5'-proximal AUG sequences as translation initiation signals on mRNAs in Escherichia coli.

Abstract: A collection of 265 E. coli, coliphage, and E. coli plasmid promoter sequences was examined for ATG triplets at or immediately downstream of transcriptional initiation sites. A majority of the observed 5'-proximal triplets were either start codons for known coding regions or were blocked by in-frame stop codons upstream of known coding regions. The possible significance of such sequences for mRNA translation is discussed.

Key words: translation, untranslated leader, Shine-Dalgarno sequence, leaderless mRNA

Introduction

Messenger RNA molecules (mRNAs) lacking 5' untranslated leader sequences have been observed in eubacteria, archaebacteria, and eukaryotic mitochondria (see 1 and references cited therein). Such mRNAs begin with a 5'-AUG-3' start codon at or very near the 5' terminus of the message, and their lack of 5' untranslated leaders means that they also lack conventional Shine-Dalgarno homologies believed to exist in most prokaryotic messages. The Shine-Dalgarno sequence [2] is a consensus (5'-AAGGAGGU-3') initially characterized in mRNAs of Escherichia coli. It is typically located 5 to 13 nucleotides upstream of a start codon and is complementary to the anti-Shine-Dalgarno sequence found near the 3' terminus of 165 ribosomal RNA. Base-pairing interactions between the Shine-Dalgarno sequence and the anti-Shine-Dalgarno sequence momentarily tether the mRNA to the face of the 30S ribosomal subunit and increase the likelihood that an appropriately spaced AUG codon will be utilized during translational initiation (reviewe d in 3).

Since any naturally occurring leaderless mRNA lacks the untranslated leader and Shine-Dalgarno homology of a "conventional" message, it is reasonable to ask if they contain novel translation initiation signals not found in leadered mRNAs. In experiments conducted by Wu and Janssen [4], removal of an untranslated leader from mRNA for the viomycin phosphotransferase (vph) gene of Streptomyces vinaceus did not prevent its heterologous translation in E. coli. Similarly, Van Etten and Janssen [5] found that removal of the leader from beta-galactosidase (lacZ) mRNA results in only a two-fold reduction in translational efficiency in E. coli compared to that of the leadered wild type message. If leaderless mutant lacZ and vph mRNAs can be translated, this not only argues against the necessity of additional unique translational initiation signals for their translation but also raises the interesting possibility that a 5'-terminal AUG sequence itself can serve as a translation initiation signal provided that it is not occluded by secondary structure within the mRNA molecule.

Two questions follow from this line of reasoning. Given that AUG sequences occur randomly throughout a given mRNA, could one of these sequences act as a start codon if positioned at or near the 5' terminus of a message, and how commonly are AUG sequences found in such a position? These questions had been posed previously in reference to research conducted on leaderless mRNA in a different eubacterial genus, and a list of sequenced wild type and mutant E. coli promoters with known transcriptional start sites was examined [6]. In light of its relevance to more recent work with E. coli [4, 5], the resulting information is reconsidered here.

Methods

A collection of E. coli, coliphage, and E. coli plasmid promoter sequences listed by Hawley and McClure [7] was later assimilated into a larger compilation by Harley and Reynolds [8]. An examination of this list for ATG sequences at or immediately downstream of transcriptional initiation sites was conducted [6]. The background references cited by Hawley and McClure [7] and Harley and Reynolds [8] were then consulted for complete sequence information on the relevant genes. Two examples (Tn1721 tetR and phage P2 gene V) cited by Janssen [1] were added to the list.

Results

Of 265 natural and mutant promoters with known transcriptional start sites [1, 8], 15 direct the synthesis of messages in which the first position of an AUG triplet is the first or second nucleotide transcribed. Three of these (the Tn1721 tetR promoter, phage [lambda] [P.sub.RM], and the phage P2 gene V promoter) are known to direct the transcription of leaderless mRNAs which encode known proteins. Interestingly, 9 of the remaining 12 examples (argCBH, argE P2, crp, dapD, Tn10xxx P3, pBR322 P4, phage T7A3, phage [phi]X174D, and phage [lambda] [P.sub.R]) promote the synthesis of mRNAs in which the leaderless AUG triplets are in-frame with termination codons anywhere from 2 to 9 codons downstream. Of the remaining 3 examples, the 5'-terminal AUG of trfB mRNA is in-frame with the trfB coding sequence and would add 21 amino acids to the gene product if translated. The last 2 examples (cit. util-431 and tyrT/77) were referenced as personal communications for which complete sequence data were not published.

