Mutations in Two Independent Pathways Are Sufficient to Create Hermaphroditic Nematodes

See allHide authors and affiliations

Science  13 Nov 2009:
Vol. 326, Issue 5955, pp. 1002-1005
DOI: 10.1126/science.1176013


Although the nematode Caenorhabditis elegans produces self-fertile hermaphrodites, it descended from a male/female species, so hermaphroditism provides a model for the origin of novel traits. In the related species C. remanei, which has only male and female sexes, lowering the activity of tra-2 by RNA interference created XX animals that made spermatids as well as oocytes, but their spermatids could not activate without the addition of male seminal fluid. However, by lowering the expression of both tra-2 and swm-1, a gene that regulates sperm activation in C. elegans, we produced XX animals with active sperm that were self-fertile. Thus, the evolution of hermaphroditism in Caenorhabditis probably required two steps: a mutation in the sex-determination pathway that caused XX spermatogenesis and a mutation that allowed these spermatids to self-activate.

The genus Caenorhabditis includes nematodes that use two different mating systems, a gonochoristic one in which individuals are either male or female and an androdioecious one in which individuals are male or hermaphrodite. In C. elegans and C. briggsae, the XX animals are self-fertile hermaphrodites, which have both male and female traits. Their bodies appear female, but germ cells that differentiate during larval development become sperm, which move into the spermathecae during the first ovulation and are stored there for self-fertilization, whereas germ cells that differentiate during adulthood become oocytes, which can be fertilized either by self-sperm or male sperm. In C. elegans, a single pathway controls sexual fate for cells in the body and the germ line. In males, the secreted protein HER-1 inhibits the TRA-2 receptor, which allows a FEM protein/E3-ubiquitin-ligase complex to target the transcription factor TRA-1 for degradation (1, 2). In hermaphrodites, TRA-1 is not degraded and represses male developmental genes such as mab-3, which regulates development of the intestine and male tail, and fog-3, which controls spermatogenesis (1, 2).

The core members of the sex determination pathway in C. elegans are found in related species. In particular, tra-2 (3, 4), fem-2 (57), fem-3 (7, 8), and fog-3 (9) are conserved in both sequence and function in C. elegans, C. briggsae, and C. remanei. Furthermore, the interaction between TRA-2 and FEM-3 is conserved in all three species (8), and a TRA-2/TRA-1 interaction has been documented in C. elegans and C. briggsae (10).

In C. elegans, hermaphrodites lack HER-1 but still make sperm. This trait is due to repression of tra-2 by FOG-2 and GLD-1, allowing spermatogenesis to occur late in larval development (1113). C. briggsae lacks an ortholog of fog-2 (14), but in this species the F-box protein SHE-1 promotes hermaphrodite spermatogenesis (15). In either case, a mutation in a single gene can cause hermaphrodites to become females (11, 15). However, the phylogeny of this genus implies that the ancestor of C. elegans and C. briggsae was gonochoristic (16, 17), and the mechanisms underlying the evolution of self-fertile hermaphrodites from male/female ancestors are unknown.

In C. remanei, which has male and female individuals, knocking down tra-2 activity by RNA interference (RNAi) causes XX animals to develop male bodies and make sperm (4). Because genes that oppose tra-2 are needed for hermaphroditic development in C. elegans and C. briggsae (12, 13, 15), we decreased but did not eliminate tra-2 activity in C. remanei females, which resulted in hermaphrodite-like animals, pseudomales, females, and animals with masculinized germ lines (18). The hermaphrodite-like animals had normal female bodies, including a bilobed gonad like that found in females containing an ovary and spermatheca in each arm, connected by a central uterus (Fig. 1, A to C). The vulva was connected to the uterus and could lay eggs under control of the nervous system and sex muscles. These animals made both male and female germ cells, although they began oogenesis about a day later than females, probably because of the time devoted to spermatogenesis. Because they could not self-fertilize (n = 0/27), we call them pseudohermaphrodites. When we crossed individual pseudohermaphrodites with males, most of them laid eggs, although their broods were smaller than those of the wild type (47 ± 14 eggs and n = 12 for pseudohermaphrodites, compared with 252 ± 46 eggs and n = 5 for wild type; P < 0.009 in an unpaired t test). Thus, many of the oocytes found in pseudohermaphrodites are functional, and their self-infertility seemed to be caused by defective sperm.

