We show that ATP-binding cassette protein 50, which binds eukaryotic initiation factor-2, plays a key role in translation initiation. ATPase-defective mutants of ABC50 usurp the accuracy of the recognition of start codons, suggesting it helps ensure the accuracy of initiation.

INTRODUCTION

The ATP-binding cassette protein (ABC) 50, also termed ABCF1, was first identified as interacting with eukaryotic initiation factor 2 (eIF2) [1,2], the protein which recruits the initiator methionyl-tRNA (Met-tRNAiMet) to the small (40S) ribosomal subunit. It is the anti-codon of this tRNA that recognizes the initiation codon in the mRNA, leading to the beginning of translation from the physiologically relevant start position. Translation normally starts from AUG codons, although certain other related codons can be used, albeit with lower efficiency. Furthermore, the sequence around the initiation codon (the ‘context’) also plays a strong role in the efficiency with which the codon is used for translation initiation [3,4].

Data from the effects of siRNA-mediated knockdown of ABC50 and for expression of mutants of ABC50 which are defective for ATP binding or hydrolysis indicate that ABC50 plays a positive role in translation both for general 5′-cap-dependent translation and for translation driven by certain internal ribosome entry sites (IRESs) [1]. However, the nature of the role of ABC50 in translation initiation has remained obscure.

In addition to the eIF2–GTP–Met-tRNAiMet ternary complex (TC), the selection of start codons is modulated by eIF1, eIF1A and eIF5 (reviewed in [5]). eIF1 and eIF1A promote an open scanning-competent conformation of the 40S subunit [6]. Recognition of the initiation codon leads to dissociation of eIF1 [7], which is accompanied by release of inorganic phosphate (Pi)from eIF2–GDP. Hydrolysis of eIF2-bound GTP is facilitated by eIF5, which acts as a classical GTPase-activator protein (GAP) for eIF2 [8,9]. eIF1 acts to impair initiation codon recognition and to enhance selectivity for AUG codons in a good sequence context [10]. Conversely, eIF5 favours acceptance of the codon in the P-site as a start site, both by promoting GTP hydrolysis by eIF2 and by enhancing the release of eIF1. In effect, eIF5 acts to reduce the stringency of start-site selection whereas eIF1, together with eIF1A, help to maintain such stringency (see [5,11] for further discussion of this and other recent developments in this area).

Several lines of evidence indicate that eIF1, eIF1A and eIF5 mediate conformational changes within the 40S subunit which modulate start-site recognition. In brief, the TC is believed to bind to the 40S subunit such that the Met-tRNAiMet is not fully engaged with the P-site, in the scanning-competent state described above, which is conducive to start-site recognition. This is referred to as the Pout conformation. Appropriate base-pairing of the Met-tRNAiMet with, normally, an AUG codon leads to a conformational change to the closed Pin state, which involves displacement of eIF1 and movement of eIF5, allowing the release of the eIF2-bound Pi and leading to completion of start-site selection [12,13]. However, the details of these conformational changes and the potential role of other components remain to be clarified.

Like other ABC proteins, ABC50 binds ATP [1]. Expression of mutants of ABC50 which are defective for ATP binding or hydrolysis inhibits overall translation initiation [1]. However, it is unclear what role the hydrolysis of ATP by ABC50 might play during translation initiation. ABC50 belongs to the family of non-membrane-associated ABC proteins involved in protein synthesis, which also include eukaryotic elongation factor 3 (eEF3) (required for translation elongation in some fungi [14]) and RNAse L inhibitor type 1 (Rli1) ABCE1 (involved in post-termination recycling of ribosomes [15]) or in its regulation (general control non-depressing (Gcn20) in yeast [16]).

In the present study, we provide evidence that, in addition to impairing overall translation initiation, mutants of ABC50 that either cannot bind or cannot hydrolyse ATP also relax the accuracy of start-site selection, allowing relatively higher rates of use of certain non-AUG codons. However, they do not appear to affect the requirement for a correct initiation context. These data are consistent with the conclusion that ABC50 plays a role in ensuring correct start codon recognition during the initiation of mRNA translation.

EXPERIMENTAL

Cell culture and transfection

Human embryonic kidney (HEK) 293 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% (v/v) FBS (Invitrogen) and 1% penicillin/streptomycin. Transient transfections were carried out by calcium phosphate precipitation of 0.1–5 μg of DNA in N,N-bis[2-hydroxyethyl]-2-aminoethanesulfonic acid-buffered saline, pH 6.96, as previously described [1]. For siRNA transfections, 20 nM siRNA oligonucleotides were transfected with Lipofectamine RNAiMAX (Invitrogen) as per the manufacturer’s instructions. HeLa cells were maintained in DMEM containing Glutamax (Life Technologies) and 10% (v/v) FBS and dissociated from culture vessels with TrypLE Express (Life Technologies). Transfections of HeLa cells were carried out with GeneJuice (Merck), with a 3:1 ratio of GeneJuice to plasmid DNA.

Creation of the pICtest2 reporter vector

To obviate any problems with co-transfection efficiency of separate reporter plasmids, we constructed a dual-luciferase vector, which could be assayed quickly and easily, named pICtest2 (plasmid for initiation codon testing; Figure 1A), where the luciferase from the sea pansy Renilla reniformis (Rluc) would act as an internal control, always being translated from an AUG in the optimal Kozak consensus GCCACCAUGG (where the start codon is underlined). Transcription of the Renilla reporter mRNA is driven by the herpes simplex virus (HSV) thymidine kinase promoter. Within the same backbone, luciferase from the firefly (Photinus pyralis; Fluc) is expressed from a Simian virus 40 (SV40) promoter and it is the initiation codon of this open reading frame (ORF) which is altered. A flowchart of plasmid construction is shown in Supplementary Figure S1, with primers listed in Supplementary Table S1. In terms of translation efficiency, the untranslated regions surrounding the Fluc ORF are simple. Its 5′-UTR is 71 nt long, with a G-C content of 42.3%, both of which will allow efficient scanning. The 3′-UTR is 170 nt with 41.9% G-C content.

pICtest2 reporter vector design

Figure 1
pICtest2 reporter vector design

(A) The pICtest2 reporter plasmid has Fluc fused to a C-terminal PEST (peptide sequence rich in proline (P), glutamic acid (E), serine (S), and threonine (T)) domain under a separate promoter to Rluc. Different initiation codon contexts or alternative initiation codons are introduced at the beginning of Fluc, as detailed in the axis labels. (B) To determine whether the amino acid after the initiating methionine affects the turnover of the luciferase reporter, the four pICtest2 vectors containing the initiation codon consensus GCCAUGN were transfected into HeLa cells. After 48 h, but prior to harvesting the cells, 10 μg/ml cycloheximide (CHX) was added to the culture medium and the cells were maintained for the times indicated. Cells were then lysed and luciferase activity was determined, to measure the decay of PEST-fused luciferase. All values were normalized to the Fluc/Rluc activity at 48 h post transfection (i.e. t=0 cycloheximide) for each plasmid. (C) Testing the minimal Kozak consensus surrounding an AUG initiation codon. At 48 h after transfection into HeLa or HEK293 cells, the activities of the two luciferases were measured. Fluc values were normalized to those for Rluc. Results are expressed relative to the optimal GCCAUGG consensus sequence. OPP (opposite) denotes a UAC codon in place of the AUG codon, within the optimal Kozak consensus. Percentage incidence of the two indicated bases surrounding the annotated initiation codons within the human transcript dataset is shown below the histogram.

Figure 1
pICtest2 reporter vector design

(A) The pICtest2 reporter plasmid has Fluc fused to a C-terminal PEST (peptide sequence rich in proline (P), glutamic acid (E), serine (S), and threonine (T)) domain under a separate promoter to Rluc. Different initiation codon contexts or alternative initiation codons are introduced at the beginning of Fluc, as detailed in the axis labels. (B) To determine whether the amino acid after the initiating methionine affects the turnover of the luciferase reporter, the four pICtest2 vectors containing the initiation codon consensus GCCAUGN were transfected into HeLa cells. After 48 h, but prior to harvesting the cells, 10 μg/ml cycloheximide (CHX) was added to the culture medium and the cells were maintained for the times indicated. Cells were then lysed and luciferase activity was determined, to measure the decay of PEST-fused luciferase. All values were normalized to the Fluc/Rluc activity at 48 h post transfection (i.e. t=0 cycloheximide) for each plasmid. (C) Testing the minimal Kozak consensus surrounding an AUG initiation codon. At 48 h after transfection into HeLa or HEK293 cells, the activities of the two luciferases were measured. Fluc values were normalized to those for Rluc. Results are expressed relative to the optimal GCCAUGG consensus sequence. OPP (opposite) denotes a UAC codon in place of the AUG codon, within the optimal Kozak consensus. Percentage incidence of the two indicated bases surrounding the annotated initiation codons within the human transcript dataset is shown below the histogram.

