Translational Efficiency of a Non-AUG Initiation Codon Is Significantly Affected by Its Sequence Context in Yeast*

Previous studies have shown that translation of mrna for yeast glycyl-tRNA synthetase is alternatively initiated from UUG and a downstream AUG initiation codon. Evidence presented here shows that unlike an AUG initiation codon, efficiency of this non-AUG initiation codon is significantly affected by its sequence context, in particular the nucleotides at positions –3 to –1 relative to the initiation codon. A/A/R (R represents A Or G) and C/G/C appear to be the most and least favorable sequences at these positions, respectively. Mutation of the native context sequence –3 to –1 from AAA to CGC reduced translation initiation from the UUG codon up to 32-fold and resulted in loss of mitochondrial respiration. although an AUG initiation codon is, in general, unresponsive to context changes in yeast, an AAA (–3 to –1) to CGC mutation still reduced its initiating activity up to 8-fold under similar conditions. these results suggest that sequence context is more important for translation initiation in yeast than previously appreciated.

Aminoacyl-tRNA synthetases are a group of primordial enzymes, each of which catalyzes the attachment of a specific amino acid to its cognate tRNAs. Typically there are 20 aminoacyl-tRNA synthetases in prokaryotes, one for each amino acid (1)(2)(3)(4). In eukaryotes, protein synthesis occurs not only in the cytoplasm but also in organelles such as mitochondria and chloroplasts (5). Thus, eukaryotes such as yeast need two distinct sets of enzymes for each aminoacylation activity, one localized to the cytoplasm and the other localized to the mitochondria. Each set aminoacylates the isoaccepting tRNAs within its respective cell compartment and is sequestered from the isoacceptors confined in other compartments. In most cases, the cytoplasmic and mitochondrial aminoacyl-tRNA synthetase activities are encoded by two distinct nuclear genes, regardless of the cellular compartments to which they are confined. However, two Saccharomyces cerevisiae genes, HTS1 (the gene encoding histidyl-tRNA synthetase) (6) and VAS1 (the gene encoding valyl-tRNA synthetase) (7), specify both the mitochondrial and cytosolic forms through alternative initiation from two in-frame AUG codons. Each of these genes encodes mRNAs with distinct 5Ј-ends. The mitochondrial form of the enzyme is translated from the first AUG on the "long" mRNAs, whereas the cytosolic form is translated from the second AUG on the "short" mRNAs that have their 5Ј-ends located between the first and second AUG codons. Hence, the mitochondrial enzymes have the same polypeptide sequences as their cytosolic counterparts, except for a short amino-terminal mitochondrial targeting sequence. The transit peptide is subsequently cleaved away upon import into the mitochondria. Because the isozymes are targeted to different compartments, the two isoforms of valyl-tRNA synthetase, for example, cannot functionally substitute for each other in vivo (7,8). A similar scenario has been observed for the genes encoding the mitochondrial and cytoplasmic forms of Arabidopsis thaliana alanyl-tRNA synthetase (AlaRS), 2 threonyl-tRNA synthetase, and valyl-tRNA synthetase (9). Although two isoforms of AlaRS are generated in a similar manner in Candida albicans, the longer form per se appears dual-targeted and thus can be functional in both compartments (10). Recently, two rare cases of one gene encoding both activities have been reported in yeast in which the protein isoforms are produced by alternative use of two in-frame initiation codons: an upstream non-AUG initiator and a downstream AUG initiator (11,12).
