Selective stimulation of translation of leaderless mRNA by initiation factor 2: evolutionary implications for translation

Translation initiation in bacteria involves a stochastic binding mechanism in which the 30S ribosomal subunit first binds either to mRNA or to initiator tRNA, fMet-tRNA(f)(Met). Leaderless lambda cI mRNA did not form a binary complex with 30S ribosomes, which argues against the view that ribosomal recruitment signals other than a 5'-terminal start codon are essential for translation initiation of these mRNAs. We show that, in Escherichia coli, translation initiation factor 2 (IF2) selectively stimulates translation of lambda cI mRNA in vivo and in vitro. These experiments suggest that the start codon of leaderless mRNAs is recognized by a 30S-fMet-tRNA(f)(Met)-IF2 complex, an intermediate equivalent to that obligatorily formed during translation initiation in eukaryotes. We further show that leaderless lambda cI mRNA is faithfully translated in vitro in both archaebacterial and eukaryotic translation systems. This suggests that translation of leaderless mRNAs reflects a fundamental capability of the translational apparatus of all three domains of life and lends support to the hypothesis that the translation initiation pathway is universally conserved.     

Isolation and characterization of ribosomes and translation initiation factors from the gram-positive soil bacterium Streptomyces lividans

A primer extension inhibition (toeprint) assay was developed using ribosomes and ribosomal subunits from Streptomyces lividans. This assay allowed the study of ribosome binding to streptomycete leaderless and leadered mRNA. Purified 30S subunits were unable to form a ternary complex on aph leaderless mRNA, whereas 70S ribosomes could form ternary complexes on this mRNA. 30S subunits formed ternary complexes on leadered aph and malE mRNA. The translation initiation factors (IF1, IF2, and IF3) from S. lividans were isolated and included in toeprint and filter binding assays with leadered and leaderless mRNA. Generally, the IFs reduced the toeprint signal on leadered mRNA; however, incubation of IF1 and IF2 with 30S subunits that had been washed under high-salt conditions promoted the formation of a ternary complex on aph leaderless mRNA. Our data suggest that, as reported for Escherichia coli, initiation complexes with leaderless mRNAs might use a novel pathway involving 70S ribosomes or 30S subunits bound by IF1 and IF2 but not IF3. Some mRNA-ribosome-initiator tRNA reactions that yielded weak or no toeprint signals still formed complexes in filter binding assays, suggesting the occurrence of interactions that are not stable in the toeprint assay.     

The translational fidelity function of IF3 during transition from the 30 S initiation complex to the 70 S initiation complex

IF3 has a fidelity function in the initiation of translation, inducing the dissociation of fMet-tRNA(fMet) from the 30 S initiation complexes (30SIC) containing a non-canonical initiation triplet (e.g. AUU) in place of a canonical initiation triplet (e.g., AUG). IF2 has a complementary role, selectively promoting initiator tRNA binding to the ribosome. Here, we used parallel rapid kinetics measurements of GTP hydrolysis, Pi release, light-scattering, and changes in intensities of fluorophore-labeled IF2 and fMet-tRNA(fMet) to determine the effects on both 30SIC formation and 30SIC conversion to 70 S initiation complexes (70SIC) of (a) substituting AUG with AUU, and/or (b) omitting IF3, and/or (c) replacing GTP with the non-hydrolyzable analog GDPCP. We demonstrate that the presence or absence of IF3 has, at most, minor effects on the rate of 30SIC formation using either AUG or AUU as the initiation codon, and conclude that the high affinity of IF2 for both 30 S subunit and initiator tRNA overrides any perturbation of the codon-anticodon interaction resulting from AUU for AUG substitution. In contrast, replacement of AUG by AUU leads to a dramatic reduction in the rate of 70SIC formation from 30SIC upon addition of 50 S subunits. Interpreting our results in the framework of a quantitative kinetic scheme leads to the conclusion that, within the overall process of 70SIC formation, the step most affected by substituting AUU for AUG involves the conversion of an initially labile 70 S ribosome into a more stable complex. In the absence of IF3, the difference between AUG and AUU largely disappears, with each initiation codon affording rapid 70SIC formation, leading to the hypothesis that it is the rate of IF3 dissociation from the 70 S ribosome during IC70S formation that is critical to its fidelity function.     

