A short translational ramp determines the efficiency of protein synthesis

Abstract

Translation initiation is a major rate-limiting step for protein synthesis. However, recent studies strongly suggest that the efficiency of protein synthesis is additionally regulated by multiple factors that impact the elongation phase. To assess the influence of early elongation on protein synthesis, we employed a library of more than 250,000 reporters combined with in vitro and in vivo protein expression assays. Here we report that the identity of the amino acids encoded by codons 3 to 5 impact protein yield. This effect is independent of tRNA abundance, translation initiation efficiency, or overall mRNA structure. Single-molecule measurements of translation kinetics revealed pausing of the ribosome and aborted protein synthesis on codons 4 and 5 of distinct amino acid and nucleotide compositions. Finally, introduction of preferred sequence motifs only at specific codon positions improves protein synthesis efficiency for recombinant proteins. Collectively, our data underscore the critical role of early elongation events in translational control of gene expression.

Fig. 1. Fluorescence-based screen identifies large differences in protein synthesis as a result of the identity of the N-terminus.

a Scheme of the reporter system to test the influence of the first five amino acids and mRNA sequence of the first ribosome footprint. Nine random nucleotides were introduced into eGFP reporter in the positions from 7 to 15 nucleotide coding for amino acids 3, 4, and 5 in protein. b Fluorescence-activated cell sorting (FACS) of induced E.coli cells into 5 bins. Bin 1–4 each represent approximately 24% of the whole cell population depending on eGFP expression. Bin 5 represents 2.5% of the E.coli cells with highest eGFP expression based on relative fluorescence values (RFUs). E.coli cells were sorted based on granularity (SSC-A) and eGFP fluorescence (FITC-A) channels. c Table of relative average fluorescence values for colonies in five separated bins. Wild type eGFP expression is approximately 250 RFUs. d Distribution of the plasmid reads based on the GFP score. GFP score represents distribution value for each independent sequence in 5 bins.

Fig. 2. Effect of A-U content and RNA structure on protein expression.

a Reporters with increased A-U content have slightly higher GFP score. Reporters are binned by the number of A or U nucleotides and plotted against GFP score b Influence of local-mRNA structure on expression of eGFP 9nt library. GFP score distribution value is plotted in correlation with the number of A or U nucleotides in 9nt randomized sequence. Boxplot whiskers indicate the furthest datum that is 1.5*Q1 (upper) or 1.5*Q3 (lower).

Fig. 3. Identification of motifs that correlate with GFP score.

a, b Enrichment analyses of sequenced constructs with average GFP score of ≥4.0 results in two motifs with DNA sequence AADUAU and AAVAUU, or amino acid sequence K|N-Y and K|N-I, respectively. Average GFP score of all sequences with two motifs (present) is compared to the rest of library (absent). c Analysis scheme of the GFP scores for two motifs by moving one nucleotide at the time. Position 1 and position 4 code for K|N-Y and K|N-I amino acid motifs as codons 3 and 4 or 4 and 5, respectively. d, e Analysis of average GFP scores for two sequences motifs based on their position in 9nt randomized sequence indicates potential amino acid dependence. Average GFP score is compared to the rest of library (absent). f Scheme of analysis of overall influence of amino acid sequence when motifs code for amino acids in positions 3 and 4 or 4 and 5, respectively. g Analysis of overall influence of amino acid sequence of motif K|N-I|Y in positions 3, 4, and 5. Average GFP score for motifs is compared to the rest of library (absent). h, i Analysis of the influence of degenerate codons for Tyr(Y) or Asn(N) and Lys(K) on the GFP score of AADUAU motif, respectively. All analyzed sequences with stop codons were filtered out to represent average coding library (absent). Comparison is shown vs all the coding constructs in the library. Boxplot whiskers indicate the furthest datum that is 1.5*Q1 (upper) or 1.5*Q3 (lower).

Fig. 4. Analysis of the effect of motifs in in vitro and in vivo bacterial expressions assays.

a Western-blot analysis of NEB Pure Express in vitro expression of eGFP constructs with motif 1 AADUAU in different positions coding for amino acids 3–5. Mp1 indicates motif1 in position 1, Mp2 indicates motif1 in position 2, Mp3 indicates motif1 in position 3 and Mp4 indicates motif1 in position 4, where insertion positions are defined as in Fig. 3c. Wild-type eGFP (WT) control is indicated. Five percent of is analyzed. Kinetics of in vitro translation reaction is shown in Supplementary Fig. 13. b Relative eGFP fluorescence from in vitro and in vivo expression of eGFP constructs with AADUAU motif in different position compared to the wild-type eGFP construct. The fluorescence ratio of each construct is plotted as a fold increase over the corresponding WT. Endpoint fluorescence for in vivo induction in E. coli cells (Supplementary Fig. 12) and in vitro reaction (in Supplementary Fig. 13) were used to calculate ratios. Error bars represent standard deviation. c Western-blot analysis indicates that the N-terminal rule does not influence the expression of eGFP variants from pBAD single copy vector in vivo in E. coli Top10 cells. Two high expression variants H1 (NCT) and H2 (LQI) and WT eGFP constructs are indicated. Letter in superscript indicates amino acid in the second position (A-alanine, V-valine, E-glutamic acid). d In vivo analysis of near cognate start codons GUG and UUG eGFP variants. eGFP antibody (JL-8, Clontech), E. coli peptide release factor I (αEcRF1), anti-mouse and anti-rabbit HRP-conjugated secondary antibodies were used to visualize the expression of eGFP and normalization of western blot data, respectively. BioRad Precision Plus marker is indicated in images.