Plasmids and siRNA

pCMV5–HA (haemagglutinin)–ABC50 plasmids were created as described in [1,17]. ABC50 siRNA was obtained as a siGENOME SMARTpool from Thermo Scientific containing four sequences that target throughout the protein coding region of the mRNA (Supplementary Table S2). As a negative control, siGENOME Non-Targeting siRNA pool #1 was used. The ORF of eIF1 was cloned into pcDNA3.1(+) from Invitrogen with both amplicon and vector cut with NheI and XhoI (Supplementary Table S3). F-box/LRR-repeat protein 3 (FBXL3) cDNA was purchased from the IMAGE clones collection (Open Biosystems). The whole 5′-UTR and a portion of the ORF was amplified and NheI and XhoI restriction sites (Supplementary Table S3) introduced to allow insertion in frame with a C-terminal 3×FLAG tag, as previously described [18]. The QuikChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies) was used to increase initiation at the predicted non-AUG codon through conversion from a GUG to an AUG (Supplementary Table S3).

Immunoblotting and antibodies

Cells were harvested in ice-cold lysis buffer as previously described [1] and subjected to SDS/PAGE, electrophoretic transfer to nitrocellulose membranes and the following antibody incubations, scanned using a LI-COR Odyssey imaging system. Antibodies were purchased as follows: anti-HA from Roche Applied Science, anti-ABC50 from Aviva Biosciences, anti-actin, anti-FLAG and α-tubulin from Sigma, anti-eIF5, anti-eIF2α and anti-DAP5 from Cell Signaling Technology, anti-eIF1A from Abcam and anti-RpS6 from Santa Cruz Biotechnology. The anti-eIF1 antibody was a gift from Dr Ariel Stanhill at Technion, Haifa, Israel.

Cell Titer-Glo and dual-luciferase assays

Both Cell Titer-Glo and dual-luciferase assays were performed as per the manufacturer's instructions and analysed with the GloMax®-Multi+ Detection System, all from Promega. To accurately assess changes in Fluc translation from altered start codons/contexts, we first normalized data to the internal Rluc to correct for transfection efficiency. These data were then normalized to a vector containing the Fluc reporter with an AUG in optimal context (gccAUGg) for each condition, to rule out any alterations caused by transcription from alternative promoters. To determine turnover of luciferase, 10 μg/ml cycloheximide was added to the culture medium for the times indicated prior to cell lysis.

mRNA expression

Total RNA from HEK293 cells was isolated using TRIzol-chloroform extraction and ethanol precipitation. One microgram of total RNA was transcribed into cDNA using the ImProm-II™ Reverse Transcription System (Promega). Quantitative PCR analysis was performed using Precision qPCR SYBR Green mastermix (PrimerDesign) and the ABI StepOnePLUS system (Applied Biosystems). Primer sets were designed and validated with PrimerDesign. Relative quantification was calculated using the 2−ΔΔCTmethod with β2-microglobulin as control for death-associated protein 5 (DAP5) and Fluc normalized to Rluc.

Sucrose density gradient centrifugation

HEK293 cells were lysed in 300 μl of 50 mM HEPES/KOH, pH 7.6, 7 mM MgCl2 and 100 mM KCl, containing 0.1% (v/v) Triton X-100 and protease inhibitors. After clearing, the lysate was incubated with 30 mM EDTA for 10 min on ice before layering on to a 10%–30% (w/v) sucrose gradient containing 30 mM Tris/HCl (pH 7.5), 100 mM NaCl and 30 mM EDTA and centrifuged using a Beckman SW41 rotor for 4 h at 234000 g. Alternatively, the lysate was incubated with 15 k-units/ml micrococcal nuclease (New England Biolabs) with 1 mM CaCl2 at 30°C for 15 min. The nuclease was then inhibited by the addition of 2 mM EDTA, before layering on to a 10%–30% (w/v) sucrose gradient as above. Fractions were collected while monitoring the absorbance at 254 nm followed by either (i) trichloroacetic acid (TCA) precipitation, with samples resolved by SDS/PAGE for immunoblotting analysis or (ii) RNA isolation using phenol/chloroform and separated on formaldehyde/agarose gels for visualization with GelRed (Biotium).

Statistical significance

Significance was assessed by Student's t test and P-values are denoted throughout as *P>0.05; **P>0.01; ***P>0.001.

RESULTS AND DISCUSSION

Characterization of pICtest2 vectors

As ABC50 helps to stabilize the binding of Met-tRNAiMet to eIF2 [2] and the anti-codon of this tRNA recognizes the start codon in the mRNA, we investigated whether mutations in ABC50 affected start-site selection.

It was important to create a suite of vectors that could be used to analyse the effects of ABC50 mutants on start codon selection. Previous reports in mammalian systems have used separate luciferase reporters to examine initiation codon selection [19] but we wished to develop a bicistronic pICtest2 vector (Figure 1A), encoding Fluc and Rluc driven by separate promoters, so as to avoid problems which can arise from inefficient or variable co-transfection efficiencies when using separate vectors. Rluc has an AUG in an optimal context whereas the nature of the initiation codon and its context for the Fluc cistron are varied. Building on previous approaches, but in order to allow rapid changes in translation to be monitored, we fused the Fluc cistron to a cDNA sequence encoding a PEST domain (SSGTRHGFPPEVEEQAAGTLPMSCSQESGMDRHPAACASARINV) [20]. The presence of this domain at the C-terminus destabilizes the Fluc resulting in a t1/2 of ∼2 h (Figure 1B) compared with the established value of 3 h for the wild-type protein [20].

The chemistry of the dual-luciferase assay means that both enzymes can be rapidly assayed in the same sample and the high sensitivity allows us to evaluate even relatively inefficient initiation. Furthermore, the use of Fluc as our reporter means that N-terminally truncated enzymes generated by leaky scanning are very unlikely to be measured in the assay, since loss of the first ten amino acids reduces the level of Fluc activity to <1% [21] and the next in-frame AUG is at the 30th codon position, with the only near-cognate codon in a favourable context a CUG at the 17th position. Thus, values truly reflect initiation efficiency at the codon of interest. We also established a negative control, ‘OPP’, where the initiation codon was replaced by UAC as this triplet is the ‘opposite’ of an AUG and would be extremely unlikely to base pair with the UAC anti-codon of the initiating Met-tRNAiMet. This was used to establish a background threshold, below which activity of the potential initiation codon was considered to be nil. Consistent with this, stop codons (UGA, UAG or UAA) in the initiation position also produced similar negligible values (results not shown).

To establish the relative efficiency of initiation codon contexts, we varied the simplest minimal Kozak consensus context (RCCAUGR; where R indicates a purine), which was preceded in our reporter vector by a run of three U bases to remove any influence of purines at positions −7 to −9. Previous work had established the bases at −3 and +4 as those which most strongly influence the efficiency of translation [22] and first we sought to confirm these findings. Altering the +4 base also changes the amino acid immediately after the initiating methionine. If lysine (A at +4) is the amino acid, then there is a noticeable increase in the turnover of the reporter protein following cycloheximide addition, compared with the other three versions, with glutamic acid (+4=G), glutamine (+4=C) or leucine (+4=U) (Figure 1B). This is consistent with the observation that lysine is a particularly destabilizing residue based on the rate of turnover of the β–galactosidase reporter in Saccharomyces cerevisiae. However, the results from these authors' experiments using mammalian reticulocytes were less clear [23]. Given that the version of luciferase we use in the present study is already destabilized at the C-terminus, it is possible that our reporter system is therefore more susceptible than others previously used to the effects of the presence of lysine at position 2 [19].

Combinations of each possible nucleotide at the −3 and +4 positions were analysed following transfection of our pICtest2 variants into either HeLa or HEK293 cells (Figure 1C). In all cases, the C-C doublet at positions −2 and −1 was maintained and we considered GCCAUGG as the benchmark as it is the most represented combination in annotated initiation codons in the Ensembl database of human transcripts; 2475 compared with 1444 occurrences of ACCAUGG in 36711 transcripts. The percentage incidence of the two indicated bases surrounding the annotated initiation codons within the human transcript dataset is shown below each combination in Figure 1(C).

As expected, the presence of the G at +4 is the most important feature in determining whether an initiation codon is recognized. In reports with this context, the presence of a pyrimidine base at +4 results in a reduction in the efficiency of translation, as also shown in previous work [24]. This is not statistically significant in the HeLa cell assay, although it is in HEK293 cells. More importantly, in reporters where the +4 base is anything other than G, the presence of a purine at −3 becomes much more important. With an A at +4 and a purine at −3, translation is reduced to around 40% of the optimal consensus, which implies that 60% of scanning ribosomes do not recognize the initiation codon and presumably continue scanning downstream. As discussed above, this could also be due to steady-state levels of Fluc caused by the presence of a lysine at the second amino acid. However, when the +4A is coupled with a pyrimidine at −3, then translation of Fluc drops further to around 20% that of the optimal consensus. In the least efficient combination, +4U/−3U, initiation occurs with only 5% of the efficiency of the optimal sequence.

