Recognition of AUG and alternative initiator codons is augmented by G in position +4 but is not generally affected by the nucleotides in positions +5 and +6
1997; Springer Nature; Volume: 16; Issue: 9 Linguagem: Inglês
10.1093/emboj/16.9.2482
ISSN1460-2075
Autores Tópico(s)RNA modifications and cancer
ResumoArticle1 May 1997free access Recognition of AUG and alternative initiator codons is augmented by G in position +4 but is not generally affected by the nucleotides in positions +5 and +6 Marilyn Kozak Marilyn Kozak Department of Biochemistry, University of Medicine and Dentistry of New Jersey, 675 Hoes Lane, Piscataway, NJ, 08854 USA Search for more papers by this author Marilyn Kozak Marilyn Kozak Department of Biochemistry, University of Medicine and Dentistry of New Jersey, 675 Hoes Lane, Piscataway, NJ, 08854 USA Search for more papers by this author Author Information Marilyn Kozak1 1Department of Biochemistry, University of Medicine and Dentistry of New Jersey, 675 Hoes Lane, Piscataway, NJ, 08854 USA The EMBO Journal (1997)16:2482-2492https://doi.org/10.1093/emboj/16.9.2482 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info A primer extension (toeprinting) assay was used to monitor selection by ribosomes of the first versus the second AUG codon as a function of introducing mutations on the 3′ side (positions +4, +5 and +6) of the first AUG codon. Six different flanking codons starting with G (GCG, GCU, GCC, GCA, GAU and GGA) strongly augmented selection of AUG#1 when compared with matched mRNAs that had A or C instead of G in position +4. Augmentation by G in position +4 failed only when it was combined with U in position +5, as in the sequence augGUA. In contrast with the usual enhancing effect of introducing G in position +4, most mutations in position +5 had no discernible effect, as shown with the series augANA (where N = C, A, G or U) and the series augCNA. AUG codon recognition was also unaffected by mutations in position +6, as shown by testing four mRNAs that had augCCN as the start site. Thus the primary sequence context that augments the recognition of AUG start codons does not appear generally to extend beyond G in position +4. When the toeprinting assay was used with mRNAs that initiate translation at CUG instead of AUG, cugGAU was not recognized better than cugGGU, contradicting the hypothesis that initiation at non-AUG codons might be favored by A instead of G in position +5. Introduction Eukaryotic ribosomes appear to select the start site for translation by a scanning mechanism. The working hypothesis is that the small, 40S ribosomal subunit, carrying Met-tRNAimet and various initiation factors, engages the mRNA at the capped 5′ end and migrates linearly until it encounters the first AUG codon. At the AUG codon, which is recognized by base pairing with the anticodon in Met-tRNAimet, the 40S ribosomal subunit stops, the 60S subunit joins and the 80S ribosome is poised to start protein synthesis. Evidence for this scanning mechanism and for the corollary first-AUG rule is summarized elsewhere (Kozak, 1989a, 1992, 1995). In higher eukaryotes, sequences flanking the AUG codon modulate its ability to halt the scanning 40S ribosomal subunit. One of the modulating elements is the GCCACC motif in positions −6 to −1, immediately preceding the AUG codon (Kozak, 1987). Mutations that weaken adherence to this consensus motif, especially mutations that substitute a pyrimidine for the A in position −3, cause some 40S ribosomal subunits to bypass the first AUG and to initiate instead at the next AUG downstream (Kozak, 1986a, 1989b; Lin et al., 1993; Ossipow et al., 1993). This context-dependent ‘leaky scanning’ has also been seen when the highly conserved G in position +4, immediately following the AUG codon, is mutated (Kozak, 1986a, 1989b). Deviations in one or both of these key positions, and the resulting leaky scanning, seem to account for the ability of certain mRNAs to produce two proteins by initiating translation from the first and second AUG codons (Kozak, 1986b, 1991). Two recent studies have raised the possibility that context effects on initiation might extend into the coding domain beyond position +4. In