Identification of features in 5' terminal fragments from reovirus mRNA which are important for ribosome binding

Four types of experiments were carried out with reovirus messenger RNAs or with 5′ terminal fragments of known sequence to identify features in mRNA which appear to be important for formation of initiation complexes with ribosomes. With a number of reovirus mRNAs, 40S initiation complexes had been previously shown to protect a significantly larger segment of the RNA (including the 5′ terminal m7G) than that protected by 80S initiation complexes. Each 80S-protected sequence had an AUG codon and was a subset of the 40S-protected sequence from the same message. When 40S- and 80S-protected fragments were tested for ability to rebind to ribosomes, the 80S-protected fragments showed considerably lower binding ability, implying that the “extra” sequences protected by 40S initiation complexes contribute to ribosome attachment. Nevertheless, wheat germ ribosomes select the same 5′ terminal initiation site in each reovirus mRNA, irrespective of the presence or absence of m7G on the message. This was demonstrated by comparing fingerprints of the ribosome-protected regions obtained with methylated versus unmethylated RNA. The contribution of m7G to formation of initiation complexes is therefore quantitative rather than qualitative. Limited T1 RNAase digestion of isolated 5′ terminal fragments from several reovirus messages generated a series of smaller fragments which were analyzed for ability to rebind to ribosomes. Partial digestion products up to 30 nucleotides in length which retained the 5′ cap but not the AUG codon were unable to associate stably with ribosomes, whereas every AUG-containing fragment that was analyzed was able to form initiation complexes. The efficiency of binding of certain AUG-containing fragments, however, was reduced by removal of either the 5′ terminal region, including the cap, or of sequences comprising the beginning of the coding region, on the 3′ side of the AUG. Complex formation between messenger RNA and ribosomes was inhibited by the trinucleotide AUG, but not by various other oligonucleotides. Although the inhibition was specific, a vast excess of trinucleotide was required for moderate inhibition of 80S complex formation, and the same concentration of AUG failed to inhibit formation of 40S initiation complexes.     

Initiation of Escherichia coli ribosomes on matrix coupled mRNAs studied by optical biosensor technique

The optical biosensor technique, based on the surface plasmon resonance (SPR) phenomenon, has been used to study the initiation of protein synthesis by E. coli ribosomes on surface coupled mRNA. mRNA was first periodate oxidized and then hydrazide coupled to the surface of a CM5 sensor chip. The formation of initiation complexes on the surface coupled mRNA was monitored in real-time with a BIACORE 2000 instrument. Mature 70S*mRNA*fMet-tRNA(Met) initiation complexes were assembled on mRNA by sequential introduction of the 30S and 50S subunits supplemented with appropriate initiation factors and fMet-tRNA(Met). We show that the formation of 70S*mRNA complexes on the surface coupled mRNA proceeds efficiently only in the presence of tRNA. Moreover, 70S*mRNA*fMet-tRNA(Met) complexes formed with fMet-tRNA(Met) are more stable than similar complexes formed with deacylated tRNAs. The efficient formation and slow dissociation of mature 70S*mRNA*fMet-tRNA(Met) initiation complexes are most easily explained by the stabilization of the interaction of the ribosomal subunits by fMet-tRNA(Met). This work demonstrates the feasibility of the BIACORE technique for studying the initiation of protein synthesis.     

Overview: mechanism of translation initiation in eukaryotes

Evidence to date has placed considerable emphasis on protein synthesis initiation as the dominant site of translational control. Two specific aspects are regulated, the binding of the initiator tRNA to the 40S subunits (as a ternary complex with eIF-2 and GTP) and the subsequent binding of mRNA to the complex of the 40S subunit with initiator tRNA. In addition to regulation, eIF-2 and Met-tRNAf are in large part responsible for selection of the initiating AUG codon. The utilization of most host mRNAs requires an m7G cap structure at the 5' end. However, many viral systems appear to use one of two alternate initiation schemes referred to as re-initiation and internal initiation. The function of specific initiation factors is presented and the consequences of altering the activity of these factors is discussed.     