We also used our pICtest2 reporter system to analyse the efficiency of near-cognate non-AUG initiation. To do this, we compared the extended full Kozak consensus with that proposed for non-AUG initiation by Wegrzyn et al. [25]. Their sequence (CGCGUCGCGxxxG) was determined by comparing the context surrounding 43 published non-AUG initiation codons and has not been functionally tested, to our knowledge, in such an assay. We compared the two contexts to each other, with initiation codons being AUG or one of the nine near-cognate initiation codons, which differ by one nucleotide at each position of the triplet (Table 1). As previously stated, we considered a UAC codon as a negative control so that only luciferase activity greater than detected from this reporter was considered as being from a full-length protein. In the vast majority of cases, translation from the near-cognate codons was barely detectable, in either the Kozak or the Wegrzyn consensus (Table 1), although all except AAG exhibited Fluc expression that was significantly higher than the UAC/OPP control in at least one context and one cell line. Furthermore, CUG and GUG initiation codons in a good context initiate with an efficiency that is similar to, if not better than, values obtained for an AUG in a weak context (compare Table 1 with Figure 1C). The only initiation codon that initiates translation more frequently when in the Wegrzyn context compared with the Kozak context is an ACG with both CUG and GUG codons weaker when in this sequence.

Table 1
Comparison of non-AUG initiation from alternative reporter vectors

RRL, rabbit reticulocyte lysate; WGE, wheat germ extract; AA, amino acid; ND, not detectable.

Author Peabody [26Kozak [24Ivanov [19Present study 
System RRL WGE COS cells HEK293 cells HeLa cells HEK293 cells 
Reporter Mouse DHFR Mouse DHFR Preproinsulin Fluc FlucP FlucP 
Context AUCNNNAUCNNNUCCACCNNNGCCACCNNNGCCACCNNNACGUCGCGNNNGCCACCNNNACGUCGCGNNN
second AA Valine Valine Alanine Glutamic acid Glutamic acid Glutamic acid 
AUG 100 100 100 100 100 (0.00)*** 111.6 (25.74)** 100 (0.00)*** 111.4 (12.61)** 
CUG 82 36 ‘Faint band’ 19.5 12.76 (1.53)** 9.41 (3.30)** 11.38 (1.92)** 5.63 (0.82)** 
GUG 36 3%–5% 9.2 3.71 (0.46)** 2.02 (0.90)* 2.64 (1.05)* 0.91 (0.14)** 
UUG 39 10 ND 1.9 0.36 (0.06)** 0.64 (0.20)** 0.20 (0.03)** 0.39 (0.09)* 
AAG 14 – 0.2 0.23 (0.09) 0.14 (0.05) 0.01 (0.01) 0.02 (0.01) 
ACG 84 45 ‘Faint band’ 6.6 0.16 (0.03)** 4.06 (1.20) 0.07 (0.03) 1.93 (0.30)** 
AGG 17 – 0.1 0.19 (0.05) 0.13 (0.05) 0.03 (0.01) 0.02 (0.01) 
AUA 59 30 – 3.3 0.84 (0.20)*** 0.35 (0.11)* 0.53 (0.12)* 0.21 (0.06)* 
AUC 47 17 – 1.7 0.35 (0.01) 0.25 (0.05)* 0.06 (0.02)** 0.13 (0.06) 
AUU 67 14 – 3.2 1.12 (0.30)* 0.77 (0.24) 0.58 (0.03)*** 0.43 (0.08) 
UAC (OPP) – – – – 0.11 (0.00) – 0.03 (0.00) – 
Author Peabody [26Kozak [24Ivanov [19Present study 
System RRL WGE COS cells HEK293 cells HeLa cells HEK293 cells 
Reporter Mouse DHFR Mouse DHFR Preproinsulin Fluc FlucP FlucP 
Context AUCNNNAUCNNNUCCACCNNNGCCACCNNNGCCACCNNNACGUCGCGNNNGCCACCNNNACGUCGCGNNN
second AA Valine Valine Alanine Glutamic acid Glutamic acid Glutamic acid 
AUG 100 100 100 100 100 (0.00)*** 111.6 (25.74)** 100 (0.00)*** 111.4 (12.61)** 
CUG 82 36 ‘Faint band’ 19.5 12.76 (1.53)** 9.41 (3.30)** 11.38 (1.92)** 5.63 (0.82)** 
GUG 36 3%–5% 9.2 3.71 (0.46)** 2.02 (0.90)* 2.64 (1.05)* 0.91 (0.14)** 
UUG 39 10 ND 1.9 0.36 (0.06)** 0.64 (0.20)** 0.20 (0.03)** 0.39 (0.09)* 
AAG 14 – 0.2 0.23 (0.09) 0.14 (0.05) 0.01 (0.01) 0.02 (0.01) 
ACG 84 45 ‘Faint band’ 6.6 0.16 (0.03)** 4.06 (1.20) 0.07 (0.03) 1.93 (0.30)** 
AGG 17 – 0.1 0.19 (0.05) 0.13 (0.05) 0.03 (0.01) 0.02 (0.01) 
AUA 59 30 – 3.3 0.84 (0.20)*** 0.35 (0.11)* 0.53 (0.12)* 0.21 (0.06)* 
AUC 47 17 – 1.7 0.35 (0.01) 0.25 (0.05)* 0.06 (0.02)** 0.13 (0.06) 
AUU 67 14 – 3.2 1.12 (0.30)* 0.77 (0.24) 0.58 (0.03)*** 0.43 (0.08) 
UAC (OPP) – – – – 0.11 (0.00) – 0.03 (0.00) – 

Although our results are similar to those of Ivanov et al. [19], who also used a Fluc reporter in a Kozak consensus in HEK293 cells, they are very different from the hierarchy and efficiency of non-AUG initiation reported by Peabody [26], plus a previous work by Kozak [24] (Table 1). That study used densitometric scanning of products from dihydrofolate reductase (DHFR) reporters translated in either rabbit reticulocyte lysates (RRL) or wheat germ extract (WGE) and ACG was reported as being 85% and 45% as efficient as an AUG in the respective in vitro translation systems. Overall, our results suggest that, whereas initiation at near cognate codons is possible, the efficiency of translation may vary depending on the reporter used and thus further surrounding sequences must contribute to efficiency of selection. Evidence for this comes from the large scale ribosomal profiling of initiation codon usage in endogenous transcripts in mouse embryonic stem (ES) cells [27], which demonstrated that the AUG is only used in ∼43% of initiation events and CUG in ∼15%, the reasons for this large amount of non-AUG translation is still under study. The present data establish our reporter system as being a valid tool for measuring the efficiencies of translation initiation in differing contexts and with different initiation codons. Having created a suitable set of vectors, we could now explore the effect of specific mutations in ABC50 on start-site choice.

Mutations in ABC50 cause decreased stringency against the use of non-AUG start codons

We previously showed that expressing mutants of ABC50, where either ATP binding or hydrolysis is impaired, caused inhibition of overall protein synthesis in HEK293 cells, causing a loss of polysomes which is a characteristic of impaired translation initiation [1]. These mutants affect both nucleotide-binding domains (NBDs) in ABC50 (Figure 2A), but in different ways: in ABC50[K304M/K626M], key lysine residues in Walker box A of the NBDs have been mutated to prevent the binding of ABC50 to ATP [1,28,29] whereas in ABC50[E439Q/E730Q], residues in the Walker box B motifs that are required for ATP hydrolysis have been changed [30]. We have previously shown that the NBDs of ABC50 are not needed for its interaction with eIF2 [17] that both mutants can still bind ribosomes and that expressing ABC50[E439Q/E730Q] inhibited protein synthesis and polysome aggregation more strongly than ABC50[K304M/K626M] [1]. The observed decrease in protein synthesis did not activate stress pathways through the phosphorylation of eIF2 in either case (results not shown). Since ABC50 stabilizes the binding of Met-tRNAMeti to eIF2 [2] and the anti-codon of this tRNA recognizes the start codon in the mRNA, we asked whether these mutations in ABC50 affected start-site selection, using our pICtest2 reporter plasmids in HEK293 cells.

Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation from non-optimal start codons

Figure 2
Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation from non-optimal start codons

(A) Schematic illustration of the ABC50 protein showing the location of the main domains and the mutations studied in the present work. The small squares labelled ‘A’ and ‘B’ indicate the Walker A and B boxes of its ABC domains.(B) The left section shows an immunoblot of lysates from untransfected or cells transfected with wild-type ABC50, analysed with anti-ABC50. Different protein amounts were loaded to achieve 297-fold (±14.9) increase in ABC50 expression upon transfection. The right section shows an immunoblot of lysates from cells transfected with the indicated vectors, analysed using anti-HA to illustrate equal transfection of wild-type and mutant ABC50 plasmids. (C) pICtest2 reporter plasmids with Fluc under alternative initiation codons as shown were co-transfected into HEK293 cells together with the indicated ABC50 expression vectors. (D) mRNA expression for both Fluc and Rluc was measured using quantitative PCR (qPCR) following co-transfection of the pICtest2 reporter with either wild-type ABC50 or E439Q/E730Q mutant ABC50. (EG) The context around the AUG initiation codon for Fluc was altered as shown and the plasmids co-transfected with the indicated ABC50 expression vectors. (C) and (EG), Fluc activity was normalized to Rluc activity and then compared with the control (GCCAUGG) plasmid data for each sample set. Results represent means ± S.D. from three independent experiments, each performed in triplicate.

Figure 2
Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation from non-optimal start codons

(A) Schematic illustration of the ABC50 protein showing the location of the main domains and the mutations studied in the present work. The small squares labelled ‘A’ and ‘B’ indicate the Walker A and B boxes of its ABC domains.(B) The left section shows an immunoblot of lysates from untransfected or cells transfected with wild-type ABC50, analysed with anti-ABC50. Different protein amounts were loaded to achieve 297-fold (±14.9) increase in ABC50 expression upon transfection. The right section shows an immunoblot of lysates from cells transfected with the indicated vectors, analysed using anti-HA to illustrate equal transfection of wild-type and mutant ABC50 plasmids. (C) pICtest2 reporter plasmids with Fluc under alternative initiation codons as shown were co-transfected into HEK293 cells together with the indicated ABC50 expression vectors. (D) mRNA expression for both Fluc and Rluc was measured using quantitative PCR (qPCR) following co-transfection of the pICtest2 reporter with either wild-type ABC50 or E439Q/E730Q mutant ABC50. (EG) The context around the AUG initiation codon for Fluc was altered as shown and the plasmids co-transfected with the indicated ABC50 expression vectors. (C) and (EG), Fluc activity was normalized to Rluc activity and then compared with the control (GCCAUGG) plasmid data for each sample set. Results represent means ± S.D. from three independent experiments, each performed in triplicate.

In cells transfected with an empty vector or one encoding wild-type ABC50 (Figure 2B), the expression of Fluc was much lower from the vectors where it initiates from a non-AUG codon, in the order (of decreasing expression) CUG>GUG>ACG, indicating discrimination against non-AUG initiation codons as expected (Figure 2C). Strikingly, expressing either ABC50 mutant led to a smaller reduction in Fluc expression from the GUG and CUG variants of the vector; it should be noted that the data are normalized to the Rluc as an internal control for any effects on overall initiation from a strong start codon (Figure 2C). The observed effects were not due to changes in mRNA levels in either Fluc or Rluc (Figure 2D), strongly suggesting effects on translation. Fluc expression from the ACG variant remained low for the ABC50[K304M/K626M] mutant but the decrease in its expression relative to the AUG version was less marked in cells expressing ABC50[E439Q/E730Q]. Thus, expressing ABC50 mutants which either cannot bind ATP or which can bind ATP but cannot hydrolyse it relaxes the discrimination against non-AUG codons, in addition to decreasing overall translation initiation [1]. In contrast, expressing wild-type ABC50 did not affect start codon choice.

Mutations in ABC50 promote the use of non-optimal initiation context start sites

Efficient translation initiation depends not only on the nature of the start codon but also on the local sequence context [22]. The Kozak consensus sequence was proposed based on the frequency of each nucleotide at each position.

Since the above data indicate that ABC50 plays a role in the stringency of initiation codon selection, it was of interest to assess whether it also influenced the use of codons within non-optimal initiation contexts. To address this, we used eight different vectors in which the codon of the Fluc cistron either conforms to the most represented human consensus (GCCAUGG, where the initiation codon is underlined) or where this has been altered. In each case, the vector again contains the Rluc cistron with an AUG codon in optimal Kozak consensus. Altering the nucleotide at the −3 position had modest effects on expression of the reporter protein (Figure 2E), whereas alteration of the +4 nucleotide decreased translation in all cases (Figure 2F), consistent with the accepted consensus [31] and our results in Figure 1(C). Expressing the ABC50[K304M/K626M] or ABC50[E439Q/E730Q] mutants did not affect initiation from start codons with single nucleotide changes in the Kozak consensus sequence (Figures 2E and 2F). Altering multiple nucleotides around the AUG, in cells expressing wild-type ABC50 or empty vector, dramatically reduced translation of Fluc (Figure 2G). However, cells expressing either of the ABC50 mutants, showed a weaker decrease on Fluc translation relative to Rluc (Figure 2G). The fact that both ABC50[K304M/K626M] and ABC50[E439Q/E730Q] mutants show similar effects suggests that, in this case, ATP binding may be the important parameter. Taken together these data suggest that, whereas ATP hydrolysis by ABC50 exerts a strong influence on the usage of non-AUG start codons, it does not influence the recognition of the Kozak consensus sequence in an analogous way.

Defective ATP hydrolysis by ABC50 increases translation from an upstream GUG

Many mRNAs contain more than one potential upstream start codon in-frame with the main ORF [27,32], although the potential upstream start codon is often not AUG, but a related codon such as GUG [33]. To investigate the influence of ABC50 mutants on the translation of such an mRNA, we used a vector which contains two possible in-frame start sites (Figure 3A). Deep sequencing profiling of ribosome-bound mRNAs frozen at the point of initiation indicated the use of an upstream GUG in FBXL3 [32]. We subsequently confirmed this upstream GUG codon using a bioinformatics pipeline search for the presence of upstream near-cognate initiation codons (Joanne L. Cowan and Mark J. Coldwell, unpublished work). We placed the 5′-UTR and first 214 amino acids of the coding region of FBXL3 upstream of a triple FLAG tag in a reporter vector used previously [18] (Figure 3A). Expression of this vector resulted in detection of two protein isoforms with molecular masses of 34 and 27 kDa, as predicted from the respective GUG and AUG initiation codons (Figure 3B). Mutating the upstream GUG, to an AUG increased translation of the upper 34 kDa isoform as predicted and in doing so prevents leaky scanning to the downstream AUG (Figure 3B). These results confirm that (i) initiation can occur from either codon, (ii) the lower band is not a degradation product of the upper band, and (iii) the upper band is not a post-translationally modified form of the lower band.

Mutating residues in ABC50 required for ATP binding or hydrolysis by ABC50 alters translation of the upstream non-AUG start codons

Figure 3
Mutating residues in ABC50 required for ATP binding or hydrolysis by ABC50 alters translation of the upstream non-AUG start codons

(A) A reporter system based on the 5′-UTR and the first 214 amino acids of the annotated coding region of the FBXL3 mRNA was created; the encoded proteins have a triple FLAG tag at their C-termini. Initiation from the upstream GUG results in a longer isoform of the protein at 34 kDa (p34), whereas the annotated AUG produces a protein of 27 kDa (p27), distinguishable by immunoblot using anti-FLAG. The upstream GUG was also mutated to AUG. (B) Plasmids were transfected into HeLa cells for 48 h and proteins were detected by immunoblotting using anti-FLAG or -actin antibodies. (C) The FBXL3–FLAG vector was co-transfected into HEK293 cells with an ABC50 expression vector (wild-type or mutant). After 48 h, cells were harvested and samples were separated by SDS/PAGE. The expression levels of ectopic ABC50 were monitored using anti-HA; actin was analysed as a loading control. A representative immunoblot is shown along with a histogram showing the ratio of expression from the GUG/AUG start codon for the FBXL3 reporter. (D) Representative FLAG immunoblot showing expression from the FBXL3–FLAG vector when co-expressed with a truncated form of ABC50 containing only the N-terminus or a full-length phosphorylation-deficient mutant (S109A/S140A). For (C) and (D), results represent means ± S.D. from three independent experiments. The expression levels of ABC50 proteins were monitored using anti-HA, with actin used as a loading control.

Figure 3
Mutating residues in ABC50 required for ATP binding or hydrolysis by ABC50 alters translation of the upstream non-AUG start codons

(A) A reporter system based on the 5′-UTR and the first 214 amino acids of the annotated coding region of the FBXL3 mRNA was created; the encoded proteins have a triple FLAG tag at their C-termini. Initiation from the upstream GUG results in a longer isoform of the protein at 34 kDa (p34), whereas the annotated AUG produces a protein of 27 kDa (p27), distinguishable by immunoblot using anti-FLAG. The upstream GUG was also mutated to AUG. (B) Plasmids were transfected into HeLa cells for 48 h and proteins were detected by immunoblotting using anti-FLAG or -actin antibodies. (C) The FBXL3–FLAG vector was co-transfected into HEK293 cells with an ABC50 expression vector (wild-type or mutant). After 48 h, cells were harvested and samples were separated by SDS/PAGE. The expression levels of ectopic ABC50 were monitored using anti-HA; actin was analysed as a loading control. A representative immunoblot is shown along with a histogram showing the ratio of expression from the GUG/AUG start codon for the FBXL3 reporter. (D) Representative FLAG immunoblot showing expression from the FBXL3–FLAG vector when co-expressed with a truncated form of ABC50 containing only the N-terminus or a full-length phosphorylation-deficient mutant (S109A/S140A). For (C) and (D), results represent means ± S.D. from three independent experiments. The expression levels of ABC50 proteins were monitored using anti-HA, with actin used as a loading control.

