g., GLD-1). In addition, the activity of an RNA-binding protein may be modulated by interactions with protein cofactors (e.g., FOG-2), which may have more recent evolutionary origins and narrower functions. As a result, target mRNAs may gain or lose regulation by a given RNA binding protein, or may have the functional significance of existing regulation (e.g., Cbr-GLD-1–puf-8) changed. Tissues where precise control of mRNA translation or localization is especially important to development or physiology would be expected to make particular use of this mechanism for adaptation. This study has highlighted the nematode germ line as one such tissue, but early embryos (56–58) and nervous systems (59) may show similar dynamics.

Materials and Methods

Cbr-gld-1 alleles nm41 and nm64 were recognized by sterile F₂ progeny of ethyl methanesulfonate -mutagenized AF16 animals, as previously described (27). Cbr-gld-1(nm68) was identified in PCR-based deletion screens (12). Microscopy and immunohistochemistry were performed according to standard methods (SI Materials and Methods). Immunoblots used a custom chicken anti–GLD-1 antibody (see below and SI Materials and Methods) or an anti–GLD-1 polyclonal antibody, generously provided by T. Schedl (Washington University, St. Louis, MO). Anti-MSP and anti–RME-2 antibodies were gifts of D. Greenstein (University of Minnesota, Minneapolis, MN) and B. Grant (Rutgers University, Piscataway, NJ), respectively. RNA interference was assessed by scoring progeny of mothers microinjected with dsRNA at concentrations of ∼2–3 μg/μL (SI Materials and Methods). C. elegans gld-1 mutant strains (see SI Materials and Methods for details) were obtained from the Caenorhabditis Genetics Center, T. Schedl, and J. Kimble (University of Wisconsin, Madison, WI). Rescue of q485 homozygotes by Cbr-gld-1 was mediated by plasmid pAD-g6, a Gateway-based (Invitrogen) derivative of pCR50 [gift of C. Richie and A. Golden (National Institutes of Health, Bethesda, MD)], (SI Materials and Methods). Biolistic transformation of C. elegans was performed by standard methods (60) (see SI Materials and Methods).

RNA immunoprecipitations (RIP) used an affinity-purified polyclonal chicken antibody developed against a peptide from near the N terminus of Cbr-GLD-1 (Open Biosystems) and goat anti-IgY agarose beads (Aves Labs) using a modified version of a published protocol (61) (SI Materials and Methods). For microarray analysis, five replicates of mock (IgY) vs. anti–GLD-1 and three of total mRNA vs. anti–GLD-1 RIPs were performed. cDNA was hybridized to Agilent 44 K dual-color arrays containing at least two probes for all predicted C. briggsae protein-coding genes. Data analysis is described in SI Materials and Methods. qRT-PCR assays on RIP RNA were performed with gene-specific primer pairs (SI Materials and Methods). Gel-shift mobility assays were performed using the purified STAR domain of Cbr-GLD-1 and in vitro transcribed, 5′ radiolabeled Cbr-puf-8 3′UTR (SI Materials and Methods).