The impact of mRNA poly(A) tail length on eukaryotic translation stages

Abstract

The poly(A) tail plays an important role in maintaining mRNA stability and influences translation efficiency via binding with PABP. However, the impact of poly(A) tail length on mRNA translation remains incompletely understood. This study explores the effects of poly(A) tail length on human translation. We determined the translation rates in cell lysates using mRNAs with different poly(A) tails. Cap-dependent translation was stimulated by the poly(A) tail, however, it was largely independent of poly(A) tail length, with an exception observed in the case of the 75 nt poly(A) tail. Conversely, cap-independent translation displayed a positive correlation with poly(A) tail length. Examination of translation stages uncovered the dependence of initiation and termination on the presence of the poly(A) tail, but the efficiency of initiation remained unaffected by poly(A) tail extension. Further study unveiled that increased binding of eRFs to the ribosome with the poly(A) tail extension induced more efficient hydrolysis of peptidyl-tRNA. Building upon these findings, we propose a crucial role for the 75 nt poly(A) tail in orchestrating the formation of a double closed-loop mRNA structure within human cells which couples the initiation and termination phases of translation.

Graphical Abstract

Graphical Abstract.

Introduction

Polyadenylation at the 3′ end is crucial for mRNA stability and regulation of protein biosynthesis. This modification plays a pivotal role in safeguarding mRNA stability, preventing its degradation. Beyond its role in stability, the poly(A) tail actively engages in the intricate orchestration of protein biosynthesis regulation through interactions with the poly(A)-binding protein (PABP), thereby exerting a profound influence on gene expression modulation (1–3). The process of primary polyadenylation occurs within the nucleus subsequent to mRNA transcription. In the immediate after polyadenylation, the average length of mammalian poly(A) tails ranges from 150 to 250 nucleotides (nt) (2,3). However, the dynamic nature of poly(A) tails becomes evident in the cytoplasm, where they undergo dynamic alterations characterized by shortening and secondary elongation processes. In mammals, the steady-state length of mRNA poly(A) tails stabilizes at approximately 80–100 nt, reflecting a finely tuned equilibrium between degradation and elongation events (2–5). Intriguingly, the length of poly(A) tails exhibits considerable variability among different mRNA types within the cell. Furthermore, even among transcripts originating from the same gene, variations in poly(A) tail length are observed (2,4,6,7).

The influence of poly(A) tail length on translation has been observed in various systems, including Xenopus laevis oocytes (8), Drosophila embryonic cells (9), mouse Krebs cells (10), rabbit reticulocyte lysate (RRL) (11,12) and human HeLa cells (13). Translation of mRNA, lacking a poly(A) tail, is markedly suppressed. Similarly, mRNA featuring a short poly(A) tail (ranging from 15 to 30 nucleotides) exhibits notably low translation efficiency. Intriguingly, as the poly(A) tail length is extended beyond this initial range, there is a discernible improvement in translation efficiency. However, it is noteworthy that this enhancement eventually reaches a plateau, indicating a saturable effect where further lengthening of the poly(A) tail ceases to confer additional translational benefits. These experiments determined the amount of synthesized product, which is affected by both translation efficiency and mRNA stability. In nuclease-untreated RRL it was demonstrated that the effect of the poly(A) tail length (ranging from 5 to 50 nucleotides) on translation of capped mRNA was weak, but translation of uncapped mRNA was dependent on the poly(A) tail length (11). Another study of translation in RRL reveals a distinct association between poly(A) tail length and the formation of polysomes. Significantly, longer poly(A) tails on mRNA facilitate a more efficient assembly of polysomes in RRL (14).

The poly(A) tail of the mRNA binds to PABP, which in turn is involved in translation and is required to maintain mRNA stability (for comprehensive reviews, refer to (1,15–19)). This protein has both cytoplasmic and nuclear forms. Cytoplasmic poly(A) binding protein (PABPC) interacts with a variety of other proteins, forming complexes with them and modulating their activity (16,18). Almost all eukaryotes have PABPCs, but the number of their isoforms varies. The most widely represented isoform in human cells is isoform 1 (PABPC1), which we studied in this work (hereinafter we will call it PABP) (16–18). PABP’s structural architecture comprises four RNA-binding RRM domains located at the N terminus. N terminus also interacts with proteins harboring the PAM1 motif, while the C terminus binds proteins containing the PAM2 motif, affording PABP a multifaceted interaction repertoire (16,20,21). PABP has an extremely high affinity for poly(A) sequences, with a minimum binding site spanning approximately12 nt. Bound to poly(A) molecule of PABP covers approximately 27 nt (16,19,22–26). Binding of PABP to poly(A) sequences is cooperative (27). Moreover, evidence suggests a nonrandom binding pattern along the poly(A) tail. The initial PABP molecule is proposed to anchor at the junction of the 3′ untranslated region (3′UTR) and the poly(A) tail, with subsequent PABP molecules binding cooperatively adjacent to the first PABP (1,28). Beyond its interactions with poly(A) sequences, PABP engages in a vast network of molecular contacts. This includes binding to diverse RNA sequences (29,30), as well as to ribosomes (31), a number of translation factors (eIF4G, eIF4B, PAIP1, PAIP2 and eRF3), RNA degradation factors (LARP1, LARP4 and Ataxin-2), deadenylation complex proteins (PAN2, PAN3, CCR4), and miRNA-induced silencing proteins (TNRC6) (refer to reviews (1,15–17)). Through this intricate network of interactions, PABP emerges as a pivotal modulator, stimulating both translation initiation (14,32,33) and termination (34–37), while also influencing deadenylation processes (1,38) and nonsense-mediated mRNA degradation (NMD) (39,40).