Translation initiation in eukaryotes is a stringent process requiring not only initiator tRNA but also many protein factors, including eIF1, eIF2, eIF3, eIF4F, and eIF5. Upon binding to the cap structure, the 43 S preinitiation complex, composed of the 40 S ribosome and initiation factors, moves along the mRNA in a 3Ј direction until it encounters the first AUG codon. At this point, GTP hydrolysis that leads to dissociation of the eIF2-GDP complex from the initiator tRNA in the preinitiation complex and subsequent P i release signifies a 3-bp codon/anticodon interaction between Met-tRNA i Met and the start codon (13). It has recently been shown that eIF1 plays a critical role in start codon selection. Mutations of eIF1 may lead to reduced interaction of this initiation factor with 40 S subunits and increased initiation at UUG codons (14). Previous studies on CYC1 (15) in S. cerevisiae suggested that AUG is the only codon recognized as the translational initiator and that the AUG codon nearest the 5Ј-end of an mRNA serves as the start site for translation. If the first AUG codon is mutated, then initiation can begin at the next available AUG from the 5Ј-end of the mRNA. The same * This work was supported by Grant NSC 95-2311-B-008-005 and Grant NSC 95-2311-B-008-006 (to C.-C. W.) from the National Science Council (Taiwan). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. rules apply to all eukaryotes. However, there are many examples in higher eukaryotes, where cellular and viral mRNAs can initiate from non-AUG codons that differ from AUG by just one nucleotide (16). The relatively weak base pairing between a non-AUG codon and the anticodon of an initiator tRNA appears to be compensated for by interactions with nearby nucleotides, in particular a purine (A or G) at position Ϫ3 and a G at position ϩ4 (17,18). Similarly, mutations in the sequence surrounding the first AUG can lead to its inefficient utilization as an initiator and subsequent use of an AUG at a downstream location. In addition to sequence context, a stable hairpin structure located 12-15 nucleotides downstream from the initiator can also facilitate recognition of a weak initiator by the 40 S ribosomal subunit (19). Although some reports suggest that sequences immediately preceding the initiation codon may also play a role in modulating the efficiency of translation initiation in yeast, the magnitude of this context effect appears relatively insignificant (20 -22). Perhaps for that reason, yeast cannot efficiently use non-AUG codons as the translation start site (23,24). Nonetheless, three yeast genes, GRS1 (one of the two glycyl-tRNA synthetase (GlyRS) genes in S. cerevisiae) (11), ALA1 (the only AlaRS gene in S. cerevisiae) (12), and CARP2A (the gene coding for acidic ribosomal protein P2A in C. albicans) (25) have recently been reported to use naturally occurring non-AUG triplets as translation initiators.
In the case of CARP2A, a non-AUG codon, i.e. GUG, serves as the exclusive translation initiator, whereas in the cases of ALA1 and GRS1, non-AUG codons act as alternative translation initiators that are accompanied by a downstream in-frame AUG initiation codon. Although two homologous GlyRS genes, namely GRS1 and GRS2, have been identified in the yeast genome, only one, i.e. GRS1, is functional and provides both cytoplasmic and mitochondrial aminoacylation activities, whereas the other, i.e. GRS2, appears to be dispensable for growth (26). Further studies by site-specific mutagenesis indicated that two functionally exclusive protein isoforms are alternatively generated from GRS1. A short form that is responsible for the cytoplasmic activity is translationally initiated from a classical AUG initiator, whereas a longer isoform that provides the mitochondrial activity is initiated from an upstream in-frame UUG codon (11). Expression of ALA1 follows a similar scenario. The mitochondrial form of AlaRS is initiated from two successive in-frame ACG triplets 69 nucleotides upstream of the AUG initiator of the cytoplasmic form (12). This feature of redundancy of ACG initiators may in itself represent a novel mechanism to improve the efficiency of a poor initiation event (27). However, in general the detailed molecular mechanism that enables the ACG (in ALA1) and UUG (in GRS1) codons to serve as translation initiation sites remains poorly understood. In particular, if other factors such as sequence context do not play a role in recognition of non-AUG initiators, the initiating ribosome should not be able to differentiate an authentic non-AUG initiation codon from other random non-AUG codons. In the work described here, we looked for the most and least favorable sequence contexts for the UUG initiator of GRS1 and compared the effects of these contexts on the efficiency of translation initiation. Contrary to many previous reports on context effect, our results argued that sequence context, in particular the nucleotides at relative position Ϫ3 to Ϫ1, is very important to the efficiency of the non-AUG initiator in GRS1. Mutation of the nucleotides at its relative position Ϫ3 to Ϫ1 from AAA to CGC reduced its initiating activity up to 32-fold. As a result, a GRS1 construct carrying this mutation was unable to maintain mitochondrial function because of drastically lowered protein levels. Although AUG was relatively insensitive to context changes under similar conditions, an AAA (Ϫ3 to Ϫ1) to CGC mutation still reduced its initiating activity up to 8-fold, suggesting that an AUG initiation codon could be subject to the same rule of modulation, albeit to a lesser extent. In this sense, a non-AUG initiator could perhaps serve as a better model in identifying the sequence and structural elements important for translation initiation in yeast.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-Cloning of GRS1 from S. cerevisiae into pRS315, a low copy number yeast shuttle vector (28), and pADH, a high copy number yeast shuttle vector with a constitutive ADH promoter, has previously been described (11). Construction of various GRS1-lexA fusions for Western blot analysis was as previously described (11,27). To clone a wild-type lexA sequence in pADH, the open reading frame of lexA was amplified by PCR as an NdeI-XhoI fragment and cloned in the appropriate sites in pADH. To clone an initiator mutant of lexA (designated here as lexA*), the NdeI site was mutated by site-directed mutagenesis to a SpeI site. Therefore, in the wild-type lexA clone, the ATG initiator of lexA is part of the NdeI site, whereas in lexA*, the native ATG initiator is mutated to ACT and becomes part of the SpeI site. A wild-type or mutant GRS1 sequence containing base pairs Ϫ88 to Ϫ12 relative to ATG1 was amplified by PCR as a PstI-NdeI or PstI-SpeI fragment and cloned in-frame into the 5Ј-end of the wildtype lexA or its initiator mutant. These constructs were expressed under the control of a constitutive ADH promoter (29). Construction of GRS1-lacZ fusions followed a similar strategy. Basically, the lexA* portion in GRS1-lexA* fusions was replaced by an initiator mutant of lacZ. Thus, the only translation initiation site for the GRS1-lacZ fusions is TTG(Ϫ23) (or its derivatives) located in the GRS1 portion.
