An AUG initiation codon, not codon–anticodon complementarity, is required for the translation of unleadered mRNA in Escherichia coli

We determined the in vivo translational efficiency of 'unleadered' lacZ compared with a conventionally leadered lacZ with and without a Shine–Dalgarno (SD) sequence in Escherichia coli and found that changing the SD sequence of leadered lacZ from the consensus 5'-AGGA-3' to 5'-UUUU-3' results in a 15-fold reduction in translational efficiency; however, removing the leader altogether results in only a twofold reduction. An increase in translation coincident with the removal of the leader lacking a SD sequence suggests the existence of stronger or novel translational signals within the coding sequence in the absence of the leader. We examined, therefore, a change in the translational signals provided by altering the AUG initiation codon to other naturally occurring initiation codons (GUG, UUG, CUG) in the presence and absence of a leader and find that mRNAs lacking leader sequences are dependent upon an AUG initiation codon, whereas leadered mRNAs are not. This suggests that mRNAs lacking leader sequences are either more dependent on perfect codon–anticodon complementarity or require an AUG initiation codon in a sequence-specific manner to form productive initiation complexes. A mutant initiator tRNA with compensating anticodon mutations restored expression of leadered, but not unleadered, mRNAs with UAG start codons, indicating that codon–anticodon complementarity was insufficient for the translation of mRNA lacking leader sequences. These data suggest that a cognate AUG initiation codon specifically serves as a stronger and different translational signal in the absence of an untranslated leader.     

Translation of the Saccharomyces cerevisiae tcm1 gene in the absence of a 5'-untranslated leader.

The role of eukaryotic 5'-untranslated messenger RNA leaders is not entirely clear, since they share little sequence similarity among each other. The importance of the leader in determining the efficiency of translation initiation was addressed here by examining the polyribosome distribution of several leader-deletion alleles of the yeast tcm1 gene (coding for ribosomal protein L3). Shortening of this 22-nucleotide leader, or complete removal of it (the first nucleotide of the mRNA becoming the A of the translation initiation codon AUG) permitted translation, albeit reduced. Further deletion of as few as the first two nucleotides of the initiation codon leads to a substantial reduction in ribosome loading, which is compatible with inefficient initiation at the next downstream, out-of-frame, AUG triplet. A second measure of translation initiation was obtained by assaying qualitatively for the production of biologically active L3 protein using growth-resistance to trichodermin. This experiment indicates that ribosomes can recognize the correct initiation codon even in the complete absence of a leader. We conclude that the 5'-untranslated leader of the yeast tcm1 gene is not essential for accurate translation initiation, but enhances its efficiency.     

Translation of the leaderless Caulobacter dnaX mRNA.

The expression of the Caulobacter crescentus homolog of dnaX, which in Escherichia coli encodes both the gamma and tau subunits of the DNA polymerase III holoenzyme, is subject to cell cycle control. We present evidence that the first amino acid in the predicted DnaX protein corresponds to the first codon in the mRNA transcribed from the dnaX promoter; thus, the ribosome must recognize the mRNA at a site downstream of the start codon in an unusual but not unprecedented fashion. Inserting four bases in front of the AUG at the 5' end of dnaX mRNA abolishes translation in the correct frame. The sequence upstream of the translational start site shows little homology to the canonical Shine-Dalgarno ribosome recognition sequence, but the region downstream of the start codon is complementary to a region of 16S rRNA implicated in downstream box recognition. The region downstream of the dnaX AUG, which is important for efficient translation, exhibits homology with the corresponding region from the Caulobacter hemE gene adjacent to the replication origin. The hemE gene also appears to be translated from a leaderless mRNA. Additionally, as was found for hemE, an upstream untranslated mRNA also extends into the dnaX coding sequence. We propose that translation of leaderless mRNAs may provide a mechanism by which the ribosome can distinguish between productive and nonproductive templates.     

