SUF12 suppressor protein of yeast. A fusion protein related to the EF-1 family of elongation factors.

Mutations at the suf12 locus were isolated in Saccharomyces cerevisiae as extragenic suppressors of +1 frameshift mutations in glycine (GGX) and proline (CCX) codons, as well as UGA and UAG nonsense mutations. To identify the SUF12 function in translation and to understand the relationship between suf12-mediated misreading and translational frameshifting, we have isolated an SUF12+ clone from a centromeric plasmid library by complementation. SUF12+ is an essential, single-copy gene that is identical with the omnipotent suppressor gene SUP35+. The 2.3 x 10(3) base SUF12+ transcript contains an open reading frame sufficient to encode a 88 x 10(3) Mr protein. The pattern of codon usage and transcript abundance suggests that SUF12+ is not a highly expressed gene. The linear SUF12 amino acid sequence suggests that SUF12 has evolved as a fusion protein of unique N-terminal domains fused to domains that exhibit essentially co-linear homology to the EF-1 family of elongation factors. Beginning internally at amino acid 254, homology is more extensive between the SUF12 protein and EF-1 alpha of yeast (36% identity; 65% with conservative substitutions) than between EF-1 alpha of yeast and EF-Tu of Escherichia coli. The most extensive regions of SUF12/EF-1 alpha homology are those regions that have been conserved in the EF-1 family, including domains involved in GTP and tRNA binding. It is clear that SUF12 and EF-1 alpha are not functionally equivalent, since both are essential in vivo. The N-terminal domains of SUF12 are unique and may reflect, in part, the functional distinction between these proteins. These domains exhibit unusual amino acid composition and extensive repeated structure. The behavior of suf12-null/SUF12+ heterozygotes indicates that suf12 is co-dominantly expressed and suggests that suf12 allele-specific suppression may result from functionally distinct mutant proteins rather than variation in residual wild-type SUF12+ activity. We propose a model of suf12-mediated frameshift and nonsense suppression that is based on a primary defect in the normal process of codon recognition.