Fig. 1

Some tra-2(RNAi) animals are hermaphrodites with inactive spermatids. DIC photomicrographs of (A) C. elegans virgin hermaphrodite, (B) SB146 C. remanei virgin female, and (C) SB146 C. remanei tra-2(RNAi) pseudohermaphrodite. The inset shows a closeup of the proximal end of the ovary from (C), which is boxed in red. Developing embryos are false-colored red, stacked oocytes are false-colored yellow, and the vulva is marked by an arrow. (D) Western blot probed with antibodies to C. elegans (C.e.) major sperm protein (MSP) isolated from samples of 10 individual animals of the indicated species and sex; C. remanei animals were from strain PB4641. (E) DIC photomicrographs of spermatids from pseudohermaphrodites in sperm media (left) or in sperm media with pronase (right). In the diagram, activated spermatids are diagrammed red with clear pseudopods to the right of the micrograph. (Right graph) Frequency of activation of spermatids from males and pseudohermaphrodites. Error bars indicate 95% confidence intervals.

In C. elegans, the major sperm protein stimulates ovulation (19) and forms the cytoskeleton of activated sperm, which allows sperm to crawl toward and fertilize oocytes (20). Because pseudohermaphrodite spermatids did not function, we assayed the major sperm protein and found that it was abundant (Fig. 1D). Next, we found that 78% of the pseudohermaphrodites ovulated within 1 day of producing oocytes (n = 27), compared with 0% of control females (n = 30). Because spermatids from pseudohermaphrodites stimulate ovulation, we concluded that they release major sperm protein. Although the spermatids were normal in these respects, differential interference contrast (DIC) optics revealed that they clustered in the ovaries and did not move into the spermathecae of young adults, suggesting that they had not activated (Fig. 1C, inset). Furthermore, older animals had spermatids in the uterus, but the ovaries and spermathecae were empty, suggesting that ovulation pushed inactive spermatids into the uterus, as in some C. elegans mutants (21).

To assay sperm activation, we examined adult pseudohermaphrodites and males and found that both released spermatids that remained round and inactive (Fig. 1E). By contrast, sperm from C. elegans adult hermaphrodites at this stage had already activated, produced pseudopodia, and begun to crawl (22). When exposed to media containing a mix of proteases that activates C. remanei male spermatids (23), the spermatids from pseudohermaphrodites produced pseudopods at a frequency that was not significantly different from that of their male counterparts (Fig. 1E, P > 0.2315 in an unpaired t test). Thus, pseudohermaphrodite spermatids do not activate in vivo but are capable of activation.

In C. elegans, male seminal fluid activates hermaphrodite spermatids (24). Furthermore, C. elegans males can mate with C. remanei females but do not produce embryos (25, 26). However, crossing C. elegans him-6 males with C. remanei pseudohermaphrodites yielded 142 eggs. Thirteen of these eggs (9%) developed into adult females, although 120 died as embryos and 9 died during larval development. Crossing pseudohermaphrodites with Cr-fog-3(RNAi) males, which cannot sire progeny because they only make oocytes, showed that C. remanei seminal fluid also activated spermatids, because eight mothers produced a total of 16 inviable eggs.

We observed a variety of terminal phenotypes among the self-progeny in these crosses (fig. S1). Some embryos died before 200 min (when apoptosis normally begins), some shortly afterward, and some after morphogenesis. Eight progeny arrested as L2 larvae, and two of these had abnormal granules in their intestines. Control crosses of wild-type strains resulted in 5 to 10% embryonic death (fig. S2), probably because of balanced deleterious mutations found in all C. remanei strains (27). However, more than 25% of the embryos produced by pseudohermaphrodites crossed with C. remanei males died, including 40% during the first day of reproduction, when mothers switch from spermatogenesis to oogenesis. These data suggest that some of the pseudohermaphrodite oocytes are defective. Because the lethality among self-progeny was much higher than among cross-progeny (91% versus 25%), some of the self-sperm are probably defective, too.