Eight more of the E. coli transcripts cited by Harley and Reynolds [8] have AUG triplets following short leaders of 3 to 7 nucleotides. One of these (transcribed from phage P22 [P.sub.RM]) has a four-nucleotide leader 5' to the c2 repressor coding sequence, but despite the lack of an upstream Shine-Dalgarno sequence, it has not been reported as a leaderless mRNA, perhaps due to the presence of the short leader. Five other transcripts (the ampC, Ipp, tufB, pBR322 Na, and [lambda] [P.sub.R]' mRNAs) have AUG triplets which are blocked by in-frame termination codons upstream of the coding regions for their known gene products. For the last 2 examples, short leaders on the malK and metA transcripts are followed by AUG triplets which are not blocked prior to known downstream coding regions.

Discussion

If a 5'-proximal AUG triplet can serve as a sufficient translational initiation signal, then any AUG sequence positioned in this manner might conceivably be translated, mRNA conformation permitting, regardless of whether or not the utilized start codon is part of a legitimate open reading frame ending with one of the three possible termination codons. Translational elongation by ribosomes which have initiated at a 5'-proximal AUG triplet might interfere with the translation of coding regions located downstream on the same mRNA. Precedent for this line of reasoning was provided by Schottel et al. [9]. Using the E. coli cat gene under the control of the lac promoter, they found that CAT (chloramphenicol acetyltransferase) production was reduced by more than 90 percent in a mutant in which an open reading frame overlapped the cat coding sequence out-of-frame. Additionally, in a mutant leaderless mRNA expressed in Streptoinyces lividans, a 5'-terminal, out-of-frame AUG triplet was translated while the downstream start codon for the aph (aminoglycoside phosphotransferase) coding sequence just one nucleotide away was not [10]. Another possible implication of the translation of illegitimate 5'-proximal start codons concerns energetic waste. Perhaps the energy consumption of translational initiation which does not lead to the synthesis of functional gene products would leave a bacterial cell at a selective disadvantage.

If translation proceeding from illegitimate 5'-proximal AUG triplets is, in fact, energetically unfavorable or potentially disruptive to the translation of overlapping downstream coding regions, then selective pressure might exist for such protein synthesis to terminate relatively quickly. In light of this consideration, it is interesting that in the 15 mRNAs with no leaders or with leaders of only one nucleotide before an AUG sequence, three of the upstream AUG triplets are known start codons (for the [lambda] cI, Tn1721 tetR, and phage P2 gene V coding regions) while 9 of the remaining twelve are blocked by in-frame stop codons from 2 to 9 codons downstream. In an additional 8 mRNAs with short leaders of from 3 to seven nucleotides and lacking a Shine-Dalgarno homology, one 5'-proximal AUG sequence is the start codon for the phage P22 c2 repressor protein, and 5 others are blocked in-frame by termination codons upstream of known coding regions.

In the minority of observed cases where 5'-proximal AUG triplets are not followed by in-frame termination codons prior to known coding regions, stable mRNA secondary structures might prevent 30S ribosomal subunits from interacting with these prospective upstream initiation codons. Alternatively, upstream initiation and subsequent translational readthrough might serve as a means of preventing harmful overexpression of certain gene products from downstream coding regions. Depending on the function of a gene, greater translational efficiency of its mRNA is not necessarily desirable. Finally, a short leader, if present, might sufficiently reduce the translational efficiency of a start codon so that it competes less effectively against a downstream start codon with an appropriately spaced Shine-Dalgarno homology. Even in the absence of a Shine-Dalgamo homology, for example, translational repression of the aph coding region of a mutant mRNA expressed in Streptomyces lividans is relieved when a short leader is adde d upstream of an extra 5'-terminal initiation codon [10].

The fact that most of the 5'-proximal AUG sequences examined were blocked by downstream stop codons is a striking coincidence and raises the further question of what the functions of short, upstream open reading frames might be aside from the possibility that they encode short polypeptides with as yet undiscovered biological functions. They might minimize (via early termination of protein synthesis) the energetic cost of translation from cryptic initiation sites, or ribosomes translating short open reading frames might reduce the access of ribonucleases to the mRNAs which contain these sequences.

Translational attenuation occurs when the translation of one open reading frame alleviates inhibitory secondary structures in a downstream open reading frame. Such a mechanism has been shown to facilitate translation of the bacteriophage MS2 lysis gene in E. coli [11]. Perhaps short, 5'-proximal open reading frames provide similar attenuating functions among the examples cited in this report.

Another possible function for short reading frames beginning at or near the 5' termini of mRNAs is increasing the translational efficiency of downstream coding regions by a reinitiation mechanism. As possible examples of this, short reading frames at the beginnings of the E. coli dapD, Tnl0 xxxP3, and [phi]X174 D transcripts (7, 8, and references cited therein) all terminate at distances of 1 to 17 nucleotides from weak Shine-Dalgarno homologies associated with downstream coding regions. These distances are well within the 46-nucleotide ribosomal scanning range for translational reinitiation proposed by Adhin and van Duin [12] for polycistronic transcripts.