For generating various mutations in the 5Ј-untranslated region of the GRS1 gene, a short GRS1 sequence containing bp Ϫ300 ϳ ϩ300 (relative to ATG1) was PCR-amplified from the full-length GRS1 clone as an EagI-XbaI fragment and cloned in the appropriate sites of pBluescript II SK (Stratagene, La Jolla, CA). This plasmid containing the 600-bp GRS1 sequence was subsequently used as template for mutagenesis. Mutagenesis was carried out following standard protocols provided by the manufacturer (Stratagene, La Jolla, CA). After mutagenesis, the GRS1 fragment (Ϫ300 to ϩ260 bp) carrying the desired mutation was removed from the plasmid by EagI/Bsu36I digestion and then fused in-frame to the 5Ј-end of a truncated version of GRS1 lacking the EagI-Bsu36I segment. Note that Bsu36I is a native cutting site between bp ϩ260 and ϩ266 of GRS1.
RT-PCR-To estimate the relative levels of specific GRS1-lexA mRNAs generated from the fusion constructs, an RT-PCR experiment was carried out following the protocols provided by the manufacturer (Invitrogen). The relative amounts of GRS1-lexA mRNAs were determined using a set of primers complementary to Ϫ40 to Ϫ20 nucleotides of GRS1 (5Ј-TTACTCT-CAGATTGTTAAAAA-3Ј) and ϩ370 to ϩ390 nucleotides of lexA (5Ј-CAAGTCACCATCCATAATGCC-3Ј), respectively. For a more accurate comparison, cDNA products for each construct were obtained from three different cycle numbers of PCR amplification as indicated in the figure. As a control, the relative amounts of actin-specific mRNAs were determined using a set of primers complementary to ϩ537 to ϩ560 nucleotides (5Ј-ACCAACTGGGACGATATGGAAAAG-3Ј) and ϩ696 to ϩ719 nucleotides (5Ј-TTGGATGGAAACGTAGAAGGC-TGG-3Ј) of actin, respectively.
Complementation Assays for the Mitochondrial Function of GRS1-The yeast GRS1 knockout strain, RJT3/II-1, has been described previously (26). To test the mitochondrial function of the GRS1 constructs, RJT3/II-1 was first transformed with a test plasmid (carrying a LEU2 marker) and plated on a selection medium lacking uracil and leucine. Following 5-fluoroorotic acid selections, a single colony of the transformants was picked and grown to stationary phase in a synthetic medium lacking leucine. Starting from 1.0 A 600 , the cultures were 10-fold serially diluted, and 5-l aliquots of each dilution were spotted on the designated YPG plates. The plates were incubated at 30°C for 3-5 days. (Photos for the complementation assays were taken at day 3 following incubation.) Because a yeast cell cannot survive on glycerol without functional mitochondria, the transformants do not grow on YPG plates unless a functional mitochondrial GlyRS is present.
Western Blot-The protein expression patterns of the GRS1-lexA fusions were determined by a chemiluminescence-based Western blot analysis. The lexA fusion constructs were first transformed into INVSc1 (Invitrogen), and the transformants were subsequently grown in a selection medium lacking leucine. Total protein extracts were prepared from each of the transformants with a buffer containing 50 mM Tris, pH 7.4, 150 mM NaCl, 0.5% SDS, 0.5% Triton X-100, 10 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride. Aliquots of the protein extracts (10 -40 g) were loaded onto a mini gel containing 10% polyacrylamide (size: 8 cm ϫ 10 cm) or a medium gel containing 18% polyacrylamide (size: 16 cm ϫ 20 cm) and electrophoresed at 100 volts for 2-4 h. Following electrophoresis, the resolved proteins were transferred using a semi-dry transfer device to a polyvinylidene difluoride membrane in a buffer containing 30 mM glycine, 48 mM Tris base, pH 8.3, 0.037% SDS, and 20% methanol. The membrane was probed with a horseradish peroxidase-conjugated anti-LexA antibody (Santa Cruz, Santa Cruz, CA) and then exposed to an x-ray film following addition of the appropriate substrates. The Western blot data were obtained from at least three independent experiments (only one blot was shown in each figure), and the relative intensity of the protein bands shown was presented as a percentage using the mean value of the intensity of the protein band translated from the native initiator or an initiator with the native context as a reference.
