Most cases of suppression by smg mutations are allele-specific ( Hodgkin et al., (1989), Genetics 123 :301). An apparent exception is the sex-determination gene
tra-3, because it was found that all four strong mutations of
tra-3 are smg-suppressible (putative nulls: three different ambers,
e1107,
e1525,
e1903, and one non-amber,
e1767). XX animals homozygous (m-z-) for any of these mutations are strongly masculinized and sterile, while smg;
tra-3 XX animals are self-fertile intersexes, exhibiting partial suppression. We suggested that in this instance, Smg suppression might be mediated by effects on another gene,
tra-2, which could compensate for the defect in
tra-3. In support of this suggestion, it is known that dominant hypermorphic
tra-2 mutations can suppress
tra-3. Also, a weak recessive
tra-2 allele,
e1209, is smg-suppressible, although other
tra-2 alleles are not. If this interpretation were correct, then the
tra-3 and smg gene products might have related, but opposite, effects on
tra-2 activity. Further investigation makes this suggestion less plausible. First, the four strong alleles of
tra-3 are not equally suppressed by smg mutations. At 20 C, all have mean self-progeny broods of less than 1. In the presence of
smg-2(
e2008), the broods are larger but significantly different for each allele: [See Figure 1] Therefore, in one sense Smg suppression of
tra-3 does show allele- specificity. Second, the nature of the
tra-3 gene product, which appears to be a protease (see article by Tom Barnes, this WBG issue), makes it unlikely that there is any direct connection between the action of smg genes and the action of
tra-3.Instead, it is more likely that the smg mutations act by stabilizing
tra-3 messages containing chain-terminating mutations, as in the case of
unc-54 (reported by Pulak & Anderson, 1991 Meeting Abstracts). Three of the
tra-3 alleles are ambers and the fourth,
e1767, could well be a UAA or UGA stop. It is unlikely that all of these mutations lead to nonsense peptides with partial activity. In particular, Tom Barnes has found that the
e1903 amber mutation is located at the putative active site of the
tra-3 'protease', so the nonsense fragment generated by translation of
e1903 message should have zero activity, and elevating the amount of such a fragment would not restore activity. However, very low-level read- through of the UAG codon might supply traces of full-length product. It is clear that the
tra-3 product is required only in minute quantities, so smg-stabilization of nonsense messages might increase the synthesis of full-length
tra-3 product to a level at which partial activity would be seen. The same explanation (stabilization plus read- through) has been proposed for the smg-suppressible allele of
lin-29,
n546, which is also a nonsense (UGA) mutation (Rougvie & Ambros, WBG 11-3: 37). Endogenous read-through might also explain why these
tra-3 mutants are distinctly less masculinized at low temperature. Bob Waterston's studies of
sup-5 and
sup-7 showed that amber suppression is more efficient at low temperature, and this might also be true of read- through.