one case, initiation at GUG appeared to be more efficient when the second codon was GAU instead of GUA (Boeck and Kolakofsky, 1994). A companion study by Grünert and Jackson (1994) reported similarly that initiation at an AUG or CUG start codon was favored by A in position +5 and U in position +6. However, documenting the involvement of these or other nucleotides on the 3′ side of the initiator codon might be complicated by the fact that mutations introduced in these positions of the mRNA may change the amino acid sequence of the encoded polypeptide. This could be a problem because the identity of the amino acid adjacent to the N-terminal methionine can affect post-translational modifications which, in turn, can affect protein turnover. To circumvent possible complications from post-translational events, an assay that directly monitors ribosome–mRNA initiation complexes was used in the present study to reinvestigate the question of whether nucleotides in positions +4, +5 and +6 affect the recognition of initiator codons. Correct definition of the context requirements for initiation is important for predicting translational start sites, which is an important aspect of interpreting cDNA sequences (Kozak, 1996). Results Preliminary test of mutations in positions +4, +5 and +6 The mRNAs used for these experiments have two start codons and two open reading frames (ORFs), as outlined in Figure 1. ORF1, which extends from AUG#1 to a UAA codon overlapping Leu45 in the chloramphenicol acetyltransferase (CAT) coding sequence, encodes a 70 amino acid polypeptide with a molecular mass of 8 kDa. This polypeptide is designated p8out (meaning out-of-frame with respect to CAT) or simply p8. ORF2 initiates with AUG#2, which is in-frame with the downstream CAT coding sequence. ORF2 thus encodes a 240 amino acid polypeptide (the 219 amino acid CAT protein with a 21 amino acid N-terminal extension), with a molecular mass of 28 kDa. The product of ORF2 is designated p28precat or simply p28. Because the sequence preceding AUG#1 includes U in position −3, which is suboptimal, leaky scanning should allow these mRNAs to produce both polypeptides: p8 from AUG#1 and p28 from AUG#2. The control mRNA in Figure 2A (lane 9) illustrates how this leaky scanning can be modulated by changes in context. Because AUG#1 in the control has the optimal A in position −3, this mRNA produces a much higher yield of p8, and a much lower yield of p28, than any other mRNA in this series. This fits with previous studies of mutations involving sequences on the 5′ side of the AUG codon (Kozak, 1986a, 1989b). Figure 1.Sequences of mRNAs used to study the effects of varying the nucleotide in positions +4, +5 and +6. The sequence in the top line is common to all mRNAs in this series. The 3′ end of this sequence indicated by an ellipsis (. . .) leads to the CAT coding domain (Kozak, 1989b). Not shown is the m7G cap at the 5′ end of all mRNAs. Mutations in the indicated positions of particular mRNAs were introduced around the first AUG codon, which initiates translation of an 8 kDa polypeptide (p8). In a different, overlapping reading frame, the second AUG codon initiates translation of a 28 kDa polypeptide (p28) which is an N-terminally extended version of CAT. Because of the suboptimal context preceding AUG#1 (notably the presence of U rather than A in position −3), some ribosomes would be expected to reach AUG#2 by leaky scanning. Thus, each of the 12 test constructs should direct translation of both p8 and p28. The improved context (A in position −3) in the control p-Con-1 should strongly shift translation in favor of p8. With the other control, p-Con-0, p28 should be the sole translation product because the upstream AUG codon is absent. Notice that mRNAs are named by stating the three bases following the first AUG codon. Download figure Download PowerPoint Figure 2.Translation of mRNAs that vary in positions +4, +5 and +6 flanking the first AUG codon. The autoradiograms show [3H]leucine-labeled proteins produced in a rabbit reticulocyte translation system using mRNAs that have point mutations in positions +4 and +5 (A) or position +6 (B). The mutations identified above each lane were