Fig. 5. The position and context of the motifs around the initiation codon is critical for protein-synthesis yield.

a Positional bias in controlling the expression of eGFP constructs with different amino acids in positions 3, 4, and 5. Western-blot analysis of NEB-Pure-Express in vitro expression of eGFP constructs with sequence XYZ (KFS, IGK, and TVG, respectively) as amino acids 3(X), 4(Y), and 5(Z) followed by 6×His tag (MV-XYZ-6xHis) or as amino acids 9(X), 10(Y), and 11(Z) preceded by 6×His tag (MV-6×His-XYZ). Five percent of in vitro translation reaction is analyzed. b Images of E. coli colonies expressing mCherry-eGFP polycistronic constructs from pBAD double vector (shown in the schematic). Four colonies of each construct expressing WT mCherry and eGFP constructs with sequence KFS, IGK, and TVG as amino acids inserted in positions 3, 4, and 5 are shown. Dylight 650 (mCherry), Alexa488 (eGFP) and epi white (optical) filters were used to image expression of mCherry and eGFP proteins in E. coli colonies. c, d Simple insertion of different amino acids in recombinant mEOS2 or human Gαi protein constructs can modulate their expression in vivo in E.coli BL21 cells, respectively. Wild type (WT) and control samples as well as amino acids in position 3, 4, and 5 for variants of mEOS2 and Gαi proteins are indicated. mEOS2 and Gαi constructs were cloned in pET16b and pBAD vector as C-terminally His-tagged proteins. Proteins were visualized based on their C-terminal 6-Hist tag using Penta-His (Qiagen) antibody. GFP antibody (JL-8, Clontech) is used to visualize the expression of eGFP and BioRad Precision Plus marker is indicated in all images. The same amount of the E.coli cells (OD600) was used for Western Blot analysis of in vivo expression of different reporter constructs.

Fig. 6. The identity of the first 5 amino acids impacts protein synthesis in a well-defined in vitro translation system.

a Thin-layer chromatography (TLC) analysis of in vitro peptide synthesis using S35-labeled methionine (M in red). Sequences and GFP scores of tetra-peptides, penta-peptides, and hexa-peptides representing starts of wild type eGFP (MVSKGK) and two high expressing clones MVKYH and MVYKH are indicated. Protein synthesis was initiated from pelleted initiation complexes at time 0 and resolved over time (300 s). Points at 1, 5, 10, 20, 30, 45, 60, 90, 120, 180, 240, 300 s are shown. Migration of tetra-peptide, penta-peptide, and hexa-peptide is indicated. Arrows indicate final hexa-peptide products of the reaction. b Analyses of accumulation of the tetra-peptide and hexa-peptide for MVSKGK, MVKYHK, and MVYKHK peptides. Amounts of radioactivity for tetra-peptide and hexa-peptide for the last three time points (180–300 s) were normalized to total radioactivity and plotted in relation to MVSK or MVSKGK peptide amounts. Error bars represent standard deviation.

Fig. 7. smFRET assay for monitoring translation of codon 3–5.

a Schematics of zero-mode waveguide (ZMW)-based single-molecule FRET assay to monitor translation. Elongation factors including fluorescently-labeled tRNA (Phe-(Cy5)-tRNA^Phe) and quencher-labeled large ribosomal subunit (BHQ-50S) are delivered to pre-initiation complex (PIC) with labeled small ribosomal subunit (Cy3B-30S) and mRNA tethered to the ZMWs. b Expected fluorescence signal observed from a translating complex utilizing FRET between Cy3B and BHQ-2 on ribosomal subunits, as well as direct excitation of Cy5 on Phe-(Cy5)-tRNA^Phe. c Measured percentage of complete translation events for different codon 3–5 mRNA constructs (n = 179 molecules for all; error bars represent s.e. based on the binomial distribution). d Representative traces for “processive” translation of K₃I₄H₅ (Left) and “abortive” translation of T₃V₄G₅ (Right).

Fig. 8. Parsing tRNA and amino acid contributions to abortive translation of codon 3–5.

a Expected fluorescence traces for successful and abortive translation. A possible sampling of tRNA to the A site may occur on the arrested ribosomal population, resulting in multiple binding event of Cy5-labeled specific tRNA such as Lys-(Cy5)-tRNA^Lys. b A pie-chart of different populations observed in the experiment. c Representative traces for each population. d Measured percentage of complete translation events for K₃Y₄Y₅ and V₃A₄A₅ mRNA codons with a correct peptide, and for translating K₃Y₄Y₅ mRNA codons with V₃A₄A₅, K₃A₄A₅ or A₃A₄A₅ peptide sequences using Flexizyme-mischarged tRNAs (n = 161, 147, 156, 131, and 142 molecules from left to right; error bars represent s.e. based on the binomial distribution).

Fig. 9. Model for translational regulation by the identity of N-terminal sequence.

a Schematic representing the initiating and elongating ribosome footprints as well as a movement during the translation of the first five codons described in this manuscript. b “Stall and drop-off” model in translational regulation by the identity of N-terminal sequence. Arrows indicate stalled and abortive translation on non-processive peptides followed by ribosome recycling and peptidyl-tRNA drop-off at codons 4 and 5. Start codon (AUG), ribosome binding site (RBS), ribosomal subunits, peptide exit channel, ribosome footprint, peptidyl-tRNAs, amino acids as well as position of detrimental 3–5 codons is indicated.