Sequence and structural features associated with translational initiator regions in yeast--a review

We have compared the translational initiator regions of 131 yeast genes. 95% utilize the first AUG from the 5' end of the message as the start codon for translation. Yeast leader regions in general are rich in adenine nucleotides (nt), have an average length of 52 nt, and are void of significant secondary structure. Sequences immediately adjacent to AUG start codons are preferred, however, the bias in nucleotide distribution (5'-A-YAA-UAAUGUCU-3') does not reflect a higher eukaryotic consensus (5'-CACCAUGG-3') with the exception of an adenine nucleotide preference at the -3 position. A minority of yeast mRNAs that contain AUG codons in the leader region that do not serve as the start codon for the primary gene product differ from the majority of mRNAs by one or more of these general properties. This analysis appears to indicate that basic features associated with yeast leader regions are consistent with a general mechanism of initiation of protein synthesis in eukaryotes, as proposed by the ribosomal 'scanning' model, but perhaps only basic features associated with ribosomal recognition of an AUG start codon are intact.     

Mutational analysis of upstream AUG codons of poliovirus RNA.

The 5' untranslated region of poliovirus type 2 Lansing RNA consists of 744 nucleotides containing seven AUG codons which are followed by in-frame termination codons, thus forming short open reading frames (ORFs). To determine the biological significance of these small ORFs, all of the upstream AUG codons were mutated to UUG. The point mutations were introduced into an infectious poliovirus cDNA clone, and RNA transcribed in vitro from the altered cDNA was transfected into HeLa cells to recover the virus. Mutation of AUG 7 resulted in a virus (called R2-5NC-14) with a small-plaque phenotype, whereas mutation of the other six AUG codons produced virus with a wild-type plaque morphology. To determine whether the small-plaque phenotype of R2-5NC-14 was due to altered translational efficiency of the viral mRNA, we constructed chimeric mRNAs containing the 5' noncoding region of poliovirus mRNA fused to the chloramphenicol acetyltransferase (CAT) coding sequence. mRNA containing a mutated AUG 7 codon showed decreased translational efficiency in vitro. The results indicate that the upstream ORFs of poliovirus RNA are not essential for viral replication and do not act as barriers to the translation of poliovirus mRNA. AUG 7 and flanking sequences may play a positive acting role in poliovirus RNA translation.     

Control of start codon choice on a plant viral RNA encoding overlapping genes.

The signals that control initiation of translation in plants are not well understood. To dissect some of these signals, we used a plant viral mRNA on which protein synthesis initiates at two out-of-frame start codons. On the large subgenomic RNA (sgRNA1) of barley yellow dwarf virus-PAV serotype, the coat protein (CP) and overlapping 17K open reading frames (ORFs) are translated beginning at the first and second AUG codons, respectively. The roles of bases at positions -3 and +4 relative to the AUG codons in efficiency of translation initiation were investigated by translation of sgRNA1 mutants in a cell-free extract and by expression of a reporter gene from mutant sgRNA1 leaders in protoplasts. The effects of mutations that disrupted and restored secondary structure encompassing the CP AUG independently of, and in combination with, changes to bases -3 and +4 were also examined. Partial digestion of the 5' end of the sgRNA1 leader with structure-sensitive nucleases gave products that were consistent with the predicted secondary structure. Secondary structure had an overall inhibitory effect on translation of both ORFs. In general, the "Kozak rules" of start codon preference predominate in determining start codon choice. Unexpectedly, for a given CP AUG sequence context, changes that decreased initiation at the downstream 17K AUG also reduced initiation at the CP AUG. To explain this observation, we propose a new model in which pausing of the ribosome at the second AUG allows increased initiation at the first AUG. This detailed analysis of the roles of primary and secondary structure in controlling translation initiation should be of value for understanding expression of any plant gene and in the design of artificial constructs.     

Analysis of the Saccharomyces cerevisiae mitochondrial COX3 mRNA 5' untranslated leader: translational activation and mRNA processing.