introduced around AUG#1, which initiates translation of p8. p28 results from initiation at the invariant AUG#2. Figure 1 gives the 5′ end sequences of these mRNAs in full. A control mRNA that lacks AUG#1 (p-Con-0 in lane 1 of A) produced only p28. For the control mRNA p-Con-1 (lane 9 in A; lane 5 in B), the context around AUG#1 was improved by changing position −3 from U to A, thus enhancing synthesis of p8 and greatly reducing synthesis of p28. In (B), the slight residual translation of p28 evident in lane 5 was abolished in lane 6 by introducing downstream the structure-prone sequence 8336, which is thought to slow scanning and thus augment recognition of AUG#1 (Kozak, 1990a). This is shown only as an illustration, inasmuch as all the other mRNAs used in this figure contained the unstructured sequence 8335 at the BamHI site. The conditions used for translation (protein accumulation assay) and subsequent fractionation by polyacrylamide gel electrophoresis are described in Materials and methods. Download figure Download PowerPoint The present study asks whether mutations on the 3′ side of AUG#1 can also modulate the selection of translational start sites. As shown in Figure 2A (lanes 2–8), the yield of p8 initiated from AUG#1 indeed varied at least 5-fold when point mutations were introduced in positions +4 or +5. However, the scanning mechanism predicts that if, for example, p-augAAA really supports initiation better than p-augAUA, as suggested by the 5-fold higher yield of p8 in lane 6 versus lane 8, then the yield of p28 should be proportionately lower in lane 6. That prediction is not met. Instead, the only mRNA in the test series that shows both elevated p8 synthesis and reduced p28 synthesis is p-augGCA (lane 2), the construct that has G instead of U, C or A in position +4. With the other mRNAs tested in Figure 2A, a possible explanation for the variable yield of p8 without concomitant reduction of p28 is that mutations in positions +4 and +5, which change the subterminal amino acid, thereby alter the turnover of polypeptide p8. In this case, the amount of radiolabeled p8 that accumulates during the hour-long incubation would not reflect the efficiency of initiation at AUG#1 accurately. To circumvent this potential problem, the mRNAs used in Figure 2A were retested using a direct initiation assay. To examine the effects of mutations in position +6, I chose a codon that specifies the same amino acid regardless of which base occurs in position +6. Thus the N-terminal sequence of the nascent polypeptide is Met–Pro when translation initiates at augCCG, augCCU, augCCC or augCCA. Among these four mRNAs there was no significant difference in the yield of p8 in a standard translation assay (Figure 2B, lanes 1–4). These mRNAs were also retested using the initiation assay described next. Direct analysis of AUG codon recognition using mRNAs with mutations in positions +4, +5 and +6 By using a reticulocyte lysate supplemented with sparsomycin and cycloheximide to inhibit elongation (see Materials and methods), initiation complexes accumulate in which the ribosome is held at the AUG codon. The particular AUG start site can be identified by using a primer extension inhibition assay in which a 32P-labeled deoxyoligonucleotide primer, annealed to the mRNA downstream from all potential initiator codons, is extended by reverse transcriptase up to the 3′ edge of the bound ribosome. Figure 3 outlines how the assay works in principle. Figure 3.Schematic representation of a primer extension assay for mapping the position of ribosomes on mRNA. The unextended 32P-labeled primer, represented by the wide black line in step (b), is shown near the bottom of the polyacrylamide gel in step (d). Extension of the primer with reverse transcriptase in the absence of bound ribosomes proceeds to the 5′ end of the mRNA. If ribosomes are allowed to bind to the mRNA before the addition of reverse transcriptase, primer extension halts prematurely; the exact size of the extension product(s) reveals which AUG codon(s) were selected, taking into account that the leading edge of an 80S ribosome extends ∼15 nucleotides 3′ of the AUG codon (Kozak and Shatkin, 1977). The basic design