Rewiring the Sex-Determination Pathway During the Evolution of Self-Fertility

Abstract

Although evolution is driven by changes in how regulatory pathways control development, we know little about the molecular details underlying these transitions. The TRA-2 domain that mediates contact with TRA-1 is conserved in Caenorhabditis. By comparing the interaction of these proteins in two species, we identified a striking change in how sexual development is controlled. Identical mutations in this domain promote oogenesis in Caenorhabditis elegans but promote spermatogenesis in Caenorhabditis briggsae. Furthermore, the effects of these mutations involve the male-promoting gene fem-3 in C. elegans but are independent of fem-3 in C. briggsae. Finally, reciprocal mutations in these genes show that C. briggsae TRA-2 binds TRA-1 to prevent expression of spermatogenesis regulators. By contrast, in C. elegans TRA-1 sequesters TRA-2 in the germ line, allowing FEM-3 to initiate spermatogenesis. Thus, we propose that the flow of information within the sex determination pathway has switched directions during evolution. This result has important implications for how evolutionary change can occur.

Keywords: evolution of gene regulation; nematodes; sex determination.

Fig. 1.

The core sex determination pathway in Caenorhabditis nematodes. a) The top line summarizes the core regulatory interactions that control sexual fates. In XX animals, the TRA-2 receptor is free to be cleaved by TRA-3, releasing an intracellular fragment that binds FEM-3, preventing the FEM complex from working. This allows the TRA-1 transcription factor to be cleaved, forming a repressor of male genes. In XO animals, the male sex hormone HER-1 binds to and inactivates TRA-2. This leaves the FEM complex free to assemble with CUL-2 and ubiquitinate TRA-1, leading to its degradation. Hence, male genes are expressed. b) The phylogeny shows that ancestral Caenorhabditis species were male/female, and three species independently evolved XX animals that reproduce as self-fertile hermaphrodites.

Fig. 2.

Caenorhabditis briggsae tra-2(mx) alleles do not feminize the germ line. a) Alignment of the C-termini of five Caenorhabditis TRA-2 proteins, prepared using MUSCLE. Identical residues are shaded dark gray, and conservative substitutions are shaded light gray. All C. elegans tra-2(mx) alleles affect one of five conserved sites in this region. We made identical C. briggsae mutations for three of the C. elegans alleles (marked in green shades). b) Differential interference contrast photomicrographs of a Cbr-tra-2(v403) hermaphrodite and male. The black arrow indicates the vulva, and one of the self-embryos in the uterus is outlined in yellow. The small blue arrow marks two self-sperm. c) Yeast two-hybrid results, with bait constructs listed before prey constructs. d) Self-brood size of hermaphrodites, with error bars indicating the mean and 95% confidence limits.

Fig. 3.

Caenorhabditis briggsae tra-2(mx) mutations suppress she-1 independent of fem-3 function. a, b, d) Graphs showing the percent of XX animals of each genotype that develop as self-fertile hermaphrodites, rather than as females. The error bars show 95% confidence intervals. a) N = 29, 62, 67, respectively. b) N = 29, 87, 62, 60. d) N = 7, 87, 88. c, f) Semi-quantitative RT-PCR results showing fog-3 expression (lower band) and control ama-1 expression (upper band). These two products were produced in separate PCRs (using equivalent amounts of the cDNA template) and run on a single gel. Each lane used cDNA template produced from a mixture of 5 mid-L4 XX larvae. e) Self-brood size of hermaphrodites, with error bars indicating the mean and 95% confidence limits. g) Diagram showing the locations of the reference mutation Cbr-tra-2(nm1) and the null allele Cbr-tra-2(v440). The v440 allele truncates the protein prior to the crucial HER-1-binding site and removes the entire intracellular domain, so it represents a molecular null.

Fig. 4.

A C. briggsae tra-1 mutation that blocks TRA-2-binding favors spermatogenesis. a) Diagram of the structure of C. briggsae TRA-1, indicating the location of the double frameshift mutant v197 v383. Orange marks the conserved gain-of-function domain, green the five conserved zinc fingers, and purple the conserved TRA-2-binding domain. b) Alignment of five Caenorhabditis TRA-1 proteins, showing part of the TRA-2-binding domain. It was prepared using MUSCLE. Identical residues are shaded dark gray, and residues with conservative substitutions are shaded light gray. c) Yeast two-hybrid results, with bait constructs listed before prey constructs. d) The number of self-progeny for animals of the indicated genotypes. The thick line indicates the mean, and the error bars represent 95% confidence intervals. e) The number of sperm in individuals of the indicated genotypes, observed following DAPI staining. f) Extensive sperm in one ovotestis of a tra-1(v197 v383) XX animal. g) Graph showing the percent of XX animals of each genotype that develop as self-fertile hermaphrodites, rather than as females. The error bars show 95% confidence intervals. N = 113, 73, 212, and 67, respectively.

Fig. 5.

Caenorhabditis elegans tra-2(mx) alleles increase TRA-2's ability to target the FEM complex. a, c) Animals of the indicated genotypes were generated from crosses using tra-2(mx) fathers as described in the Materials and Methods and raised at 20 °C. On its own, each fem mutation is recessive. The error bars show 95% confidence intervals. a) N = 30, 92, 79, 25, 76, 73, 59, 100, and 61, respectively. c) N = 88, 89, 61, 69, 75, 49, 61, 75, 100, 64, 69, and 72, respectively. b) Location of the nonsense alleles we used in C. elegans fem genes. d) Semi-quantitative RT-PCR results showing fog-3 expression (lower band) and control ama-1 expression (upper band). These two products were produced in separate PCRs (using equivalent amounts of the cDNA template) and run on a single gel. Each lane used cDNA template produced from single mid-L4 XX larvae. The quantitation on the right was done using Carestream MI software. The fog-3 values are normalized to the average wild-type values for ama-1. (All calculations were done in arbitrary units.) Each genotype shows bars for the mean and standard deviation. The P-value was computed for a comparison of wild type with combined mutant data using the Mann–Whitney U test.

Fig. 6.

Recent evolutionary changes have rewired Caenorhabditis sex determination. a) Models for how the relative concentrations of TRA-2ic (purple loop), TRA-1 (pink), and FEM-3 (blue) specify germ cell fates. The stippled ovals enclose factors that promote spermatogenesis. In C. briggsae, TRA-2_ic binds TRA-1 to prevent spermatogenesis. We present one possible mechanism by which it might act, in which full-length TRA-1 promotes the expression of sperm genes and TRA-2_ic blocks this activation step. (Although TRA-1 repressor also participates, we omit it for clarity.) By contrast, C. elegans FEM-3 plays an essential role in spermatogenesis downstream of TRA-1. Thus, if TRA-2_ic fails to bind TRA-1, it is free to inactivate FEM-3 and prevent spermatogenesis. b) Model for sex determination in the germ cells of larval C. briggsae hermaphrodites. Genes or proteins that promote male fates are in blue, and those that promote female fates in red. Proteins are in capitals and genes in lowercase italics. Arrows denote positive interactions, and “—|” denotes negative ones. Critical relationships in the regulation of spermatogenesis are highlighted in black and the key TRA-2/TRA-1 interaction in red. c) Similar model for C. elegans.