Extensive genetic experiments have indicated that seven genes involved in sex determination in C. elegans (
her-1,
tra-2 &
tra-3,
fem-1,-2,-3,
tra-1) are organized in a regulatory hierarchy. In all previous tests examining somatic sexual phenotype, mutations of
tra-2 and
tra-3 were found to be epistatic to mutations of
her-1, mutations of the fem genes were found to epistatic to
her-1,
tra-2 and
tra-3, and mutations of
tra-1 were epistatic to mutations in any of the other six genes. Hence the basic model:
her-1 --|
tra-2,-3 --|
fem-1, -2, -3 --|
tra-1 At the 1995 International C. elegans Meeting, a surprising result was described that contrasts with this consistent pattern. Betsy Goodwin reported that the strongest hypermorphic (feminizing) mutation of
tra-2,
e2020, is epistatic to loss-of-function (masculinizing) mutations of
tra-1. In view of this report, which could modify our current understanding of the sex determination pathway, I have re-examined the epistatic relationship between
tra-2(
e2020) and
tra-1(lf) mutations. I was unable, however, to repeat Betsy's finding. The following results indicate that
tra-1(lf) mutations appear to be fully epistatic to
tra-2(
e2020), in all aspects of somatic phenotype. The germ line phenotype also exhibits partial epistasis. First,
tra-2(
e2020);
tra-1(lf) XX animals were generated by carrrying out single pair matings between individual
tra-2(
e2020)
unc-4/++;
tra-1/+ XX females and
tra-2(
e2020)
unc-4/++;
tra-1 XX males. Initial crosses were carried out using a standard strong lf mutation of
tra-1,
e1099. Parental genotypes were verified by progeny testing. A total of 241 Unc progeny were scored, of which half (129/241) were clearly male in phenotype. These must have been
e2020;
e1099 XX homozygotes, apart from a few rare
e2020/+;
e1099 recombinants. 51 of these were examined by Nomarski microscopy: all were unambiguously and perfectly male with respect to tail structure, absence of yolk, body size and so on. Somatic gonad phenotypes were variable, like
tra-1(
e1099) alone, but never feminized. Germ line phenotypes exhibited more oogenesis than
e1099 alone (about half the animals contained oocyte-like cells), but also sperm were seen in some individuals, in contrast to
e2020 XX females, which never make sperm. Similar crosses were carried out using
tra-1(
e1834), which is a deletion of most of the 5' half of
tra-1, with essentially identical results to those obtained with
tra-1(
e1099). Second, in order to test for possible maternal effects, and to examine the male fertility of the
tra-2(
e2020);
tra-1(lf) XX homozygotes, a male/female strain was constructed using
e2020 and
e1099. The strain CB5172 consists of
tra-2(
e2020);
tra-1(
e1099)/+ XX females and
tra-2(
e2020);
tra-1(
e1099) XX males. This strain was propagated for many generations, segregating 50% females and 50% males at each generation. No animals of ambiguous sexual phenotype were observed, even after starvation, which enhances the feminizing properties of
tra-2(
e2020). Survival of this strain necessarily depends on the fertility of the
e2020;
e1099 XX males. All exhibit mating behavior, but only about 1 in 9 is able to sire progeny (as compared to about 1 in 5 for normal
e1099 XX males). This is nevertheless enough to permit propagation. When crossed with marked hermaphrodites, the males sire only XX female progeny, confirming homozygosity for
e2020 and
e1099. As expected, the XX strain is unstable: when populations expand above a few hundred individuals, rare
e2020;
e1099/+ XO animals arise. Mating by these high fertility males and their
e2020 XO sons soon leads to a pure
e2020 strain. This has the same distinctive feminizing properties as the original
e2020, indicating that no modifiers have accumulated. A stable version of CB5172 was also constructed (CB5190), by including
xol-1. These results, and the strain CB5172, were communicated to Betsy Goodwin last fall, in the hope of clarifying the source of the discrepancy between our observations.