of this ‘toeprinting’ assay was developed by Hartz et al. (1988) for studies with prokaryotic ribosomes. The primer used in the present studies was 23 nucleotides long, the full-length extension product was 184 nucleotides and the extension inhibition products obtained when a ribosome was bound at AUG#1 or AUG#2 were 123 and 109 nucleotides, respectively. Download figure Download PowerPoint Two control reactions in Figure 4 illustrate how the assay works in practice. With the control mRNA p-Con-1 in which AUG#1 resides in a nearly optimal context (ACAaugG, see Figure 1), one prominent primer extension product is evident in Figure 4A (lanes 11 and 12) and the size of this product indicates that it derives from ribosomes bound at AUG#1, the start codon for p8. This primer extension product was absent when p-Con-0 mRNA was used for ribosome binding (Figure 4A, lanes 1 and 2), consistent with the fact that p-Con-0 lacks the upstream AUG codon (see Figure 1). With p-Con-0 the toeprinting assay maps ribosomes instead at the p28 start codon. (The p28 start site is labeled AUG#2 in Figure 4 because it is the second start codon in all mRNAs except p-Con-0.) Figure 4.Primer extension analysis of ribosome–mRNA complexes. Initiation at the first and second AUG codons was monitored as a function of introducing mutations around AUG#1. The assay is explained in Figure 3. The primer (PR) and primer extension products are labeled along the left margin. (A and B) The mRNAs used in lanes 3–10 varied only in position +4 or position +5, as indicated at the top of each panel. The sequences of these mRNAs as well as the two control transcripts (lanes 1, 2, 11 and 12) are depicted in full in Figure 1. Adjacent bracketed lanes show that, with a given mRNA, the ratio of initiation at AUG#1 versus AUG#2 was the same when low (lanes 1, 3, 5, 7, 9 and 11) and 3-fold higher (lanes 2, 4, 6, 8, 10 and 12) levels of initiation complexes were analyzed. Because of small variations in the amount of radioactivity applied to each lane, the important comparison is not the intensity of the AUG#1 band from lane to lane, but the ratio of AUG#1 to AUG#2 in each lane. These ratios are given in Table I. (C) Toeprint analyses with mRNAs that differed in position +6, as indicated above lanes 5–8. Lanes 1–4 display the complementary strand sequence of p-augCCA mRNA. A series of black dots within these sequencing lanes highlight the C–A–T bands that correspond to the first, second and (silent) third AUG codons. When the electrophoresis was run for longer, the foreshortened primer extension products caused by bound ribosomes could be mapped, by reference to the sequencing lanes, 15–16 nucleotides downstream from the first and second AUG codons. In the absence of ribosomes, the primer was extended all the way to the 5′ end of the mRNA, as shown in a control reaction (D). Download figure Download PowerPoint The rest of Figure 4A tests the effects of introducing mutations in position +4 flanking AUG#1. Because all four test transcripts (the first four mRNAs in Figure 1) have U instead of the optimal A in position −3, some ribosomes are able to reach AUG#2 by leaky scanning. The question is whether the ratio of initiation at AUG#1 versus AUG#2 differs among these four mRNAs which are identical except for position +4. Quantitation of the data from Figure 4A (Table I, measurement 1, entries 1–4) indeed shows a 2.6-fold shift in favor of AUG#1 when that codon is followed by G in position +4 (henceforth written G+4). There was no real hierarchy among the other three nucleotides in position +4. Table 1. Effects of mutations in positions +4, +5 and +6 as monitored by primer extension analysis of ribosome–mRNA initiation complexes Entry No. Sequence flanking AUG#1a Ratio of initiation at AUG#1 versus AUG#2b Measurement #1 (2.0 mM Mg2+) Measurement #2 (1.4 mM Mg2+) Measurement #3 (2.0 mM Mg2+) Measurement #4 (2.7 mM Mg2+) Actual Normalized Actual Normalized Actual Normalized Actual Normalized 1 Position +4 is G 2.6 2.6 0.46 2.4 2.2 2.2 6.3 3.9 2 U 1.0 1.0 0.19 1.0 1.0 1.0 1.6 1.0 3 C 0.8 0.8 0.16 0.8 0.8 0.8 1.6 1.0 4 A 0.9 0.9 0.19 1.0 0.8 0.8 1.7 1.1 5 Position +5 is C 0.8 1.1 0.13 1.0 0.6 1.2 