The cells were then lysed at 4°C using a bead beater, followed by centrifugation at 12,000 ϫ g for 2 min. Aliquots of the supernatants (5-200 g) were diluted to 0.8 ml with Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 , and 50 mM 2-mercaptoethanol). ␤-Gal activity assays were initiated (at 37°C) by adding 0.2 ml of o-nitrophenyl ␤-D-galactoside (4 mg/ml). The reaction mixtures were then incubated with constant shaking at 37°C for 30 min, followed by addition of 0.4 ml of 1 M Na 2 CO 3 . The reaction mixtures were centrifuged at 12,000 ϫ g for 2 min, and the A 420 of the supernatants was determined. Relative ␤-gal activities were calculated from A 420 readings normalized to protein concentrations.

Production of GlyRS Isoforms through Leaky Scanning-Our
previous study showed that two distinct protein isoforms are generated from the yeast GRS1 gene via alternative initiation of translation. The longer form that accounts for the mitochondrial activity is initiated from UUG(Ϫ23) (the number "Ϫ23" in UUG(Ϫ23) refers to the 23rd codon triplet that precedes AUG1), whereas the short form that is responsible for the cytoplasmic activity is initiated from AUG1 ( Fig. 1) (11). The leader peptide that is encoded by the nucleotide sequence between the two initiators acts as a targeting signal for the mitochondrial precursor form of GlyRS. Because only one mRNA species having its 5Ј-end located at position Ϫ88 relative to AUG1 was transcribed from GRS1 (11), and UUG(Ϫ23) is a poor initiation codon, translation of the GRS1 mRNA was expected to follow the mechanism of "leaky scanning" proposed by Kozak (30).
To provide direct evidence that scanning ribosomes can skip the relatively poor UUG initiation site and initiate at a downstream AUG, we employed Western blot analysis. Because the processed form of mitochondrial GlyRS is almost indistinguishable in size from its cytoplasmic counterpart, it is unfeasible to directly assay the relative amounts of the isoforms using Western blots. To circumvent this problem, we chose to use lexA as a reporter to measure the relative initiation activities of the alternative initiators of GRS1 (11,31). In addition, to avoid mitochondrial import and cleavage of the signal peptide, only part of the leader sequence of GRS1 (bp Ϫ88 to Ϫ12) was used in construction of the GRS1-lexA fusions. Signal peptide deduced from the truncated leader sequence (containing amino acid residue Met*(Ϫ23) to Val(Ϫ5)) lacks the predicted matrix processing peptidase cleavage site (Fig. 1) and was shown to be unable to target GlyRS into mitochondria (data not shown). As

M* S F F N I S R R F Y S Q I V K K S V K I K R M S V E D I K K A R A A V P F N R E Q L
FIGURE 1. The 5-end sequence of GRS1. The transcription start site at nucleotide position Ϫ88 relative to ATG1 of GRS1 is labeled on top of the sequence. The alternative translation initiation sites, TTG(Ϫ23) and ATG1, are underlined. The mitochondrial targeting signal that extends from Met*(Ϫ23) to Arg(Ϫ1) is shaded for clarity. The predicted mitochondrial matrix processing peptidase cleavage site is marked with OE. Met* stands for methionine encoded by the TTG(Ϫ23) initiator. schematized in Fig. 2, a wild-type or mutant GRS1 fragment containing bp Ϫ88 to Ϫ12 was PCR-amplified and fused inframe to the 5Ј-end of lexA, yielding various GRS1-lexA fusions. These fusions were cloned in the yeast shuttle vector pADH and expressed under the control of a constitutive ADH promoter. As expected, two distinct protein bands were generated from the parental GRS1-lexA fusion (Fig. 2, number 1). Inactivation of the TTG or ATG initiator on the fusions specifically impaired the production of the upper and lower bands, respectively (compare numbers 2 and 4), suggesting that the upper and lower bands were initiated from the TTG and ATG codons, respectively. Consistently, inactivation of both initiators impaired the production of both bands (see number 5). Interestingly but not surprisingly, mutation of TTG(Ϫ23) to ATG strongly increased the levels of the upper band and completely abolished the production of the lower band (see number 3). Further mutation of the downstream ATG initiation codon to ACT in this construct did not alter the pattern of protein expression (see number 6). These results suggest that TTG(Ϫ23) can be readily bypassed by scanning ribosomes, and two protein isoforms can be generated from the GRS1 mRNA through leaky scanning.