We used transformation of yeast mitochondria and homologous gene replacement to study features of the 613-base COX3 mRNA 5' untranslated leader (5'-UTL) required for translational activation by the protein products of the nuclear genes PET54, PET122, and PET494 in vivo. Elimination of the single AUG triplet in the 5'-UTL had no detectable effect on expression, indicating that activator proteins do not work by allowing ribosomes to bypass that AUG. Deletion of the entire 5'-UTL completely prevented translation, suggesting that the activator proteins do not function by antagonizing any other negative element in the 5'-UTL. Removal of the 15 terminal bases from the 5' end of the 5'-UTL did not block activator-dependent translation. The largest internal deletion that did not interfere with translation removed 125 bases from the upstream portion of the leader. However, two large deletions that blocked translation could be reverted to activator-dependent expression by secondary changes in the remaining 5'-UTL sequences, indicating that the original deletions had not removed the translational activator target but only deformed it. Taken together, the deletion mutations and revertants define a region of 151 bases (between positions -480 and -330 relative to the start codon) containing sequences that are sufficient for translational activation when modified slightly. Suppression of the respiratory phenotypes of two 5'-UTL mutations by overexpression of PET54, PET122, and PET494 indicated functional interactions between the leader and the three activator proteins. The mature COX3 mRNA is cleaved from a precursor immediately downstream of the preceding tRNAVal in a fashion resembling mRNA processing in vertebrate mitochondria. Our results indicate that the site of this cleavage in Saccharomyces cerevisiae is determined solely by the position of the tRNA.     

Alternative translation of human fibroblast growth factor 2 mRNA occurs by internal entry of ribosomes.

Alternative initiations of translation of the human fibroblast growth factor 2 (FGF-2) mRNA, at three CUG start codons and one AUG start codon, result in the synthesis of four isoforms of FGF-2. This process has important consequences on the fate of FGF-2: the CUG-initiated products are nuclear and their constitutive expression is able to induce cell immortalization, whereas the AUG-initiated product, mostly cytoplasmic, can generate cell transformation. Thus, the different isoforms probably have distinct targets in the cell. We show here that translation initiation of the FGF-2 mRNA breaks the rule of the cap-dependent ribosome scanning mechanism. First, translation of the FGF-2 mRNA was shown to be cap independent in vitro. This cap-independent translation required a sequence located between nucleotides (nt) 192 and 256 from the 5' end of the 318-nt-long 5' untranslated region. Second, expression of bicistronic vectors in COS-7 cells indicated that the FGF-2 mRNA is translated through a process of internal ribosome entry mediated by the mRNA leader sequence. By introducing additional AUG codons into the RNA leader sequence, we localized an internal ribosome entry site to between nt 154 and 318 of the 5' untranslated region, just upstream of the first CUG. The presence of an internal ribosome entry site in the FGF-2 mRNA suggests that the process of internal translation initiation, by controlling the expression of a growth factor, could have a crucial role in the control of cell proliferation and differentiation.     

Mutational analysis of the HIS4 translational initiator region in Saccharomyces cerevisiae.

We have mutated various features of the 5' noncoding region of the HIS4 mRNA in light of established Saccharomyces cerevisiae and mammalian consensus translational initiator regions. Our analysis indicates that insertion mutations that introduce G + C-rich sequences in the leader, particularly those that result in stable stem-loop structures in the 5' noncoding region of the HIS4 message, severely affect translation initiation. Mutations that alter the length of the HIS4 leader from 115 to 39 nucleotides had no effect on expression, and sequence context changes both 5' and 3' to the HIS4 AUG start codon resulted in no more than a twofold decrease of expression. Changing the normal context at HIS4 5'-AAUAAUGG-3' to the optimal sequence context proposed for mammalian initiator regions 5'-CACCAUGG-3' did not result in stimulation of HIS4 expression. These studies, in conjunction with comparative and genetic studies in S. cerevisiae, support a general mechanism of initiation of protein synthesis as proposed by the ribosomal scanning model.     