In C. elegans, swm-1 encodes a protease inhibitor with two TIL domains, and a short inhibitor with only one, and prevents the premature activation of sperm in males (28). Analysis of predicted proteins suggests that C. remanei, C. briggsae, C. brenneri, and C. japonica all have conserved orthologs of swm-1 (Fig. 2B and figs. S3 and S5). In addition, putative paralogs of SWM-1 were seen in C. elegans and C. briggsae (Fig. 2B and figs. S4 and S5). In the C. elegans genome, the paralog C25E10.8 is an adjacent duplicate of swm-1 (fig. S5C). It is only 81% similar over 137 residues, and its reactive sites differ from SWM-1. Lastly, when deleted, C25E10.8 does not affect sperm activation (28). In C. briggsae, the paralog CBG19173 is also an adjacent duplicate (fig. S5A). It encodes a protein only 75% similar to Cbg-SWM-1 over 137 residues, and both of its reactive sites are different.

Fig. 2

SWM-1 is a conserved regulator of sperm activation. (A) Phylogeny of the elegans group of nematodes (16, 17), as exemplified by a neighbor-joining analysis of the B0238.12 family of TIL domain proteins (green). Bootstrap replication values are shown at the node of each branch, and the TIL protein C25E10.7 is the outgroup [blue, see (18)]. (B) Neighbor-joining tree of the SWM-1 family of proteins.

We dissected C. remanei swm-1(RNAi) males and saw that 65/94 sperm had activated to the spiky stage, whereas control spermatids were inactive. Thus, C. remanei swm-1 appears to regulate sperm activation, like its C. elegans ortholog. When we characterized tra-2(RNAi); swm-1(RNAi) XX animals that produced both sperm and oocytes, we found that 53.3% were self-fertile hermaphrodites. Out of 103 eggs produced, 15 hatched and 10 grew to adulthood; all of the adults were female, as expected for XX self-progeny.

Thus, we propose that self-fertile hermaphrodites originated from the alteration of both the sex-determination and XX sperm-activation pathways.

In C. elegans, the development of self-fertile hermaphrodites requires that FOG-2 bind GLD-1 (13), which interacts with tra-2 mRNAs to block their translation (12). Although there is no FOG-2/GLD-1 complex in C. briggsae (14), the novel gene she-1 drives hermaphroditic development in that species, and genetic tests suggest that she-1 also acts upstream of tra-2 to regulate its activity (15). Lastly, our data show that lowering tra-2 activity in C. remanei can create pseudohermaphrodites. Thus, mutations that affect tra-2 activity might play a general role in the evolution of hermaphrodites, although it remains possible that other genes in the pathway are also important (7).

Because C. elegans germ cell fates can be affected by mutations that change either the HER-1 or TRA-1 binding sites on TRA-2 or the GLD-1 binding site in tra-2 mRNA or that alter a protease that cleaves TRA-2 (1), tra-2 might be more sensitive to changes in upstream regulators than other genes. In addition, it might be better positioned to alter germline fates, because TRA-2 acts through the FEM proteins to control TRA-1 stability and also binds TRA-1 directly (1) (Fig. 3A). Thus, mutations that cause hermaphroditic development by targeting tra-2 might be more frequent than mutations that act elsewhere to cause hermaphroditic development.

Fig. 3

Evolution of self-fertile hermaphrodites. (A) Model of the sex-determination pathway in C. remanei, indicating the branch at TRA-2. Proteins promoting spermatogenesis are blue, and those promoting oogenesis are red. (B) Model of an adaptive landscape in which pseudohermaphrodites are under negative selection because of the resources devoted to sperm production. Red arrows indicate paths opposed by selection, green indicates paths favored by selection, and yellow marks neutral paths. (C) Model of an adaptive landscape in which the negative effect of delayed oogenesis in pseudohermaphrodites is offset by self-fertilization after mating.