In cells co-transfected with wild-type ABC50 or with an empty vector, both start sites in the FBXL3 reporter are used, generating the two possible FLAG-tagged products in an approximate ratio of 2:1 for initiation at GUG compared with AUG (Figure 3C). Data from cells where the ABC50[K304M/K626M] mutant was overexpressed show a trend towards increased translation from the upstream GUG, but this effect is not statistically significant. However, overexpression of the ABC50[E439Q/E730Q] mutant significantly increased expression from the upstream GUG (Figure 3C). Analysis of single mutants revealed a similar effect, with point mutations in the Walker B domain again showing a stronger effect than the individual Walker A mutants (Figure 3C). These data further demonstrate that ATP binding and/or hydrolysis by ABC50 can modulate the selection of non-AUG codons as start sites.

The NBDs of ABC50 are not needed for its interaction with eIF2 [17] and indeed the N-terminal region suffices for this. However, although the N-terminal domain of ABC50 can still bind eIF2, it is not sufficient to interact with ribosomes [17]. Overexpression of a truncated form of ABC50, expressing only the N-terminal portion and lacking both ABC domains, did not affect initiation codon selection (Figure 3D), supporting the idea that it is the ATP-binding domains of ABC50 which are important for this role. Although there is a trend towards an increase in FLAG expression compared with cells overexpressing wild-type ABC50, this was not significant. We have previously identified two phosphorylation sites in the N-terminal portion of ABC50 that may play an important role in its function [17]. These serine residues (Ser109 and Ser140) are not required for the interaction of ABC50 with eIF2 [17]. Furthermore, the data in Figure 3(D) show that these mutations have no effect on translation from the FBXL3 reporter. It should be noted that immunoblotting for the HA tag on ABC50 confirmed similar expression levels of the wild-type and mutant ABC50 proteins. These data indicate that mutating the ABC domain relaxes the bias against the use of GUG for the initiation of translation, perhaps by affecting the stringency with which start sites are monitored. In comparison, neither the N-terminal domain of ABC50 nor mutations of the phosphorylation sites within full-length ABC50 affected this (Figure 3D). This suggests that the ABC domain mutants exert a dominant interfering effect, which is not overcome by the endogenous ABC50 present within the cells.

Initiation of specific mRNAs from non-canonical codons in mammalian cells has become more widely accepted [26]. One such example is DAP5/eIF4G2 (also known as p97/NAT1 (N-acetyltransferase)), a member of the eIF4G family whose translation initiates solely from a GUG initiation codon in a good context (AAAGUGG) (Figure 4A) [3436]. eIF4GI is the archetypal member of the family and can be translated from multiple AUG initiation codons [37,38] and eIF4GII is translated from CUG and AUG initiation codons [18]. To determine whether the ABC50 mutants affected endogenous mRNA translation from non-canonical codons, in addition to the reporters already used, we investigated the expression of the DAP5 protein, using immunoblotting. As shown in Figure 4(B), the ABC50[K304M/K626M] and the ABC50[E439Q/E730Q] mutants each significantly increased the expression of the DAP5 protein. No changes in DAP5 mRNA levels were observed with any of the ABC50 vectors indicating the effect is probably at the level of translation (Figure 4C). The role of DAP5 in mediating both cap-dependent and IRES-driven translation of specific mRNAs has been linked to cell survival during both mitosis and endoplasmic reticulum stress [3941]. The importance of ABC50 under these conditions requires further investigation.

Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation of DAP5

Figure 4
Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation of DAP5

(A) Schematic representation of members of the eIF4G family of proteins, showing conserved domains (MIF4G, MA3 and W2) and interaction sites with poly(A)-binding protein (PABP) and eIF4E. Isoforms arising from the use of alternative initiation codons are indicated by arrows and a–f. (B) Representative immunoblot showing endogenous DAP5 expression in HEK293 cells expressing the ABC50 variants indicated. (C) HEK293 cells were transfected with the ABC50 expression vectors shown. Using qPCR DAP5 mRNA levels were measured relative to β2-microglobulin. For (B) and (C), results represent means ± S.D. from three independent experiments.

Figure 4
Mutating residues in ABC50 required for ATP binding or hydrolysis alters translation of DAP5

(A) Schematic representation of members of the eIF4G family of proteins, showing conserved domains (MIF4G, MA3 and W2) and interaction sites with poly(A)-binding protein (PABP) and eIF4E. Isoforms arising from the use of alternative initiation codons are indicated by arrows and a–f. (B) Representative immunoblot showing endogenous DAP5 expression in HEK293 cells expressing the ABC50 variants indicated. (C) HEK293 cells were transfected with the ABC50 expression vectors shown. Using qPCR DAP5 mRNA levels were measured relative to β2-microglobulin. For (B) and (C), results represent means ± S.D. from three independent experiments.

Knocking down ABC50 decreases overall translation with no influence on start codon selection

It was possible that the effects of the ABC mutants reflected a decrease in the availability of functional eIF2–ABC50 complexes due to association of some of the eIF2 with mutant ABC50. To test this, we used siRNA against ABC50 to knockdown endogenous levels of ABC50 and asked whether this elicited a similar effect. Treating cells with siRNA directed against the ABC50 mRNA caused a marked decrease in ABC50 expression, whereas the scrambled negative control siRNA did not (Figure 5A). In control cells or cells treated with either the scrambled or anti-ABC50 siRNA, a similar ratio of the longer and shorter FBXL3 polypeptides was observed (Figure 5A). Knocking down ABC50 decreased the total levels of both isoforms, to the same extent, consistent with our earlier data showing that ABC50 is required for efficient translation initiation [1] (Figure 5A). Additionally, silencing ABC50 did not alter the expression of DAP5, confirming that silencing of ABC50 does not influence start-codon selection (Figure 5A).

Knocking down ABC50 decreases overall translation but does not alter start codon stringency

Figure 5
Knocking down ABC50 decreases overall translation but does not alter start codon stringency

(A) Multiple siRNAs against ABC50 (or a scrambled non-targeting control) were transfected into HEK293 cells; after 72 h the efficiency of knockdown was monitored by immunoblotting for ABC50. The FBXL3–FLAG vector was then transfected into cells silenced for ABC50; after a further 24 h cells were harvested and lysates were analysed by immunoblotting for the FLAG tag. Endogenous DAP5 expression was analysed using anti-DAP5 antibody. A representative immunoblot is shown along with quantification histograms for each antibody, the signals being normalized to actin. Results represent means ± S.D. from three independent experiments. (B) siRNA-treated HEK293 cells as in (A) were transfected with pICtest2 vectors containing Fluc under initiation from a non-canonical GUG or a severely altered consensus sequence as shown. Fluc activity was normalized to Rluc activity and then compared with the untreated optimal sequence control (gccAUGg) vector data for each sample set. Results represent means ± S.D. from three independent experiments, measured in triplicate. (C) At 72 h after transfection of the indicated siRNAs, cells were trypsinized and counted. (D) At 72 h after siRNA treatment, cells were replated into a 96-well plate and left for a further 24 h. Cell viability was assessed using the Cell Titer-Glo assay (Promega) and data were normalized to cell number. For (C) and (D), results represent means ± S.D. from three independent experiments, each normalized to its ‘mock’-transfected control (RNAiMAX only).

Figure 5
Knocking down ABC50 decreases overall translation but does not alter start codon stringency

(A) Multiple siRNAs against ABC50 (or a scrambled non-targeting control) were transfected into HEK293 cells; after 72 h the efficiency of knockdown was monitored by immunoblotting for ABC50. The FBXL3–FLAG vector was then transfected into cells silenced for ABC50; after a further 24 h cells were harvested and lysates were analysed by immunoblotting for the FLAG tag. Endogenous DAP5 expression was analysed using anti-DAP5 antibody. A representative immunoblot is shown along with quantification histograms for each antibody, the signals being normalized to actin. Results represent means ± S.D. from three independent experiments. (B) siRNA-treated HEK293 cells as in (A) were transfected with pICtest2 vectors containing Fluc under initiation from a non-canonical GUG or a severely altered consensus sequence as shown. Fluc activity was normalized to Rluc activity and then compared with the untreated optimal sequence control (gccAUGg) vector data for each sample set. Results represent means ± S.D. from three independent experiments, measured in triplicate. (C) At 72 h after transfection of the indicated siRNAs, cells were trypsinized and counted. (D) At 72 h after siRNA treatment, cells were replated into a 96-well plate and left for a further 24 h. Cell viability was assessed using the Cell Titer-Glo assay (Promega) and data were normalized to cell number. For (C) and (D), results represent means ± S.D. from three independent experiments, each normalized to its ‘mock’-transfected control (RNAiMAX only).

