Reassignment of a stop codon in bacteria can simply involve loss of the appropriate RF, as previously mentioned for the absence of RF2 in Mycoplasma genitalium. However, in eukaryotes, there is only one factor that decodes all the stop codons. Therefore, changes in codon reassignment, i.e., the reassignment of a stop codon as sense, should be reflected by changes in the sequences of the eRF1, since this factor should no longer recognize the reassigned stop codon as sense.
In this respect, the use of alternative nuclear genetic codes makes the ciliates perfect for this type of analysis, for example, some ciliate species, such as the Euplotes have reassigned UGA as Cys (C), whereas other species, such as Tetrahymena, translate both UAA and UAG with Glu (E) (; reviewed in ). Validating this assumption, it was recently demonstrated that, indeed, the ciliate eRF1 does not recognize the reassigned stop codons in vitro [111, 112]. A flurry of sequencing activity saw a rapid increase in the number of available ciliate eRF1 gene sequences [100, 113-116]. These types of analyses revealed that most of the convergent changes were indeed associated with domain I; however, the number of positions decreased significantly as the number of ciliate eRF1 gene sequences available increased, challenging the reliability of this method.
Consistent with the assignment of domain I as the codon-recognition domain, random mutagenesis of yeast eRF1 identified numerous locations scattered through domain I, which altered the stop-codon recognition specificity . The mutations located to a groove formed by two helices (a2 and a3), which was proposed to form a binding pocket into which a triplet stop codon was modeled . The involvement of domain I in codon recognition was convincingly demonstrated when hybrid eRF1, containing the domain I from Tetrahymena eRF1 (which recognizes only UGA) and domains II and III from Saccharomyces cerevisiae eRF1 (which recognizes all stop codons), terminated only at UGA stop codons . Consistently, it was demonstrated recently that a combination of four substitutions in two different regions of domain I had a profound effect on the stop-codon specificity of human eRF1 in vitro, such that it only terminated efficiently at UGA stop codons, similar to ciliate eRF1s . This result suggests that in fact two distinct regions within domain I are involved in codon recognition, and that the protein-anticodon mimicry concept  may, in contrast with the situation in bacteria where decoding occurs through a simple tripeptide motif, be far too simplistic to describe the situation in eukaryotes and bacteria.
The two prime candidates thought to be involved in stop-codon recognition in eRF1 are two loop regions located in domain I, one containing a heptapeptide sequence 58TASNIKS64 (human eRF1 numbering) and the other a consensus sequence i25YxCxxxFi3i (reviewed in ). The close proximity of the TASNIKS sequence to the stop codon was confirmed when the Lys (K) residue was found to be crosslinked when synthetic mRNAs containing 4-thiouridine at the first position of the stop codon were used . This suggests that like the situation in bacteria, the protein factor may be directly decoding the stop codon; however, the mechanism may differ significantly.
It is noteworthy that the conditional lethality associated with a mutation in domain I of the yeast eRF1 (P86A) is rescued by compensating mutations A1491G and U1949 located in helix 44 of the decoding region . Interestingly, the mutation G1491 creates a base-pair with C1409 yielding yeast cells that are extremely sensitive to paromomycin. Furthermore, second-site mutations were identified in the switch region at U912C and G886A of the 18S rRNA . Whether this region actually represents a universal switch has recently been brought into doubt by the creation of the equivalent switch mutants in yeast, which did not exhibit the predicted ram or restrictive phenotypes although they did support the involvement of this region in ribosomal fidelity . In any case, the complexity of stop-codon recognition seems to be a conserved feature between eukaryotes and eubacteria and will require dissection of the (e)RF:termination complex by cryo-EM and crystallization to understand the mechanism fully.
Was this article helpful?