Binding of wheat germ ribosomes to bisulfite-modified reovirus messenger RNA: Evidence for a scanning mechanism

A scanning mechanism has been proposed (Kozak, 1978) to explain how eukaryotic ribosomes select the correct AUG codon for initiation of protein synthesis. The hypothesis is that a 40 S ribosomal subunit binds initially at or near the 5′-terminus of a message and subsequently migrates toward the interior of the messenger RNA, stopping when it encounters the first AUG codon, at which point a 60 S subunit joins and peptide bond formation begins. The scanning mechanism predicts that if a message were modified by introduction of a new AUG triplet upstream of the existing initiator codon, the adventitious AUG should be the preferred site for formation of an 80 S initiation complex. This prediction has been confirmed in the present studies with two reovirus messenger RNAs, in which sodium bisulfite was used to convert an ACG sequence (located in the 5′ untranslated region of each message) to AUG. Analysis of the ribosome-protected mRNA fragments recovered from sparsomycin-blocked 80 S initiation complexes revealed that a high percentage of wheat germ ribosomes were centered around the “unnatural” 5′-proximal AUG created by the bisulfite treatment, although some ribosomes were also positioned at the second (normal) initiator codon. The bisulfite modification was carried out in 7 m-urea at 37 °C. resulting in quantitative conversion of cytosine to uracil. Thus, both the primary and secondary structure of the message were drastically altered. These perturbations did not impair the efficiency of ribosome binding, nor did the highly unfolded state of the mRNA permit ribosomes to attach to spurious sites in the interior of the message. The data support a mechanism in which the initiator codon is selected by virtue of its position in a message (i.e. closest to the 5′-terminus), without regard to either the primary or secondary structure of the flanking regions.     

Interaction of the Escherichia coli fdhF mRNA hairpin promoting selenocysteine incorporation with the ribosome.

The codon UGA located 5' adjacent to an mRNA hairpin within fdhF mRNA promotes the incorporation of the amino acid selenocysteine into formate dehydrogenase H of Escherichia coli. The loop region of this mRNA hairpin has been shown to bind to the special elongation factor SELB, which also forms a complex with selenocysteinyl-tRNA(Sec) and GTP. We designed seven different mRNA constructs derived from the fdhF mRNA which contain a translation initiation region including an AUG initiation codon followed by no, one, two, three, four, five or six UUC phenylalanine codon(s) and the UGA selenocysteine codon 5' adjacent to the fdhF mRNA hairpin. By binding these different mRNA constructs to 30S ribosomal subunits in vitro we attempted to mimic intermediate steps of elongation of a structured mRNA approaching the ribosome by one codon at a time. Toeprint analysis of the mRNA-ribosome complexes showed that the presence of the fdhF mRNA hairpin strongly interferes with binding of the fdhF mRNA to 30S ribosomal subunits as soon as the hairpin is placed closer than 16 bases to the ribosomal P-site. Binding is reduced up to 25-fold compared with mRNA constructs where the hairpin is located outside the ribosomal mRNA track. Surprisingly, no toeprint signals were observed in any of our mRNA constructs when tRNA(Sec) was used instead of tRNA(fMet). Lack of binding of selenocysteinyl-tRNA(Sec) to the UGA codon was attributed to steric hindrance by the fdhF mRNA hairpin. By chemical probing of the shortest mRNA construct (AUG-UGA-fdhF hairpin) bound to 30S ribosomal subunits we demonstrate that the hairpin structure is not unfolded in the presence of ribosomes in vitro; also, this mRNA is not translated in vivo when fused in-frame 5' of the lacZ gene. Therefore, our data indicate that the fdhF mRNA hairpin has to be unfolded during elongation prior to entering the ribosomal mRNA track and we propose that the SELB binding domain within the fdhF mRNA is located outside the ribosomal mRNA track during decoding of the UGA selenocysteine codon by the SELB-selenocysteinyl-tRNA(Sec)-GTP complex.     

Ribosomes bind leaderless mRNA in Escherichia coli through recognition of their 5'-terminal AUG

Leaderless mRNAs are translated in the absence of upstream signals that normally contribute to ribosome binding and translation efficiency. In order to identify ribosomal components that interact with leaderless mRNA, a fragment of leaderless cI mRNA from bacteriophage λ, with a 4-thiouridine (4^S-U) substituted at the +2 position of the AUG start codon, was used to form cross-links to Escherichia coli ribosomes during binary (mRNA+ribosome) and ternary (mRNA+ribosome+initiator tRNA) complex formation. Ribosome binding assays (i.e., toeprints) demonstrated tRNA-dependent binding of leaderless mRNA to ribosomes; however, cross-links between the start codon and 30S subunit rRNA and r-proteins formed independent of initiator tRNA. Toeprints revealed that a leaderless mRNA's 5′-AUG is required for stable binding. Furthermore, the addition of a 5′-terminal AUG triplet to a random RNA fragment can make it both competent and competitive for ribosome binding, suggesting that a leaderless mRNA's start codon is a major feature for ribosome interaction. Cross-linking assays indicate that a subset of 30S subunit r-proteins, located at either end of the mRNA tunnel, contribute to tRNA-independent contacts and/or interactions with a leaderless mRNA's start codon. The interaction of leaderless mRNA with ribosomes may reveal features of mRNA binding and AUG recognition that are distinct from known signals but are important for translation initiation of all mRNAs.     