Further analysis using the luciferase expression vectors confirmed that silencing ABC50 did not alter expression of Fluc from either a GUG initiation codon or an AUG in a poor context, as compared with the scrambled siRNA control (Figure 5B). Consistent with the impairment of reporter protein synthesis caused by knocking down ABC50 (Figure 5A), both cell number (Figure 5C) and cell viability (Figure 5D) were decreased under this condition. Thus, although decreasing cellular levels of ABC50 does indeed impair protein synthesis, it does not affect the normal strong preference for AUG start codons in a good context. Paytubi et al. [17] have previously shown that silencing ABC50 decreased general protein translation as assessed by labelling with [35S] methionine. The effects on start codon preference therefore are not due to the unavailability of ABC50–eIF2 complex and the effect of ABC50 mutants on start-site choice may be through a dominant effect on overall PIC (plasmid for initiation codon (testing)) function in start-site selection.

Decreased start-codon stringency observed with ABC50 mutants can be restored by eIF1

It is widely accepted that eIF1 plays a role in maintaining start-site selection stringency [5,11]. We noted that overexpression of eIF1 decreased overall translation including that of the ABC50 expression vectors (Figure 6A). For analysis, we therefore focused on the change in ratio between the AUG- and GUG-initiated products for FBXL3. Interestingly, the increased translation from the upstream GUG codons in cells expressing ABC50 mutants was also decreased in the presence of eIF1, as shown by an increase in the AUG band of FBXL3, indicating that eIF1 still promotes start-site stringency under these conditions (Figures 6A and 6B). Similar data were also observed when using the dual-luciferase vectors (Figures 6C and 6D), with an even greater effect of eIF1 expression, most probably due to the increased sensitivity of this assay. Taken together, these data suggest that the impaired initiation codon fidelity observed with the ABC50[K304M/K626M] and ABC50[E439Q/E730Q] mutants is not caused by blocking the ability of eIF1 to exert its function.

Overexpression of eIF1 in the presence of ABC50 mutants restores start codon stringency

Figure 6
Overexpression of eIF1 in the presence of ABC50 mutants restores start codon stringency

(A) Expression of FBXL3 was analysed by immunoblotting for the FLAG tag in lysates from cells co-transfected with the indicated ABC50 expression vectors with or without a vector expressing eIF1. A representative immunoblot is shown in (A) together with histograms (B) showing the ratio of expression from the GUG/AUG for the FBXL3 reporter. Results represent means ± S.D. from three independent experiments. (C) Plasmids encoding Fluc with a GUG initiation codon were co-transfected into HEK293 cells with the indicated ABC50 expression vectors with and without eIF1 overexpression. Results represent mean relative Fluc/Rluc activity ± S.D. from three independent experiments, each performed in triplicate. (D) A representative immunoblot confirming expression of both HA–ABC50 vectors and eIF1 is shown.

Figure 6
Overexpression of eIF1 in the presence of ABC50 mutants restores start codon stringency

(A) Expression of FBXL3 was analysed by immunoblotting for the FLAG tag in lysates from cells co-transfected with the indicated ABC50 expression vectors with or without a vector expressing eIF1. A representative immunoblot is shown in (A) together with histograms (B) showing the ratio of expression from the GUG/AUG for the FBXL3 reporter. Results represent means ± S.D. from three independent experiments. (C) Plasmids encoding Fluc with a GUG initiation codon were co-transfected into HEK293 cells with the indicated ABC50 expression vectors with and without eIF1 overexpression. Results represent mean relative Fluc/Rluc activity ± S.D. from three independent experiments, each performed in triplicate. (D) A representative immunoblot confirming expression of both HA–ABC50 vectors and eIF1 is shown.

ABC50 mutants do not substantially alter translation initiation factors binding to the 40S subunit

Recruitment of the TC and other initiation factors to the 40S ribosome is essential for translation initiation [5]. We have previously shown that ABC50 binds to ribosomes and to eIF2 [1,2]. It was possible that ABC50 mutants relaxed the stringency of start-site choice by altering the association of the relevant translation initiation factors to the 40S subunit; for example, impaired recruitment of eIF1 might account for the effects observed in the present study. We therefore investigated the association of eIF1 and other initiation factors with the 40S subunit in the presence of wild-type or mutant ABC50. Cytoplasmic extracts from transfected HEK293 cells were separated by ultracentrifugation on linear sucrose gradients containing EDTA to disrupt polysomes (Figure 7A). The amount of 40S subunits was normalized by immunoblotting for the 40S protein rpS3 (ribosomal protein S3) (Figure 7B). Analysis of fractions containing the 40S subunits showed no alteration in the association with eIF2, eIF5, eIF1A or eIF1 with 40S particles in the presence of the ABC50 mutants (Figures 7B and 7C). In addition, disruption of polysomes into mRNA-bound monoribosomes by mild micrococcal ribonuclease treatment also showed no change in the distribution of eIF2 or eIF5 in cells overexpressing mutant or wild-type ABC50 (Figure 7D). Data for eIF1 were not possible to obtain for this treatment due to the low levels of the factor on the 40S and the sensitivity of the antibody. These data indicate that the effects of the ABC50 mutants on start-site selection are not a consequence of marked changes in the recruitment of key factors of the 40S complex but, more probably, of effects on the function of the assembled complex.

ABC50 mutants do not alter binding of initiation factors to the 40S subunit

Figure 7
ABC50 mutants do not alter binding of initiation factors to the 40S subunit

(A) Transfected HEK293 lysates were separated by a 10%–30% (w/v) sucrose density gradient in the presence of EDTA to dissociate polysomes; subsequently, 18 fractions were collected from each gradient while monitoring the absorbance at 254 nm. To confirm which fractions contained the 40S subunits, RNA was isolated from each fraction. 18S and 28S RNAs were separated on a formaldehyde/agarose gel and visualized with GelRed (fraction numbers are shown below each gel). Immunoblotting for eIF5 was used to confirm the distribution of the translation initiation factors. (B) Proteins were precipitated from each fraction by TCA and the 40S fractions pooled and then separated by SDS/PAGE and immunoblotted for the presence of the indicated translation initiation factors. (C) Ribosomal protein S3 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were analysed as controls for 40S subunits and non-ribosomal material respectively. The ratio of protein associated with 40S subunits compared with the total protein was quantified. Results represent means ± S.D. from three independent experiments. (D) Transfected HEK293 lysates were treated with micrococcal nuclease to disrupt polysomes and separated by a 10%–30% (w/v) sucrose density gradient as above. Proteins were precipitated from each fraction by TCA and the 40S fractions pooled and then separated by SDS/PAGE and immunoblotted for the presence of the indicated translation initiation factors.

Figure 7
ABC50 mutants do not alter binding of initiation factors to the 40S subunit

(A) Transfected HEK293 lysates were separated by a 10%–30% (w/v) sucrose density gradient in the presence of EDTA to dissociate polysomes; subsequently, 18 fractions were collected from each gradient while monitoring the absorbance at 254 nm. To confirm which fractions contained the 40S subunits, RNA was isolated from each fraction. 18S and 28S RNAs were separated on a formaldehyde/agarose gel and visualized with GelRed (fraction numbers are shown below each gel). Immunoblotting for eIF5 was used to confirm the distribution of the translation initiation factors. (B) Proteins were precipitated from each fraction by TCA and the 40S fractions pooled and then separated by SDS/PAGE and immunoblotted for the presence of the indicated translation initiation factors. (C) Ribosomal protein S3 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were analysed as controls for 40S subunits and non-ribosomal material respectively. The ratio of protein associated with 40S subunits compared with the total protein was quantified. Results represent means ± S.D. from three independent experiments. (D) Transfected HEK293 lysates were treated with micrococcal nuclease to disrupt polysomes and separated by a 10%–30% (w/v) sucrose density gradient as above. Proteins were precipitated from each fraction by TCA and the 40S fractions pooled and then separated by SDS/PAGE and immunoblotted for the presence of the indicated translation initiation factors.

