Bifunctional messenger RNAs in eukaryotes

Abstract

Most eukaryotic mflNAs have a single open reading frame and a single functional initiation site, which is usually the AUG codon that lies closest to the 5’-end (Kozak, Cell 15, 1109-1123, 1978; Microbial. Rev. 47, l-45, 1983). The mechanisms that enforce this monocistronic rule have been elucidated in part by studying the small subset of mRNAs that break the rule: mRNAs, constituting some 50/o-10% of the total, in which spurious AUG codons occur upstream from the start of the long open reading frame, and the even rarer mRNAs, all of viral origin, that direct the synthesis of two separately initiated polypeptides. My primary objective in reviewing these bifunctional viral mRNAs and the mechanisms that underlie their misbehavior is to call attention to certain cellular mRNAs that may also be able to produce two proteins. This exercise in forecasting seems justified by past experience: with every bifunctional viral message, the second protein went undetected until it was realized, based on the nucleotide sequence, that a second protein should be produced. Synthesis of Two Proteins from Overlapping Reading Frames The monocistronic character of most eukaryotic mRNAs, and the occasional deviations therefrom, can be rationalized by ascanning mechanism in which both position (i.e., proximity to the Y-terminus) and flanking sequences dictate which AUG codon will initiate translation. The scanning model postulates that a 40s ribosomal subunit binds initially at the capped 5’-end of a message and migrates linearly until it reaches the first AUG codon. If the first AUG codon lies in an optimal context, which in higher eukaryotes is CCACCAUGG (Kozak, Cell 44,283-292,1986), the 40s subunit migrates no farther; it couples with a 60s subunit and protein synthesis initiates uniquely at that site. When the context around the first AUG codon is less favorable (positions -3 and +4 are especially critical), some 40s subunits nevertheless stop and initiate there, while some bypass the first site and initiate at the next AUG codon downstream. This “leaky scanning” mechanism rationalizes most of the known instances of dual initiation in animal virus systems, as summarized in Table 1. By analogy with these well-studied viral mRNAs, it seems reasonable to expect certain cellular mRNAs to be bifunctional. When the V-proximal AUG codon lies in an unfavorable context and in the same reading frame as the second AUG codon, ribosomes should initiate at both sites, producing “long” and “short” versions of the encoded polypeptide. This is the predicted pattern of expression for the gene that is abnormal in human chronic granulomatous disease (Royer-Pokora et al., Nature 322, 32-38,1986), as well as for some oncogenes (Rosson and Reddy, Nature 379, 604-606, 1986; Rao et al., PNAS 83, 2392-2396, 1986), hormone receptors (Meriino et al., MC6 5, 1722-1734, 1985; Miesfeld et al., Ceil 46,389-399, 1986) and growth factors (Gray et ai.: Nature 312,721-724, 1984; Derynck et al., Nature 376, 701-705, 1985). Since N-terminal amino acids often determine the intracellular distribution, and sometimes the activity of proteins (e.g., oncogene products), the predicted synthesis of long and short forms might have regulatory implications. The consequences are even more interesting when the first and second AUG codons lie in different reading frames, and an unfavorable context around the first AUG codon allows some ribosomes to reach the second. The prediction that such mRNAs can synthesize two proteins from different, overlapping reading frames has again been verified with viral mRNAs, as indicated in Table 1. Although no cellular mRNA has yet been proven to work this way, one promising candidate is the message that encodes a murine lymphokine-a 20,000 dalton protein that initiates at the AUG codon nearest the 5’-end of the message (Noma et al., Nature 379,640-646, 1986). Because the first AUG codon lies in an unfavorable context for initiation (UUGAUGG), some 40s subunits should bypass that