Scanning for Sequence Elements Important for Non-AUG Initiation-The question arose as to why UUG(Ϫ23), and not other non-AUG codons that also differ from AUG by a nucleotide in the leader sequence, can be recognized by the preinitiating complex as an alternative translation initiator in GRS1. One possibility is that this non-AUG initiation site contains an unusual sequence context that can effectively compensate for the poor interaction between the initiator codon and initiator tRNA. To test this hypothesis, a trinucleotide scanning strategy was carried out to map the approximate sequence elements important for non-AUG initiation. The mutations were first aimed at the nucleotide sequences flanking the initiating site. In practice, the 5Ј-flanking sequences that are rich in A were mutated to T in a fashion of one codon at a time, whereas the 3Ј-flanking sequences that are rich in T and C were mutated to A in a similar fashion (Fig. 3A). Fig. 3B shows that mutations of the nucleotides at positions Ϫ3 to Ϫ1 relative to the TTG initiator to TTT had the strongest effect among all the mutations tested and reduced the initiating activity up to 10-fold (Fig. 3B,  number 4), whereas mutations at other positions, including positions Ϫ9 to Ϫ7 (to TTT), Ϫ6 to Ϫ4 (to TTT), ϩ4 to ϩ6 (to AAA), ϩ7 to ϩ9 (to AAA), and ϩ10 to ϩ12 (to AAA), altered the initiating activity less than 2-fold (Fig. 3B, numbers 2, 3,  and 5-7). These results suggest that the nucleotides at position Ϫ3 to Ϫ1 relative to the TTG initiator are the most important sequence elements for translation among the flanking sequences tested.  TTT TTT GAA AAA TTG TCG TTC TTC AAT   TTT CAA TTT AAA TTG TCG TTC TTC AAT   TTT CAA GAA TTT TTG TCG TTC TTC  The junction between GRS1 and lexA sequences is a SpeI site. To obtain a more accurate estimate of the relative levels of protein expression, protein extracts prepared from the wild-type construct were 2-fold serially diluted (indicated at the 'bottom of the blot) and compared with those from the mutants. The quantitative data for the relative initiating activities of the mutants are shown as a separate diagram at the bottom.

Effect of Sequence Context
Identifying the Best and Worst Sequence Contexts-The trinucleotide scanning strategy provided a rough picture about which triplet might be important for non-AUG initiation. However, it was still not clear which nucleotide of the triplet affects the initiating activity the most and to what extent. Pursuant to this objective, the nucleotides at position Ϫ3 to Ϫ1 (relative to TTG(Ϫ23)) were each mutated from A to T, G, and C, and the resultant mutants were analyzed for their initiating activities by Western blots. As shown in Fig. 4, mutation of A(Ϫ3) to T, C, and G reduced the initiating activity 6-, 12-, and 4-fold, respectively. Mutation of A(Ϫ2) to T, C, and G reduced the initiating activity 3-, 2-, and 4-fold, respectively. Although mutation of A(Ϫ1) to T and C reduced the initiation activity 3-and 6-fold, respectively, mutation of A(Ϫ1) to G slightly increased the initiating activity, around 1.2-fold. Thus, nucleotides at all these positions play an important role in modulating the efficiency of initiation from the non-AUG codon, and the nucleotide at position Ϫ3 appears to be the most influential one among the three. In addition, A/A/R (R represents A or G) and C/G/C appear to be the most and least favorable nucleotides at these positions, respectively.
Previous studies in higher eukaryotes showed that recognition of an AUG initiator is significantly modulated by the nucleotides at its relative positions Ϫ3 and ϩ4 (16). In contrast, our current study showed that mutations of the nucleotides at positions ϩ4 to ϩ6 had barely detectable effect in yeast (Fig. 3). To rule out the possibility that triple A mutations cannot faithfully reflect the effect of individual nucleotides on initiation, nucleotides at these positions were individually mutated and tested. As shown in Fig. 4A, except for mutation of T(ϩ4) to A, which increased the initiating activity ϳ2-fold, mutations at ϩ4, ϩ5, and ϩ6 had little effect on initiation. It should be noted that changing C(ϩ5) to A would create an in-frame stop codon, TAG, and in turn block protein translation. To avoid this unfortunate situation, the nucleotide at position ϩ6 was concomitantly mutated to C when testing the effect of the nucleotide ϩ5. Taken together, the result suggests that the nucleotides at positions ϩ4 to ϩ6 are not as significant to the initiating activity as those at position Ϫ3 to Ϫ1, and the ϩ4 nucleotide appears to be the only one with a discernible effect on initiation among the three.