CONCLUSIONS

The present data show that crippling the ability of ABC50 to bind or hydrolyse ATP relaxes the normally tight stringency of start-site selection, allowing the use of non-AUG codons. Interestingly, the role of ATP hydrolysis by ABC50 appears to be restricted to the nature of the start codon itself, rather than the adjacent sequence, since ABC50 mutants did not relax the need for a correct Kozak consensus. However, data from siRNA-mediated knockdown experiments indicate that the availability of ABC50 itself does not affect start-site fidelity. It is possible that the lack of effect of knocking down ABC50 on start-site selection reflects redundancy with another member(s) of the ABC family, although ABC50 shows very limited similarity to other ABC proteins outside its ABC domains (Supplementary Figure S2A). As knocking down its expression does markedly impair overall translation initiation [1], ABC50 may have two distinct functions, an essential role in overall translation and a separate role in start-site choice that was revealed here by the dominant effects of the mutants we have used in the present study. When ABC50 mutants are present in excess over wild-type (endogenous) ABC50, it is possible that wild-type ABC50 can still promote PIC formation and become incorporated into them. However, the mutant ABC50 may act on the PIC, either as a second copy perturbing the equilibrium between open and closed states. Alternatively, mutant ABC50 may itself become incorporated into the PIC and interfere with its function.

Since ATP hydrolysis by ABC proteins is widely associated with mechanical action [42,43], ABC50 may be involved in conformational alterations, perhaps the Pout → Pin change in the P-site during start-site selection [12]; this is consistent with the finding that such mutants do not affect the actual association with 40S ribosomal subunits of proteins involved in start-site recognition such as eIF2, eIF1 and eIF5. However, since the nature of this change is poorly understood, it is not yet possible to test this hypothesis. ABC50 mutants may perturb the interactions between the components involved in start-site recognition, which are now known to involve a multifactor complex in mammals [44] as well as in yeast where it was initially discovered [45]. Further work is needed to study how ABC50 mutants influence start codon selection.

These studies therefore identify ABC50 as influencing start-site choice in mammalian cells. Although all ABC proteins show strong homology in the ABC regions, it is their N-terminal domains which provide the specificity for their function. In fact the N-terminal portion of Gcn20 was found to be both necessary and sufficient for its complex formation with Gcn1 and subsequent activation of Gcn2 [46]. In agreement with this, it is the N-terminal domain of ABC50 that interacts with its partner eIF2, although the ABC domains are required for interaction with the ribosomes [17]. Sequence alignments confirm that the N-terminal domain of ABC50 is not orthologous to any of the human homologues of the yeast ABC proteins known to be involved in translation (Supplementary Figure S2A). This is consistent with it playing an alternative role in start-site choice as revealed by the present data. Additionally, the N-terminal domain of ABC50 only shows high conservation within mammals (Supplementary Figure S2B), suggesting its role may have developed with increased species complexity.

Abbreviations

     
  • ABC

    ATP-binding cassette

  •  
  • DHFR

    dihydrofolate reductase

  •  
  • DMEM

    Dulbecco's modified Eagle's medium

  •  
  • eIF

    eukaryotic initiation factor

  •  
  • Fluc

    firefly luciferase

  •  
  • HEK

    human embryonic kidney

  •  
  • IRES

    internal ribosome entry site

  •  
  • Met-tRNAiMet

    methionyl-tRNA

  •  
  • NBD

    nucleotide-binding domain

  •  
  • ORF

    open reading frame

  •  
  • Rluc

    Renilla luciferase

  •  
  • TC

    ternary complex

  •  
  • TCA

    trichloroacetic acid

AUTHOR CONTRIBUTION

Joanna Stewart designed and performed experiments, analysed data and wrote the paper. Joanne Cowan and Lisa Perry performed experiments, analysed data and edited the paper before submission. Christopher Proud and Mark Coldwell supervised all aspects of the present work and wrote the paper. All authors were involved in the research design and discussion and approved the paper.

FUNDING

This work was supported by the U.K. Biotechnology and Biological Sciences Research Council [grant numbers BB/H006834/1 (to M.J.C.) and BB/J007706/1 (to C.G.P.)]; and the Biotechnology and Biological Sciences Research Council Doctoral Training Award reference BB/F017235/1 (to L.S.P.)].

References

References
1
Paytubi
S.
Wang
X.
Lam
Y.W.
Izquierdo
L.
Hunter
M.J.
Jan
E.
Hundal
H.S.
Proud
C.G.
ABC50 promotes translation initiation in mammalian cells
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
24061
-
24073
)
[PubMed]
2
Tyzack
J.K.
Belsham
G.J.
Wang
X.
Proud
C.G.
ABC50 interacts with eukaryotic initiation factor (eIF) 2 and binds to ribosomes in an ATP-dependent manner
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
34131
-
34139
)
[PubMed]
3
Cavener
D.R.
Ray
S.C.
Eukaryotic start and stop translation sites
Nucleic Acids Res.
1991
, vol. 
19
 (pg. 
3185
-
3192
)
[PubMed]
4
Kozak
M.
How do eucaryotic ribosomes select initiation regions in messenger RNA?
Cell
1978
, vol. 
15
 (pg. 
1109
-
1123
)
[PubMed]
5
Hinnebusch
A.G.
Molecular mechanism of scanning and start codon selection in eukaryotes
Microbiol. Mol. Biol. Rev.
2011
, vol. 
75
 (pg. 
434
-
67
)
[PubMed]
6
Passmore
L.A.
Schmeing
T.M.
Maag
D.
Applefield
D.J.
Acker
M.G.
Algire
M.A.
Lorsch
J.R.
Ramakrishnan
V.
The eukaryotic translation initiation factors eIF1 and eIF1A induce an open conformation of the 40S ribosome
Mol. Cell
2007
, vol. 
26
 (pg. 
41
-
50
)
[PubMed]
7
Maag
D.
Fekete
C.A.
Gryczynski
Z.
Lorsch
J.R.
A conformational change in the eukaryotic translation preinitiation complex and release of eIF1 signal recognition of the start codon
Mol. Cell
2005
, vol. 
17
 (pg. 
265
-
275
)
[PubMed]
8
Das
S.
Ghosh
R.
Maitra
U.
Eukaryotic translation initiation factor 5 functions as a GTPase-activating protein
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
6720
-
6726
)
[PubMed]
9
Paulin
F.E.
Campbell
L.E.
O'Brien
K.
Loughlin
J.
Proud
C.G.
Eukaryotic translation initiation factor 5 (eIF5) acts as a classical GTPase-activator protein
Curr. Biol.
2001
, vol. 
11
 (pg. 
55
-
59
)
[PubMed]
10
Pestova
T.V.
Kolupaeva
V.G.
The roles of individual eukaryotic translation initiation factors in ribosomal scanning and initiation codon selection
Genes Dev.
2002
, vol. 
16
 (pg. 
2906
-
2922
)
[PubMed]
11
Asano
K.
Why is start codon selection so precise in eukaryotes?
Translation
2014
, vol. 
2
 pg. 
e28387
 