Changes to the sex-determination pathway are not sufficient to create self-fertile hermaphrodites because spermatids in affected animals do not self-activate. In C. elegans, mutations in spe-8, spe-12, spe-19, spe-27, or spe-29 block activation of spermatids in hermaphrodites but not in males (29, 30), so the signals that activate spermatids might differ between the sexes. This difference could be due to selection, because male spermatids must remain inactive before mating to improve transfer (28), but hermaphrodite spermatids need to activate during ovulation to avoid being swept out of the gonad.

Before mating, sperm activation in C. elegans males is blocked by SWM-1. Because spe-29; swm-1 hermaphrodites have active sperm but spe-29 hermaphrodites do not, SWM-1 also functions in XX animals (28). This result can be explained by the expression of SWM-1 in germ cells during spermatogenesis, as implied by genetic mosaic analyses (28). Thus, mutations that prevented swm-1 from blocking sperm activation in XX animals, without compromising its role in males, might have been required for the evolution of a hermaphroditic mating system.

Although the gonochoristic species do not form a clade (16, 17), their SWM-1 proteins were more similar to each other than to those of the hermaphrodites (Fig. 2, A and B). Furthermore, maximun likelihood tests indicate that the ratio of nonsynonymous to synonymous substitutions (called Ka/Ks) along the hermaphroditic swm-1 branches differs from that of the rest of the tree [P < 0.0018; see (18)]. Thus, the type of mating system appears to influence the selective pressures acting on swm-1. This difference might have been caused by positive selection on some residues of SWM-1 during the evolution of hermaphroditism. Alternatively, it might be due to a relaxation of negative selection on SWM-1 in hermaphrodites. In either case, these changes could have been facilitated by the presence of paralogs of swm-1, which only exist in the hermaphroditic taxa.

On the basis of our results, we suggest two models describing the evolution of hermaphroditism. Both models assume that selection would strongly favor the ability of individual XX animals to colonize new environments because these nematodes reproduce in large numbers and can quickly exhaust local food sources. Such an advantage would be needed to offset problems caused by inbreeding depression in originally outbreeding populations of nematodes (27). (i) XX animals first acquired the ability to activate self-sperm by a neutral mutation. If so, a subsequent mutation conferring the ability to produce XX sperm and self-fertilize would have immediately been under positive selection when sexual partners were rare (left path, Fig. 3, B and C). (ii) XX animals first acquired the ability to produce spermatids and oocytes. Because our data imply that such animals could not self-fertilize, the production of spermatids should have resulted in reduced fitness (right path, Fig. 3B). However, selection against these incipient hermaphrodites would be alleviated if self-sperm were occasionally activated by male seminal fluid (Fig. 3C). Because C. remanei females are attractive to males of related species (31), a pseudohermaphrodite that mated with such a male would be stimulated to have self-progeny. However, a C. remanei female that mated with a male of another species would not have offspring. Despite this possibility, we favor the model in Fig. 3B.

In Drosophila, male wing spots have been independently gained or lost in different lineages because of mutations in cis regulatory elements that drive the yellow gene (32). Similarly, independent duplications of the gene encoding a pancreatic ribonuclease, followed by similar functional changes, occurred in leaf-eating monkeys (33). However, both examples involve novel traits that were created by modification of only a single gene. Thus, they differ from this study, which focuses on a novel trait that seems to have required independent changes in at least two regulatory pathways. To explain these changes, we favor stepwise models, in which a neutral mutation allowed for selection to act on a second, favorable change. Several species closely related to C. remanei have been isolated recently and will allow us to test whether the evolution of hermaphroditism follows our model throughout the Caenorhabditis genus.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank the NSF for support (grant 0543828), D. Greenstein for antibodies, Y. Shifmann for technical assistance, A. Singson and S. L’Hernault for advice, and E. Moss and R. Dutch for comments on this manuscript.

Stay Connected to Science

Navigate This Article