Footprinting mRNA-ribosome complexes with chemical probes.

We footprinted the interaction of model mRNAs with 30S ribosomal subunits in the presence or absence of tRNA(fMet) or tRNA(Phe) using chemical probes directed at the sugar-phosphate backbone or bases of the mRNAs. When bound to the 30S subunits in the presence of tRNA(fMet), the sugar-phosphate backbones of gene 32 mRNA and 022 mRNA are protected from hydroxyl radical attack within a region of about 54 nucleotides bounded by positions -35 (+/- 2) and +19, extending to position +22 when tRNA(Phe) is used. In 70S ribosomes, protection is extended in the 5' direction to about position -39 (+/- 2). In the absence of tRNA, the 30S subunit protects only nucleotides -35 (+/- 2) to +5. Introduction of a stable tetraloop hairpin between positions +10 and +11 of gene 32 mRNA does not interfere with tRNA(fMet)-dependent binding of the mRNA to 30S subunits, but results in loss of protection of the sugar-phosphate backbone of the mRNA downstream of position +5. Using base-specific probes, we find that the Shine-Dalgarno sequence (A-12, A-11, G-10 and G-9) and the initiation codon (A+1, U+2 and G+3) of gene 32 mRNA are strongly protected by 30S subunits in the presence of initiator tRNA. In the presence of tRNA(Phe), the same Shine-Dalgarno bases are protected, as are U+4, U+5 and U+6 of the phenylalanine codon. Interestingly, A-1, immediately preceding the initiation codon, is protected in the complex with 30S subunits and initiator tRNA, while U+2 and G+3 are protected in the complex with tRNA(Phe) in the absence of initiator tRNA. Additionally, specific bases upstream from the Shine-Dalgarno region (U-33, G-32 and U-22) as well as 3' to the initiation codon (G+11) are protected by 30S subunits in the presence of either tRNA. These results imply that the mRNA binding site of the 30S subunit covers about 54-57 nucleotides and are consistent with the possibility that the ribosome interacts with mRNA along its sugar-phosphate backbone.     

The initiation codon affects ribosome binding and translational efficiency in Escherichia coli of cI mRNA with or without the 5' untranslated leader

Translational efficiency of an AUG, CUG, GUG, or UUG initiation codon was measured for the naturally leaderless cI mRNA from bacteriophage lambda. In a cI-lacZ translational fusion, only AUG supported a high level of expression; GUG supported a low level of expression, while UUG and CUG expression was barely above background levels. Addition of an untranslated lac leader and Shine-Dalgarno sequence to cI increased expression but still showed a dependence on an AUG for maximum expression. cI-lacZ mRNA with an AUG initiation codon showed a greater in vitro ribosome binding strength and a higher level of full-length in vivo mRNA, suggesting that the initiation codon is an important determinant of ribosome binding strength and translational efficiency for mRNA with or without the 5' untranslated leader.     

The effect of oligonucleotides complementary to the 3' end of the 16S rRNA on the formation of the bacterial 30S initiation complex

Although much attention was focused on the role of the 16S RNA in mRNA selection by the 30S ribosomal subunit no true consensus site has emerged as yet. Oligonucleotides such as GAGG, UGAU and CCAA which are complementary to the 3' end of the 16S RNA stimulate the AUG-dependent binding of fMet-tRNA to 30S subunits. If those tetranucleotides are used in combination or if the octanucleotides GAGGUGAU and UGAUCCAA are applied, the degree of stimulation remains unchanged. Effects are strictly dependent on the presence of initiation factor 2 (IF-2) and cannot be produced by using A4 or U4. With sequences covalently linked to the AUG as in CCAAAUG and UGAUCCAAAUG, the efficiency of the initiation complex formation decreases significantly as compared to AUG with UGAUCCAAAUG being the least efficient mRNA analogue. The pentadecanucleotide GAGGUGAUCCAAAUG, however, shows the highest efficiency in directing the binding of the fMet-tRNA to 30S subunits and is clearly superior to AUG. Initiation factor 2 (IF-2), which stimulates tRNA binding significantly with AUG and CCAAAUG, both in terms of slope and plateau values of the binding curves, does not effect the initial rate of tRNA binding to GAGGUGAUCCAAAUG. In another set of experiments, where GAGG and AUG are separated by oligo(U) or oligo(A) sequences, the effect of chain length was investigated. mRNA analogues with a spacer of 6-9 nucleotides show the highest binding efficiencies, with a U spacer being superior to an A spacer, indicating that a more flexible spacer favours tRNA binding.     