Comparing the Effect of Various Sequence Contexts on AUG and Non-AUG Initiators-To examine whether sequence context has a similar effect on translation from various initiation codons in yeast, we next mutated the TTG initiation codon to ACG (another non-ATG initiation codon found in yeast) and ATG (a canonical initiation codon) and then tested their responses to various contexts. As shown in Fig. 5A, ACG and ATG could each substitute for the initiator function of TTG(Ϫ23) in GRS1 (Fig. 5A, numbers 1, 5, and 9). However, when nucleotides Ϫ3 to Ϫ1 relative to the TTG or ACG initiation site were mutated from AAA (presumably the best context) to CGC (presumably the worst context), the resultant constructs lost much of their initiating activity and failed to provide the necessary mitochondrial GlyRS function to the knockout strain (Fig. 5A, numbers 2 and 10). This finding, although interesting, was not totally unexpected in view of the fact that the TTG initiator is highly sensitive to its context changes (Fig. 4). However, it was surprising to find that a similar mutation also affected the complementing activity of the ATG initiator, albeit to a much lesser extent (Fig. 5A, numbers 5 and 6). In contrast to the dramatic effect caused by the mutation of AAA to CGC, mutation of AAA to CAA or TTT had only mild or little effect on the complementing activity of the resultant constructs (Fig.  5A, numbers 3, 4, 7, 8, 11, and 12).
Quantitative analysis of the initiating activity by Western blots showed that these three initiators, TTG, ATG, and ACG, had a relative initiating activity of 1:10:0.5 (Fig. 5B, numbers 1, 5,  and 9). When the context sequences Ϫ3 to Ϫ1 were mutated from AAA to TTT or CAA, the initiating activity of the TTG codon was reduced by 10 -12-fold (Fig. 5, B and C, compare  numbers 1, 3, and 4). In contrast, when the context sequences were mutated to CGC, the initiating activity of the TTG codon was reduced up to ϳ32-fold (Fig. 5, B and C, compare numbers  1 and 2). This result may explain why only pSJ198 lost its complementing function, whereas pSJ215 and pKJ30 did not (Fig.  5A, compare numbers 2-4). A similar hierarchy of context preference was found for the ACG initiator (Fig. 5, B and C,  numbers 9 -12). In contrast to the non-ATG initiators, an ATG  FEBRUARY 8, 2008 • VOLUME 283 • NUMBER 6

JOURNAL OF BIOLOGICAL CHEMISTRY 3177
initiator appeared to be less sensitive toward its context changes; mutation of the context sequences Ϫ3 to Ϫ1 from AAA to CAA or TTT had no discernible effect on its initiating activity (Fig.  5B, compare numbers 5, 7, and 8). However, when the context sequences were mutated to CGC, the initiating activity of the ATG initiator was reduced by 8-fold (Fig. 5, B and C, compare numbers 5 and 6), suggesting that an ATG initiator is subjected to the same rule of modulation, albeit to a much lesser extent. Interestingly, although the ATG initiator with the worst context had an initiating activity similar to that of the TTG or ACG initiator with the best context (Fig. 5B, compare numbers 6, 1, and 9), transformants carrying this compromised ATG initiator showed a relatively poor growth phenotype on YPG (Fig. 5A, compare numbers 6, 1, and 9). To check whether the fusions shown in Fig. 5B express similar levels of mRNAs, a semi-quantitative RT-PCR experiment was conducted. Three different cycle numbers of PCR amplification were carried out for each construct. Fig. 5D shows that similar levels of the cDNA products were generated from these fusions, suggesting that these mutations did not compromise the stability of the specific mRNAs.
Context Effect Determined Using lacZ as a Reporter Gene-To verify whether the quantitative assays shown in Fig. 5 truthfully reflect the initiating activities of the initiation codons with various contexts, we next used lacZ as a reporter. The lexA portion in each of the GRS1-lexA fusions (Fig. 5B, numbers  1-8) was replaced by a lacZ reporter gene where the native initiator codon ATG had been mutated to ACT. Consequently, the initiator codon TTG(Ϫ23) (or its derivatives) in the GRS1 portion becomes the only initiation site for the production of the GlyRS-LacZ fusion proteins (schematized in Fig. 6). Consistent with the results obtained using lexA as a reporter, the GRS1-lacZ fusions showed a similar pattern of expression. Mutation of the context sequence Ϫ3 to Ϫ1 from AAA to CGC, TTT, and CAA reduced the initiating activities of the TTG codon 30-, 6-, and 9-fold, respectively (Fig. 6, compare numbers  1-4). Similar mutations reduced the initiating activities of the ATG codon 4-, 1-, and 1-fold, respectively (compare numbers [5][6][7][8]. In addition, the initiating activity of the TTG codon (with AAA context) was only 3-fold lower than that of the ATG codon with the same context (compare numbers 1 and 5). Although the relative levels of the LacZ fusions were slightly different from those of the LexA fusions, the overall patterns of  PCR cycle  24  18  21   GRS1-lexA   24  18  21  24  18  21  24  18  21   1  2  3  4   24  18  21  24  18  21  24  18  21  24  18

Effect of Sequence Context
their protein expression were consistent. Thus, the results obtained from the LexA fusions appear consistent.