1.4 1.1 6 A 0.7 1.0 0.14 1.1 0.5 1.0 1.5 1.2 7 G 0.7 1.0 0.15 1.2 0.5 1.0 1.5 1.2 8 U 0.7 1.0 0.13 1.0 0.5 1.0 1.3 1.0 9 Position +6 is G 1.1 1.1 0.23 1.0 1.2 1.0 2.2 0.9 10 U 1.0 1.0 0.24 1.0 1.2 1.0 2.4 1.0 11 C 0.9 0.9 0.16 0.7 0.9 0.8 2.0 0.8 12 A 0.8 0.8 0.19 0.8 1.0 0.8 1.9 0.8 a The codon following AUG#1 was NCA for the +4 series, ANA for the +5 series and CCN for the +6 series. The mRNA sequences are given in full in Figure 1. b Because the sequence flanking AUG#2 was constant, an increase in the AUG#1/AUG#2 ratio indicates improved recognition of AUG#1. The data were obtained by densitometric scanning of the autoradiograms in Figure 4 (for measurement 1) and Figure 5 (for measurements 2–4). Because the mRNAs were tested four at a time (i.e. in Figure 5, four mRNAs were tested at three Mg2+ concentrations in each of three experiments), the strictly valid comparison is among mRNAs 1–4 or among mRNAs 5–8 but not between mRNAs 1–4 and mRNAs 5–8. However, by using the entries in columns 4, 6, 8 and 10, wherein the values have been normalized to the mRNA in each series that has U in the test position, it is possible to compare mRNAs between experiments. Ratio of initiation at AUG#1 versus AUG#2b 2+ Measurement #4 (2.7 mM Mg) The toeprinting experiment was repeated in Figure 4B using four mRNAs that were identical except for position +5. Quantitation of these results showed no significant shift in the AUG#1/AUG#2 ratio (Table I, measurement 1, entries 5–8). Thus there was no evidence that the nucleotide in position +5 affects the selection of translational start sites. Nor was there any significant effect when mutations in position +6 were tested (Figure 4C; Table I, measurement 1, entries 9–12). Because leaky scanning in cell-free translation systems was shown previously to be sensitive to the concentration of Mg2+ (Kozak, 1989b, 1990b), I repeated the test of mutations in positions +4, +5 and +6 at three different Mg2+ concentrations. The results of these toeprinting assays are shown in Figure 5 and the quantitation is given in Table I (measurements 2, 3 and 4). As reported previously, when a given mRNA is tested at different Mg2+ concentrations, the tendency to scan past AUG#1 and initiate instead at AUG#2 increases as the Mg2+ concentration is decreased. This can be seen in Figure 5A, for example, by comparing the translation of p-augGCA in lanes 1, 5 and 9. The point is sustained by comparing any other mRNA in Figure 5A at low, medium and high concentrations of Mg2+ (e.g. p-augUCA in lanes 2, 6 and 10; p-augCCA in lanes 3, 7 and 11; or p-augACA in lanes 4, 8 and 12). Figure 5.Selection of AUG start sites as a function of magnesium concentration and sequence variation following AUG#1. (A) Twelve toeprinting reactions using mRNAs that differed from one another only in position +4 following AUG#1. The identity of the base in position +4 is marked above each lane. The primer extension reactions were carried out after 4 min incubation in a reticulocyte translation system in which the magnesium concentration had been adjusted to 1.35 (lanes 1–4), 2.02 (lanes 5–8) or 2.70 mM (lanes 9–12). The experiment was repeated using mRNAs that differed from one another only in position +5 (B) or position +6 (C). The mRNA sequences are given in full in Figure 1. Autoradiograms (similar to Figure 4) have been cropped. Quantitation of these results is given in Table I. Download figure Download PowerPoint The real purpose of this experiment was to compare, at a given concentration of Mg2+, the translation of four mRNAs that differ in a single position downstream of AUG#1. In Figure 5A (lanes 1–4) this four-way comparison reveals that G is the only nucleotide in position +4 that augments recognition of AUG#1 at low Mg2+, and that conclusion holds at moderate (lanes 5–8) and high (lanes 9–12) Mg2+ concentrations. In Figure 5B, a similar experiment using four mRNAs that differ only in position +5 shows that, at any given Mg2+ concentration, AUG#1 is recognized with equal efficiency irrespective of nucleotide changes in position +5. Figure 5C shows that, at any given Mg2+ concentration, recognition of AUG#1 is indifferent also to the nucleotide in position +6. The conclusion from this analysis is that G+4 appears to be the only nucleotide on the 3′ side of the AUG codon that augments initiation. Distinguishing between particular codon effects and generalized context effects Although G+4 augments AUG codon recognition under a variety of reaction conditions, as shown above, in all those studies the G in position +4 was part of the codon GCA. To determine if the augmentation is attributable specifically to G+4 or if it is the flanking codon GCA that happens to favor initiation, I tested mRNAs that had six different GNN codons adjacent to AUG#1. In the toeprint analyses shown in Figure 6A, each mRNA was compared with a matched construct that had C or A instead of G in position +4. Quantitation of the results (Table II) reveals that AUG#1 was indeed recognized ∼3-fold better in five out of six cases where G+4 was the flanking nucleotide. Since five different flanking codons starting with G (GCG, GCU, GCC, GCA and GAU) strongly augmented the recognition of AUG#1, it seems reasonable to attribute the enhancement to the G residue in position +4 rather than to a particular flanking codon. Figure 6.Additional testing of various codons following AUG#1. Primer extension assays were carried out using a reticulocyte translation system with the Mg2+ concentration adjusted to 1.7 mM. The codon flanking AUG#1 is identified above each lane of the gel. Except for this first codon variation, the mRNA sequences were as given in Figure 1. (A) The positive effect of G+4 is seen with a variety of codons flanking AUG#1. The first eight lanes, compared two at a time, show the shift in initiation when AUG#1 is followed by G versus C in position +4. Thus the flanking codons are GCG versus CCG in lanes 1 and 2, GCU versus CCU in lanes 3 and 4, GCC versus CCC in lanes 5 and 6 and GCA versus CCA in lanes 7 and 8. The last four lanes show the shift when AUG#1 is followed by G versus A in position +4. The flanking codons are GAU, lane 9; AAU, lane 10; GUA, lane 11; and AUA, lane 12. Quantitation of these results is given in Table II. (B) The stimulatory effect of G+4, evident when AUG#1 is followed by GAU (lane 1) or GGA (lane 6), fails when the flanking codon is GUN (lanes 2–5). (C) Variations in position +5 do not significantly affect recognition of AUG#1, as shown with CNA as the flanking codon (lanes 1–4). The mRNAs used in lanes 5 and 6 are controls. Warming during electrophoresis slightly retards the mobility of samples near the edges of the gel. Download figure Download PowerPoint Table 2. The positive effect of G in position +4 occurs with a variety of codons flanking AUG#1 Entry No. Sequence at AUG#1 Ratio of initiation at AUG#1/AUG#2a Actualb Normalizedc 1 augGCG 2.00 3.3 2 augCCG 0.61 1.0 3 augGCU 1.70 4.2 4 augCCU 0.41 1.0 5 augGCC 1.60 3.4 6 augCCC 0.47 1.0 7 augGCA 1.40 3.0 8 augCCA 0.46 1.0 9 augGAU 1.20 2.4 10 augAAU 0.50 1.0 11 augGUA 0.35 0.9 12 augAUA 0.40 1.0 a Because the sequence flanking AUG#2 was constant, an increase in this ratio indicates improved recognition of AUG#1. b The autoradiogram in Figure 6A was used for these measurements. c To normalize the results, the AUG#1/AUG#2 ratio for each mRNA that has G in position +4 is expressed relative to the matched C+4 or A+4 mRNA, which is set at 1.0. In Figure 6A, augGUA was the only mRNA in which G+4 unexpectedly failed to enhance initiation. To determine whether it is specifically the flanking codon GUA that disfavors initiation or whether the 3′ sequence GU somehow undermines recognition of the preceding AUG codon, I tested initiation at augGUG, augGUU and augGUC along with augGUA. Figure 6B (lanes 2–5) shows equally poor recognition of AUG#1 with all four constructs in this series. Thus the usual stimulatory effect of G+4, seen in Figure 6B with the control transcripts augGAU and augGGA (lanes 1 and 6), fails for some reason when G+4 is followed by U+5. Although U in position +5 prevents the usual stimulatory effect of G+4, U in position +5 is not generally deleterious. Thus, augAUA was not recognized less efficiently than augACA, aug AAA or augAGA in Figure 5B. The point is confirmed in Figure 6C where augCUA (lane 4) was recognized as