Factors governing the expression of a bacterial gene in mammalian cells.

Cultured monkey kidney cells transfected with simian virus 40 (SV40)-pBR322-derived deoxyribonucleic acid (DNA) vectors containing the Escherichia coli gene (Ecogpt, or gpt) coding for the enzyme xanthine-guanine phosphoribosyltransferase (XGPRT) synthesize the bacterial enzyme. This paper describes the structure of the messenger ribonucleic acids (mRNA's) formed during the expression of gpt and an unexpected feature of the nucleotide sequence in the gpt DNA segment. Analyses of the gpt-specific mRNA's produced during infection of CV1 cells indicate that in addition to the mRNA's expected on the basis of known simian virus 40 RNA splicing patterns, there is a novel SV40-gpt hybrid mRNA. The novel mRNA contains an SV40 leader segment spliced to RNA sequences transcribed from the bacterial DNA segment. The sequence of the 5'-proximal 345 nucleotides of the gpt DNA segment indicates that the only open translation phase begins with an AUG about 200 nucleotides from the end of the gpt DNA. Two additional AUGs as well as translation terminator codons in all three phases precede the XGPRT initiator codon. Deletion of the two that are upstream of the putative start codon increases the level of XGPRT production in transfected cells; deletion of sequences that contain the proposed XGPRT initiator AUG abolishes enzyme production. Based on the location of the XGPRT coding sequence in the recombinants and the structure of the mRNA's, we infer that the bacterial enzyme can be translated from an initiator AUG that is 400 to 800 nucleotides from the 5' terminus of the mRNA and preceded by two to six AUG triplets.     

The Escherichia coli threonyl-tRNA synthetase gene contains a split ribosomal binding site interrupted by a hairpin structure that is essential for autoregulation

The expression of the gene encoding Escherichia coli threonyl-tRNA synthetase (ThrRS) is negatively autoregulated at the translational level. ThrRS binds to its own mRNA leader, which consists of four structural and functional domains: the Shine–Dalgarno (SD) sequence and the initiation codon region (domain 1); two upstream hairpins (domains 2 and 4) connected by a single-stranded region (domain 3). Using a combination of in vivo and in vitro approaches, we show here that the ribosome binds to thrS mRNA at two non-contiguous sites: region −12 to +16 comprising the SD sequence and the AUG codon and, unexpectedly, an upstream single-stranded sequence in domain 3. These two regions are brought into close proximity by a 38-nucleotide-long hairpin structure (domain 2). This domain, although adjacent to the 5' edge of the SD sequence, does not inhibit ribosome binding as long as the single-stranded region of domain 3 is present. A stretch of unpaired nucleotides in domain 3, but not a specific sequence, is required for efficient translation. As the repressor and the ribosome bind to interspersed domains, the competition between ThrRS and ribosome for thrS mRNA binding can be explained by steric hindrance.     

The actin gene in yeast Saccharomyces cerevisiae: 5'' and 3'' end mapping, flanking and putative regulatory sequences

The 5' and 3' flanking regions of the yeast actin gene have been sequenced and the ends of the actin mRNA were determined by the single-strand nuclease mapping procedure. The mRNA starts with a pyrimidine residue 141 (or 140) nucleotides upstream from the initiation codon. The actin gene lacks a typical "TATA" box 30 base pairs upstream from the mRNA start site but it contains a region homologous to the canonical sequence 5'-GGCTCAATCT-3' which is found in several eukaryotic genes 70 to 80 bp upstream from the mRNA cap site. Judging from the S1 nuclease mapping, there are two populations of actin mRNA terminating 98 and 107 nucleotides downstream from the stop codon. The 3' termini are preceded by three AATAAA sequences found in most eukaryotic polyadenylated mRNAs.