Finally, engaging ribosomes in the synthesis of short polypepties at 5'-proximal open reading frames might provide a concentration effect. Ribosomes momentarily associated with mRNAs in this manner might be more likely to initiate translation at larger coding regions on the same messages due to their physical proximity to those mRNAs after translational termination and subsequent template release.

It should be noted that these explanations represent speculation based on circumstantial evidence. Aside from the start codons for the [lambda] [P.sub.RM], Tnl721 tetR, P2 gene V, and P22 [P.sub.RM] transcripts, it is not known if any of the other AUG sequences mentioned in the results are actually translated. The data and interpretations discussed here do suggest, however, that taking a new look at old sequences could provide a basis for interesting mutational studies on the effects of short, 5'- proximal open reading frames or unblocked AUG triplets on the translational efficiency of known downstream coding regions in prokaryotic messages.

References

(1.) Janssen, G. R. (1993). Eubacterial, archaebacterial, and eucaryotic genes that encode leaderless mRNA. In Industrial Microorganisms: Basic and Applied Molecular Genetics. Baltz, R. H., G. D. Hegeman, and P. L. Skatrud (eds.). Washington, DC: American Society for Microbiology, pp. 59-67.

(2.) Shine, J., and L. Dalgarno (1974). The 3'-terminal sequence of Escherichia coli 16S ribosomal RNA: Complementarity to nonsense triplet and ribosome binding sites. Proceedings of the National Academy of Science USA 71: 1342-1346.

(3.) Gold, L. (1988). Posttranscriptional regulatory mechanisms in Escherichia coli. Annual Reviews in Biochemistry 57: 199-233.

(4.) Wu, C.-J., and G. R. Janssen (1996). Translation of vph mRNA in Streptomyces lividans and Escherichia coli after removal of the 5' untranslated leader. Molecular Microbiology 22: 339-355.

(5.) Van Etten, W., and G. R. Janssen (1997). An AUG initiation codon, not codon-anticodon complementarity, is required for the translation of unleadered mRNA in Escherichia coli. Molecular Microbiology 27: 987-1002.

(6.) Jones III, R. L. (1991). Translational characterization of the leaderless aph message from Streptomyces fradiae. Doctoral dissertation, Indiana University, Bloomington, pp. 130-132.

(7.) Hawley, D. K., and W. R. McClure (1983). Compilation and analysis of Escherichia coli promoter DNA sequences. Nucleic Acids Research 11: 2237-2255.

(8.) Harley, C. B., and R. P. Reynolds (1987). Analysis of E. coli promoter sequences. Nucleic Acids Research 15: 2343-2361.

(9.) Schottel, J. L., J. J. Sninsky, and S. N. Cohen (1984). Effects of alterations in the translational control region on bacterial gene expression: Use of cat gene constructs transcribed from the lac promoter as a model system. Gene 28: 177-193.

(10.) Jones, III, R. L., J. C. Jaskula, and G. R. Janssen (1992). In vivo translational start site selection on leaderless mRNA transcribed from the Streptomyces fradiae aph gene. Journal of Bacteriology 174: 4753-4760.

(11.) Berkhout, B., B. B. Schmidt, A. van Strien, J. van Boom, J. van Westrenen, and J. van Duin (1987). Lysis gene of bacteriophage is activated by translation termination at the overlapping coat gene. Journal of Molecular Biology 195: 517-524.

(12.) Adhin, M. R., and J. van Duin (1990). Scanning model for translational reinitiation in eubacteria. Journal of Molecular Biology 213: 811-818.

              E. coli mRNAs with 5'-proximal AUG triplets
mRNA source (promoter)    5'-proximal AUG category [*]
phage [lambda][P.sub.RM]               1
Tn1721 tetR                            1
phage P2 gene V                        1
argCBH                                 2
argE P2                                2
crp                                    2
dapD                                   2
Tn10xxx P3                             2
pBR322 P4                              2
phage T7A3                             2
phage [phi]X174D                       2
phage [lambda] [P.sub.R]               2
trfB                                   3
cit. util-431                          4
tyrT/77                                4
phage P22 [P.sub.RM]                   1
ampC                                   2
lpp                                    2
tufB                                   2
pBR322 bla                             2
phage [lambda][P.sub.R]'               2
malK                                   3
metA                                   3
(*.)1 = known start codon
2 = blocked by in-frame stop codon
3 = unblocked by in-frame stop codon
4 = undetermined due to incomplete
sequence information