A quantitative kinetic scheme for 70 S translation initiation complex formation

Association of the 30 S initiation complex (30SIC) and the 50 S ribosomal subunit, leading to formation of the 70 S initiation complex (70SIC), is a critical step of the translation initiation pathway. The 70SIC contains initiator tRNA, fMet-tRNA(fMet), bound in the P (peptidyl)-site in response to the AUG start codon. We have formulated a quantitative kinetic scheme for the formation of an active 70SIC from 30SIC and 50 S subunits on the basis of parallel rapid kinetics measurements of GTP hydrolysis, Pi release, light-scattering, and changes in fluorescence intensities of fluorophore-labeled IF2 and fMet-tRNA(f)(Met). According to this scheme, an initially formed labile 70 S complex, which promotes rapid IF2-dependent GTP hydrolysis, either dissociates reversibly into 30 S and 50 S subunits or is converted to a more stable form, leading to 70SIC formation. The latter process takes place with intervening conformational changes of ribosome-bound IF2 and fMet-tRNA(fMet), which are monitored by spectral changes of fluorescent derivatives of IF2 and fMet-tRNA(fMet). The availability of such a scheme provides a useful framework for precisely elucidating the mechanisms by which substituting the non-hydrolyzable analog GDPCP for GTP or adding thiostrepton inhibit formation of a productive 70SIC. GDPCP does not affect stable 70 S formation, but perturbs fMet-tRNA(fMet) positioning in the P-site. In contrast, thiostrepton severely retards stable 70 S formation, but allows normal binding of fMet-tRNA(fMet)(prf20) to the P-site.     

tRNA binding sites on the subunits of Escherichia coli ribosomes

Programmed 30 S subunits expose only one binding site, to which the different classes of tRNA (deacylated tRNAPhe, Phe-tRNAPhe, and N-acetylphenylalanyl (AcPhe)-tRNAPhe) bind with about the same affinity. Elongation factor Tu within the ternary complex does not contribute to the binding of Phe-tRNA. Binding of acylated or deacylated tRNA to 30 S depends on the cognate codon; nonprogrammed 30 S subunits do not bind tRNA to any significant extent. The existence of only one binding site/30 S subunit (and not, for example, two sites in 50% of the subunits) could be shown with Phe-tRNAPhe as well as deacylated tRNAPhe pursuing different strategies. Upon 50 S association the 30 S-bound tRNA appears in the P site (except the ternary complex which is found at the A site). Inhibition experiments with tetracycline demonstrated that the 30 S inhibition pattern is identical to that of the P site but differs from that of the A site of 70 S ribosomes. In contrast to 30 S subunits the 50 S subunit exclusively binds up to 0.2 and 0.4 molecules of deacylated tRNAPhe/50 S subunit in the absence and presence of poly(U), respectively, but neither Phe-tRNA nor AcPhe-tRNA. Noncognate poly(A) did not stimulate the binding indicating codon-anticodon interaction at the 50 S site. The exclusive binding of deacylated tRNA and its dependence on the presence of cognate mRNA is reminiscent of the characteristics of the E site on 70 S ribosomes. 30 and 50 S subunits in one test tube expose one binding site more than the sum of binding capacities of the individual subunits. The results suggest that the small subunit contains the prospective P site and the large subunit the prospective E site, thus implying that the A site is generated upon 30 S-50 S association.     

Preparation and characterization of eukaryotic initiation factor EIF-3. Formation of binary (EIF-3-Met-tRNAf) and ternary (EIF-3-Met-tRNAf-GTP) complexes.

