A profusion of controls

YONE who teaches knows there are two questions to steer clear of on exams: the one about evolution, to which there is no certifiable answer; and the one about gene control, to which every conceivable answer is correct. Or so it seems at times. The profusion of transcriptional and posttranscriptional controls that have emerged in recent years can be simplified by recognizing categories. One category involves the regulation of mRNA stability. Short sequence motifs that confer sensitivity or resistance to nucleases, thereby controlling mRNA turnover, have been identified in both prokaryotes and eukaryotes (reviewed by Brawerman, 1987). Here I would like to focus attention on three other categories of gene regulation: (a) programmed variations in mRNA structure, particularly in ways that affect translation; (b) translation-linked control of mRNA metabolism; and (c) regulation of translation by RNA-binding proteins. The structure of mRNA determines both the form and yield of the encoded protein, mRNA structure in turn depends on where transcription initiates and how transcripts are edited during the splicing and polyadenylation steps. The use of alternative start sites for transcription can affect the translatability of eukaryotic mRNAs in two ways. Promoter switching may produce from one gene two forms of mRNA, one of which begins slightly farther upstream than the other. In cases where the longer transcript includes an upstream in-frame AUG codon, initiation of translation from that site will add an "extra" NH2-terminal domain to the protein, as illustrated in Fig. 1; the biological consequences of the extra domain are significant, as indicated in Table I. An important point, predicted by theory (Kozak, 1980, 1983) and verified experimentally (see references in Table I), is that the shorter version of the protein can be synthesized only from the 5'-truncated form of mRNA even though its initiation site is present, internally, in the longer transcript. Thus, to produce two versions of the protein, two versions of mRNA are required. (There is a way to generate two proteins by initiating at the first and second AUG codons in a single mRNA, but that "leaky scanning" process requires that the upstream AUG codon occur in an unfavorable context for initiation [Kozak, 1986]; the mRNAs in Table I do not meet that requirement.) In other cases, promoter switching does not affect the form of the encoded protein, but the 5'-noncoding sequence is changed in a way that makes translation more or less difficult. The transcription …