DISCUSSION
We presented evidence here that recognition of the alternative translation initiators of GRS1, UUG(Ϫ23) and AUG1, could be mediated by a mechanism known as leaky scanning (30), wherein only a small portion of the scanning ribosomes (ϳ10% in this case) initiate from the first initiator, i.e. UUG(Ϫ23), whereas the rest of the ribosomes skip this start site and continue scanning toward the 3Ј-end of the mRNA until they encounter the next available initiator, i.e. AUG1. As a consequence, mutations that replaced the upstream UUG initiator codon with AUG effectively impaired the production of the short protein form (Fig. 2). On the other hand, because only a minor portion of the scanning ribosomes start from the native upstream initiator, inactivation of this initiator did not result in a drastic increase in initiation from the downstream AUG initiator. Although leaky scanning might also be involved in the synthesis of AlaRS isoforms in S. cerevisiae (12,31), no solid experimental evidence has been presented so far to confirm this hypothesis. In addition, it should be noted that three distinct mRNAs (with their 5Ј-ends located at nucleotide positions Ϫ117, Ϫ105, and Ϫ54 relative to the AUG initiator, respectively) are alternatively transcribed from the AlaRS gene. Thus, the cytoplasmic form of AlaRS could be translated from the longer two mRNAs via leaky scanning or directly from the shortest mRNA (12). In this sense, it is interesting to mention that expression of two yeast genes, MOD5 (coding for isopentenyl pyrophosphate:tRNA isopentenyl transferase) (32) and CCA1 (coding for ATP (CTP):tRNA nucleotidyltransferase) (33), also involves leaky scanning. However, leaky scanning occurs in these two genes not because the upstream initiator is a non-AUG codon or an initiator with a suboptimal sequence context, but because the first AUG codon in the two genes is positioned too close to the 5Ј-end of the mRNA. In contrast to the few cases of leaky scanning found in yeast, this mechanism has been observed frequently in mammals (34 -37).
Many cellular and viral mRNAs have been shown to use non-AUG codons, such as AUU, CUG, and GUG, as translation start sites (30). Occasionally, non-AUG codons act as exclusive translation initiators, such as in the mRNAs derived from the Arabidopsis AGAMOUS gene (38) and the latently expressed kaposin locus of Kaposi's sarcoma-associated herpesvirus (39). However, in most cases, non-AUG codons serve as alternative translation start sites that are accompanied by a downstream in-frame AUG initiator (17). Recognition of the non-AUG initiators in these mRNAs by initiating ribosomes appears to be compensated for by interactions with nucleotide sequences flanking the initiators. A recent study suggested that components of the 48 S translation initiation complex, in particular eIF2␣ and 18 S rRNA, might be involved in specific recognition of the context nucleotides at positions Ϫ3 and ϩ4 (18). So, an optimal sequence context is important to the efficient recognition of a non-AUG initiator. Likewise, recognition of a canonical AUG initiator can be severely compromised by a suboptimal sequence context. In contrast to the active participation of sequence context in translation initiation in higher eukaryotes, many reports argued that sequence context does not appear to substantially affect the initiating activity of an AUG initiator in yeast (21,22,40). Mostly for this reason, yeast was previously thought to be unable to use non-AUG codons as translation start sites (24). Surprisingly, two genes in S. cerevisiae (ALA1 and GRS1) and one gene in C. albicans (CARP2A) have recently been reported to use naturally occurring non-AUG triplets as translation initiators. Even more surprising was the finding reported here that the UUG initiator of GRS1 is strongly affected by its 5Ј-flanking sequences (Figs. 4 -6). As it turned out, this noncanonical initiator happens to have a very good, if not the best, sequence context, which may explain why only this particular UUG triplet, and not other non-AUG codons that also differ from AUG by a nucleotide elsewhere in the leader sequence, can function as a translation start site (11). Perhaps for that same reason, an ACG triplet can functionally substitute for the TTG initiator (Fig. 5A, number 9). In addition, it is interesting to point out that the yeast AlaRS gene, which uses ACG as an alternative initiation codon, also contains the optimal 5Ј sequence context AAR (AAG in this case), lending further support to the findings of this report.