efficiently as augCCA, augCAA or augCGA (lanes 1–3). Effects of mutations flanking a CUG start codon In view of some earlier reports about effects of downstream mutations (Boeck and Kolakofsky, 1994; Grünert and Jackson, 1994), it seemed useful to retest some of the foregoing conclusions with mRNAs that initiate translation at a non-AUG codon. In the mRNAs depicted in Figure 7A, CUG replaces AUG#1 as the start codon for p8. When these mRNAs were used as templates in a standard translation assay, some [3H]leucine-labeled p8 was produced (Figure 7B, lanes 2–5), albeit less than with AUG as the start site for p8 (Figure 7B, lanes 1 and 6). That AUG as the p8 start codon is much stronger than CUG is also evident from the greater inhibition of p28 synthesis in lanes 1 and 6 compared with lanes 2–5 in Figure 7B. The complete absence of p8 when the CUG codon was mutated (Figure 7C, lane 1) confirms that CUG is the source of p8 in lanes 2–5. Figure 7.Effects of mutations in positions +4 and +5 on recognition of a CUG initiator codon. (A) Sequences of the mRNAs used for translation. The major difference from Figure 1 is that AUG#1 has been replaced here by a CUG start codon. To compensate for the weakness of the CUG codon, the preceding sequence includes the optimal A in position −3. Each construct was tested with an unstructured sequence (8335) and a secondary structure-forming sequence (8336) downstream. (B and C) Autoradiograms of [3H]leucine-labeled proteins resulting from translation under standard conditions (2 mM Mg2+) of the mRNA indicated above each lane. The autoradiogram in (C) was exposed twice as long as that in (B). (D) Cropped autoradiograms of toeprinting reactions carried out with p-cugAGU mRNA (lanes 3 and 4), p-cugAAU (lanes 5 and 6), p-cugGGU (lanes 7 and 8) and p-cugGAU (lanes 9 and 10). Lanes 1 and 2 show a control reaction with mRNA #9 which lacks the upstream CUG start site. For each mRNA, adjacent bracketed lanes show toeprinting reactions carried out using the first two 32P-containing fractions eluted from the Sepharose CL-4B column. Because of difficulty in synchronizing the collection when several columns are run at the same time, the first fraction (lanes 1, 3, 5, 7 and 9) contains less radioactivity in some cases. Consequently, the mRNAs are most easily compared by focusing on lanes 2, 4, 6, 8 and 10, where equal radioactivity was applied. Download figure Download PowerPoint Because earlier studies showed that appropriately positioned downstream secondary structure aids the recognition of weak initiator codons (Kozak, 1990a), I tested the CUG mRNAs with both an unstructured sequence (oligonucleotide 8335) and a structure-forming sequence (oligonucleotide 8336) downstream. The inclusion of oligonucleotide 8336 significantly elevated initiation from the CUG codon in Figure 7B (compare the yield of p8 in lane 4 versus lane 2, or in lane 5 versus lane 3); therefore, this downstream sequence was retained in the mRNAs used in Figure 7C and D. In Figure 7C, I examined the effects of mutations in position +5 flanking the CUG start codon for p8. I specifically tested A versus G in this position because Grünert and Jackson (1994) reported the biggest effect when A+5, which they considered optimal, was changed to G+5. Although I too observed a higher yield of p8 with the mRNA that initiates at cugGAU instead of cugGGU (lanes 4 and 5 in Figure 7B and C), closer inspection of the results argues against concluding that A+5 enhances initiation at CUG. Notice, for example, that the yield of p8 was not augmented by A in position +5 when the flanking codons were AGU versus AAU (Figure 7C, lanes 2 and 3) instead of GGU and GAU. Because differences in protein stability might distort the results of the protein accumulation assays in Figure 7B and C (for example, N-terminal acetylation might stabilize the form of p8 initiated from cugGAU), the critical test was whether mutations in position +5 would affect recognition of the CUG codon when initiation was monitored directly, using the primer extension assay. As shown in Figure 7D, although G in position +4 augmented recognition of the upstream CUG codon (compare cugAGU with cugGGU
Referência(s)