12
Nanda
J.S.
Saini
A.K.
Munoz
A.M.
Hinnebusch
A.G.
Lorsch
J.R.
Coordinated movements of eukaryotic translation initiation factors eIF1, eIF1A, and eIF5 trigger phosphate release from eIF2 in response to start codon recognition by the ribosomal preinitiation complex
J. Biol. Chem.
2013
, vol. 
288
 (pg. 
5316
-
5329
)
[PubMed]
13
Luna
R.E.
Arthanari
H.
Hiraishi
H.
Nanda
J.
Martin-Marcos
P.
Markus
M.A.
Akabayov
B.
Milbradt
A.G.
Luna
L.E.
Seo
H.C.
, et al. 
The C-terminal domain of eukaryotic initiation factor 5 promotes start codon recognition by its dynamic interplay with eIF1 and eIF2β
Cell Rep.
2012
, vol. 
1
 (pg. 
689
-
702
)
[PubMed]
14
Andersen
C.B.
Becker
T.
Blau
M.
Anand
M.
Halic
M.
Balar
B.
Mielke
T.
Boesen
T.
Pedersen
J.S.
Spahn
C.M.
, et al. 
Structure of eEF3 and the mechanism of transfer RNA release from the E-site
Nature
2006
, vol. 
443
 (pg. 
663
-
668
)
[PubMed]
15
Pisarev
A.V.
Skabkin
M.A.
Pisareva
V.P.
Skabkina
O.V.
Rakotondrafara
A.M.
Hentze
M.W.
Hellen
C.U.
Pestova
T.V.
The role of ABCE1 in eukaryotic posttermination ribosomal recycling
Mol. Cell
2010
, vol. 
37
 (pg. 
196
-
210
)
[PubMed]
16
Garcia-Barrio
M.
Dong
J.
Ufano
S.
Hinnebusch
A.G.
Association of GCN1-GCN20 regulatory complex with the N-terminus of eIF2a kinase GCN2 is required for GCN2 activation
EMBO J.
2000
, vol. 
19
 (pg. 
1887
-
1899
)
[PubMed]
17
Paytubi
S.
Morrice
N.A.
Boudeau
J.
Proud
C.G.
The N-terminal region of ABC50 interacts with eukaryotic initiation factor eIF2 and is a target for regulatory phosphorylation by CK2
Biochem. J.
2008
, vol. 
409
 (pg. 
223
-
231
)
[PubMed]
18
Coldwell
M.J.
Hashemzadeh-Bonehi
L.
Hinton
T.M.
Morley
S.J.
Pain
V.M.
Expression of fragments of translation initiation factor eIF4GI reveals a nuclear localisation signal within the N-terminal apoptotic cleavage fragment N-FAG
J. Cell Sci.
2004
, vol. 
117
 (pg. 
2545
-
2555
)
[PubMed]
19
Ivanov
I.P.
Loughran
G.
Sachs
M.S.
Atkins
J.F.
Initiation context modulates autoregulation of eukaryotic translation initiation factor 1 (eIF1)
Proc. Natl. Acad. Sci. U.S.A.
2010
, vol. 
107
 (pg. 
18056
-
18060
)
[PubMed]
20
Thompson
J.F.
Hayes
L.S.
Lloyd
D.B.
Modulation of firefly luciferase stability and impact on studies of gene regulation
Gene
1991
, vol. 
103
 (pg. 
171
-
177
)
[PubMed]
21
Sung
D.
Kang
H.
The N-terminal amino acid sequences of the firefly luciferase are important for the stability of the enzyme
Photochem. Photobiol.
1998
, vol. 
68
 (pg. 
749
-
753
)
[PubMed]
22
Kozak
M.
Context effects and inefficient initiation at non-AUG codons in eucaryotic cell-free translation systems
Mol. Cell. Biol.
1989
, vol. 
9
 (pg. 
5073
-
5080
)
[PubMed]
23
Gonda
D.K.
Bachmair
A.
Wunning
I.
Tobias
J.W.
Lane
W.S.
Varshavsky
A.
Universality and structure of the N-end rule
J. Biol. Chem.
1989
, vol. 
264
 (pg. 
16700
-
16712
)
[PubMed]
24
Kozak
M.
At least six nucleotides preceding the AUG initiator codon enhance translation in mammalian cells
J. Mol. Biol.
1987
, vol. 
196
 (pg. 
947
-
950
)
[PubMed]
25
Wegrzyn
J.L.
Drudge
T.M.
Valafar
F.
Hook
V.
Bioinformatic analyses of mammalian 5'-UTR sequence properties of mRNAs predicts alternative translation initiation sites
BMC Bioinformatics
2008
, vol. 
9
 pg. 
232
 
[PubMed]
26
Peabody
D.S.
Translation initiation at non-AUG triplets in mammalian cells
J. Biol. Chem.
1989
, vol. 
264
 (pg. 
5031
-
5035
)
[PubMed]
27
Ingolia
N.T.
Lareau
L.F.
Weissman
J.S.
Ribosome profiling of mouse embryonic stem cells reveals the complexity and dynamics of mammalian proteomes
Cell
2011
, vol. 
147
 (pg. 
789
-
802
)
[PubMed]
28
Moody
J.E.
Millen
L.
Binns
D.
Hunt
J.F.
Thomas
P.J.
Cooperative, ATP-dependent association of the nucleotide binding cassettes during the catalytic cycle of ATP-binding cassette transporters
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
21111
-
21114
)
[PubMed]
29
van der Does
C.
Presenti
C.
Schulze
K.
Dinkelaker
S.
Tampe
R.
Kinetics of the ATP hydrolysis cycle of the nucleotide-binding domain of Mdl1 studied by a novel site-specific labeling technique
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
5694
-
5701
)
[PubMed]
30
Tombline
G.
Bartholomew
L.A.
Urbatsch
I.L.
Senior
A.E.
Combined mutation of catalytic glutamate residues in the two nucleotide binding domains of P-glycoprotein generates a conformation that binds ATP and ADP tightly
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
31212
-
31220
)
[PubMed]
31
Kozak
M.
Point mutations define a sequence flanking the AUG initiator codon that modulates translation by eukaryotic ribosomes
Cell
1986
, vol. 
44
 (pg. 
283
-
292
)
[PubMed]
32
Lee
S.
Liu
B.
Lee
S.
Huang
S.X.
Shen
B.
Qian
S.B.
Global mapping of translation initiation sites in mammalian cells at single-nucleotide resolution
Proc. Natl. Acad. Sci. U.S.A.
2012
, vol. 
109
 (pg. 
E2424
-
E2432
)
[PubMed]
33
Ivanov
I.P.
Firth
A.E.
Michel
A.M.
Atkins
J.F.
Baranov
P.V.
Identification of evolutionarily conserved non-AUG-initiated N-terminal extensions in human coding sequences
Nucleic Acids Res.
2011
, vol. 
39
 (pg. 
4220
-
4234
)
[PubMed]
34
Yamanaka
S.
Poksay
K.S.
Arnold
K.S.
Innerarity
T.L.
A novel translational repressor mRNA is edited extensively in livers containing tumors caused by the transgene expression of the apoB mRNA-editing enzyme
Genes Dev.
1997
, vol. 
11
 (pg. 
321
-
333
)
[PubMed]
35
Levy-Strumpf
N.
Deiss
L.P.
Berissi
H.
Kimchi
A.
DAP-5, a novel homolog of eukaryotic translation initiation factor 4G isolated as a putative modulator of gamma interferon-induced programmed cell death
Mol. Cell Biol.
1997
, vol. 
17
 (pg. 
1615
-
1625
)
[PubMed]
36
Imataka
H.
Olsen
H.S.
Sonenberg
N.
A new translational regulator with homology to eukaryotic translation initiation factor 4G
EMBO J.
1997
, vol. 
16
 (pg. 
817
-
825
)
[PubMed]
37
Coldwell
M.J.
Morley
S.J.
Specific isoforms of translation initiation factor 4GI show differences in translational activity
Mol. Cell. Biol.
2006
, vol. 
26
 (pg. 
8448
-
8460
)
[PubMed]
38
Byrd
M.P.
Zamora
M.
Lloyd
R.E.
Generation of multiple isoforms of eukaryotic translation initiation factor 4GI by use of alternate translation initiation codons
Mol. Cell. Biol.
2002
, vol. 
22
 (pg. 
4499
-
4511
)
[PubMed]
39
Weingarten-Gabbay
S.
Khan
D.
Liberman
N.
Yoffe
Y.
Bialik
S.
Das
S.
Oren
M.
Kimchi
A.
The translation initiation factor DAP5 promotes IRES-driven translation of p53 mRNA
Oncogene
2013
, vol. 
33
 (pg. 
611
-
618
)
[PubMed]
40
Marash
L.
Liberman
N.
Henis-Korenblit
S.
Sivan
G.
Reem
E.
Elroy-Stein
O.
Kimchi
A.
DAP5 promotes cap-independent translation of Bcl-2 and CDK1 to facilitate cell survival during mitosis
Mol. Cell
2008
, vol. 
30
 (pg. 
447
-
459
)
[PubMed]
41
Lewis
S.M.
Cerquozzi
S.
Graber
T.E.
Ungureanu
N.H.
Andrews
M.
Holcik
M.
The eIF4G homolog DAP5/p97 supports the translation of select mRNAs during endoplasmic reticulum stress
Nucleic Acids Res.
2008
, vol. 
36
 (pg. 
168
-
178
)
[PubMed]
42
Higgins
C.F.
Linton
K.J.
The ATP switch model for ABC transporters
Nat. Struct. Mol. Biol.
2004
, vol. 
11
 (pg. 
918
-
926
)
[PubMed]
43
Locher
K.P.
Structure and mechanism of ABC transporters
Curr. Opin. Struct. Biol.
2004
, vol. 
14
 (pg. 
426
-
431
)
[PubMed]
44
Sokabe
M.
Fraser
C.S.
Hershey
J.W.
The human translation initiation multi-factor complex promotes methionyl-tRNAi binding to the 40S ribosomal subunit
Nucleic Acids Res.
2012
, vol. 
40
 (pg. 
905
-
913
)
[PubMed]
45
Asano
K.
Clayton
J.
Shalev
A.
Hinnebusch
A.G.
A multifactor complex of eukaryotic initiation factors, eIF1, eIF2, eIF3, eIF5 and initiator tRNAMet is an important translation initiation intermediate in vivo
Genes Dev.
2000
, vol. 
14
 (pg. 
2534
-
2546
)
[PubMed]
46
Marton
M.J.
Vazquez de Aldana
C.R.
Qiu
H.
Chakraburtty
K.
Hinnebusch
A.G.
Evidence that GCN1 and GCN20, translational regulators of GCN4, function on elongating ribosomes in activation of eIF2alpha kinase GCN2
Mol. Cell. Biol.
1997
, vol. 
17
 (pg. 
4474
-
4489
)
[PubMed]

Supplementary data