In higher eukaryotes, the most critical context positions are Ϫ3 and ϩ4 relative to the initiator, and the best (or most favorable) context contains a purine (A or G) at position Ϫ3 and a G at position ϩ4 (30). Although sequences immediately flanking AUG initiation codons are somewhat preferred in yeast, the bias in nucleotide distribution (5Ј-(A/Y)A(A/Y)AaugUCY-3Ј) is highly diverged from the higher eukaryotic consensus (5Ј-CACCaugG-3Ј) with the exception of an A preference at the Ϫ3 position (20). In addition, many studies suggested that sequence context plays only a minor role in translation of yeast mRNAs initiated from AUG. In this sense, the yeast consensus deduced solely from sequence analysis might not actually represent the best context for translation initiation in yeast. We report here that the best context for the UUG initiator of the yeast GRS1 gene is AAR (R represents A or G) at nucleotide positions Ϫ3 to Ϫ1, and the most critical context position is position Ϫ3 (Figs. 4 and 5). Mutation of A(Ϫ3) to C alone reduced the initiating activity of the UUG initiator (or an ACG codon positioned in the same context) up to . To our knowledge, this is the most dramatic context effect that has been reported for an initiator in yeast. Consistent with these findings, many highly expressed yeast genes are found to have an A at the Ϫ3 position and an A-rich leader region (20). In contrast to position Ϫ3, the nucleotide at position ϩ4 appears to have only minor effect on the efficiency of the UUG initiator, which is very different from the scenario seen in higher eukaryotes. Mutation of the nucleotide at this position to any other nucleotide affected the initiating activity at most by 2-fold, which is much less than the effect manifested by any of the nucleotides at positions Ϫ3, Ϫ2, and Ϫ1 (Fig. 4). Interestingly, recent studies suggested that the nature of the penultimate amino-terminal residue may influence the substrate preferences of the methionine aminopeptidase and N-␣-acetyltransferase and in turn affect the stability of the produced protein (41,42). Therefore, changes in the ϩ4, ϩ5, or ϩ6 position may affect not only the context preference but also the stability of the protein. A more thorough study is needed to delineate the effect of nucleotides at these positions. So far, it is not known whether secondary structure also plays a role in modulating the efficiency of the UUG initiator. Although most of the mutations shown in Figs. 3 and 5 did not lead to formation or loss of a stable secondary structure near the initiation site, mutation of AAA (position Ϫ3 to Ϫ1 relative to the UUG initiator) to CGC or of UCG (position ϩ4 to ϩ6 relative to the UUG initiator) to AAA was predicted to facilitate the formation of a weak secondary structure (with ⌬G of Ϫ10 and Ϫ7 kcal/mol, respectively). In the CGC (position Ϫ3 to Ϫ1) mutant, AAGAACG (position Ϫ8 to Ϫ2) pairs with CGUUCUU (position ϩ5 to ϩ11) (Fig. 5,  number 2), whereas in the AAA (position ϩ4 to ϩ6) mutant, UUUCAA (Ϫ12 to Ϫ7) pairs with UUGAAA (ϩ1 to ϩ6) (Fig. 3,  number 5). Both stem-loop structures include the UUG initiation site; UUG is either in the loop or in the stem. Strangely enough, the CGC mutation significantly reduced the efficiency of translation initiation (Fig. 5B, compare numbers 1 and 2), whereas the AAA mutation had little effect on UUG initiation (Fig. 3B, compare numbers 1 and 5). Further investigation is under way to check whether the weak structure formed in the CGC mutant actually plays a role in UUG initiation.
It should be noted that the context preference reported here was deduced mainly from a non-AUG initiator model (Figs. 3 and 4), and an AUG initiator appeared to be much less sensitive under similar conditions (Figs. 5 and 6). For example, mutation of A(Ϫ3) to C or AAA (Ϫ3 to Ϫ1) to UUU had little effect on an AUG initiation codon placed in the same position as UUG(Ϫ23) (Figs. 5 and 6). This feature appears to be consistent with many of the previous reports, where various mutations in the 5Ј sequence context affected the translation initiation from an AUG codon at most 2-3-fold (21,22). Nevertheless, mutations that changed the context of an AUG initiator from AAA (Ϫ3 to Ϫ1) to CGC significantly impaired its initiation activity (4 -8-fold) (Figs. 5 and 6). This result is particularly noteworthy because a single nucleotide change even in the most critical position has little effect on translation from an AUG codon. Therefore, context effect is likely to be less obvious if a relatively insensitive system with AUG as the initiation codon is used as the test model (24). A non-AUG initiation codon such as the one present in GRS1 can perhaps serve as a better model in identifying the sequence, and structural elements important for translation initiation that may have previously been overlooked in yeast.