The 133,000 X g supernatant fraction prepared from ascites cells in 20 mM KCl (low CKl supernatant) contained the initiation factors EIF-1 and EIF-2 (and the elongation factore EF-1 and EF-2) but lacked EIF-3; thus, low KCl supernatant could be used to assay for EIF-3. EIF-3 was prepared from a crude initiation factor perparation (a 250 mM KCl extract of ascites cell ribosomes precipitated with 70% saturated ammonium sulfate) by chromatography on DEAE-Sephadex A-50 and hydroxylapatite. The EIF-O had no detectable EIF-1 and little or no EIF-2. Factor EIF-3 was required fro translation of encephalomyocarditis virus RNA. The molecular weight of EIF-3 was estimated by Sephadex G-200 filtration to be 139,000; the sedimentation coefficient was calculated to be about 5.8. EIF-3 formed a binary complex specifically with the initiator tRNA, Met-tRNAf, and if GTP was present the factor formed a ternary complex (EIF-3-Met-tRNAf-GTP). The EIF-3 preparation had no methionyl-tRNA synthetase activity to account for binding. Complex-formation was with eukaryotic Met-tRNAf and no other aminoacyl-tRNA. The binary and ternary complexes were retained quantitatively on Millipore filters (which was the most convenient assay), but they could also be demonstrated by filtration through Sephadex G-100 or by glycerol gradient centrifugation. GTP increased the rate, the amount, and the stability of complex formed; the ration of GTP to Met-tRNAf in the ternary complex appeared to be 1. The binary and the ternary complexes transferred Met-tRNAf to the 40 S ribosomal subunits, but not to 60 S subparticles. The factor-dependent binding of Met-tRNAf to the 40 S subunit did not require mRNA (or GTP). In the presence of 60 S subunits, the initiator tRNA bound to 40 S subunits was not transferred to 80 S ribosomes even if mRNA was added--that reaction may require another initiation factor. Treatment of EIF-3 with N-ethylmaleimide led to loss of its activity in complex formation and in support of the translation of encephalomyocarditis virus RNA. In addition to forming the binary and ternary complexes, and supporting the translation of encephalomyocarditis virus RNA, EIF-3 also increases the number of free ribosomal subunits by either preventing their association or causing dissociation of 80 S couples.     

Yeast initiator tRNA identity elements cooperate to influence multiple steps of translation initiation

All three kingdoms of life employ two methionine tRNAs, one for translation initiation and the other for insertion of methionines at internal positions within growing polypeptide chains. We have used a reconstituted yeast translation initiation system to explore the interactions of the initiator tRNA with the translation initiation machinery. Our data indicate that in addition to its previously characterized role in binding of the initiator tRNA to eukaryotic initiation factor 2 (eIF2), the initiator-specific A1:U72 base pair at the top of the acceptor stem is important for the binding of the eIF2.GTP.Met-tRNA(i) ternary complex to the 40S ribosomal subunit. We have also shown that the initiator-specific G:C base pairs in the anticodon stem of the initiator tRNA are required for the strong thermodynamic coupling between binding of the ternary complex and mRNA to the ribosome. This coupling reflects interactions that occur within the complex upon recognition of the start codon, suggesting that these initiator-specific G:C pairs influence this step. The effect of these anticodon stem identity elements is influenced by bases in the T loop of the tRNA, suggesting that conformational coupling between the D-loop-T-loop substructure and the anticodon stem of the initiator tRNA may occur during AUG codon selection in the ribosomal P-site, similar to the conformational coupling that occurs in A-site tRNAs engaged in mRNA decoding during the elongation phase of protein synthesis.     

Ribosome-initiator tRNA complex as an intermediate in translation initiation in Escherichia coli revealed by use of mutant initiator tRNAs and specialized ribosomes.

For functional studies of mutant Escherichia coli initiator tRNAs in vivo, we previously described a strategy based on the use of tRNA genes carrying an anticodon sequence change from CAU to CUA along with a mutant chloramphenicol acetyltransferase (CAT) gene carrying an initiation codon change from AUG to UAG. Surprisingly, under conditions where the mutant initiator tRNA is optimally active, the CAT gene with the UAG initiation codon produced more CAT protein (3- to 9-fold more depending on the conditions) than the wild-type CAT gene. Here we show that two new mutant CAT genes having GUC and AUC initiation codons also produce more of the CAT protein in the presence of the corresponding mutant initiator tRNAs. These results are most easily understood if assembly of the 30S ribosome-initiator tRNA-mRNA initiation complex in vivo proceeds with the 30S ribosome binding first to the initiator tRNA and then to the mRNA. In cells overproducing the mutant initiator tRNAs, most ribosomes would carry the mutant initiator tRNA and these ribosomes would select the mutant CAT mRNA over the other mRNAs.