PABP stands as a highly abundant protein in cells (41,42). Quantitative assessments across various human cell lines reveal a substantial prevalence of PABP, surpassing its main partners in translation, namely, eukaryotic translation initiation factor 4G (eIF4G) and eukaryotic release factor 3 (eRF3), by factors ranging from 3000 to 10 300 and 1.1 to 15, respectively (41,43,44). Remarkably, it has been proposed that a significant fraction, ranging from 30 to 75%, of cellular PABP may exist independently of poly(A) mRNA sequences and/or polysomes (42,45,46). In solution, PABP exhibits a propensity to form dimers, a structural arrangement that hinders its binding with eIF4G. This interaction prevents the displacement of eIF4G from initiation, thereby preventing translational suppression (47). Interestingly, in the studies determining the efficiency of cell translation, it was found that the length of the poly(A) tail correlated with translation efficiency only under PABP-deficiency conditions. Under PABP-saturated conditions, the correlation between poly(A) tail length and translation efficiency disappeared (1,5,7,48).

Upon interaction with the poly(A) tail, PABP binds with eIF4G, a component of the cap-binding complex, inducing the formation of a closed-loop structure. This conformational arrangement brings the 5′ and 3′ ends of mRNA into close proximity, significantly enhancing the initiation of translation (15,49–51). In addition to its role in translation initiation, PABP, interacting with eRF3, contributes to the loading of release factors onto the stop codon in the ribosome, thereby stimulating translation termination (34–37). Translation termination, the process by which the ribosome terminates protein synthesis, is triggered by the appearance of a stop codon at the A site of the ribosome (52,53). In eukaryotes, all three stop codons are recognized by the release factor 1 (eRF1) (54–58), and the partner factor eRF3 stimulates its termination activity (59–63). Despite the demonstrated involvement of PABP in various stages of translation, an exploration of the impact of poly(A) tail length on these stages has not been studied.

Since the influence of poly(A) tail length on translation efficiency has previously been demonstrated, and there was no clear data on the influence of PABP on different stages of translation, we conducted a comprehensive study of this issue in various in vitro systems. Initially, the effect of poly(A) tail length on translation was studied using model capped and uncapped mRNAs in cell-free translation systems. Subsequently, a reconstituted mammalian translation system was employed. Initiation, elongation, and termination complexes were assembled on mRNAs featuring varied poly(A) tail lengths. Additionally, pre-termination complexes were obtained using Rabbit Reticulocyte Lysate (RRL), providing a platform to assess the influence of poly(A) tail length on the hydrolysis of peptidyl-tRNA during translation termination. As a result, we have identified stages of translation that depend on the length of the poly(A) tail and proposed a model of translation regulation by this mRNA element.

Materials and methods

Model mRNAs

Model mRNAs were synthesized via run-off transcription using the T7 RiboMAX™ Large Scale RNA Production System (Promega) kit following the manufacturer's protocol. All used model mRNAs contained UAA stop codon. To obtain capped mRNA, 3 mM 3′-O-Me-m⁷G(5′)ppp(5′)G RNA Cap Structure Analog (ARCA, NEB) was added to the reaction mixture in a 3.33-fold excess relatively to GTP. As an alternative, Fluc mRNA was capped by Vaccinia Capping System (M2080S, NEB) following the manufacturer's protocol. After transcription, mRNA was isolated by acidic phenol, precipitated with 3 M LiCl, washed with 80% ethanol and purified by gel filtration using NAP-5 column (Cytiva).

Templates for run-off transcription were obtained through PCR amplification of 5′ and 3′UTRs, along with coding sequences of firefly luciferase (Fluc) from pGEM-Fluc-GN, nanoluciferase (Nluc) from pNL-globin (64),or the short peptide MVHL from pET28-MVHL-UAA (59) with forward primers, containing T7 promoter, and reverse primers, additionally containing 0, 10, 25, 50, 75 or 100 poly(T) (Supplementary Table S1).

pGEM-Fluc-GN was constructed from the original plasmid pGEM-Fluc (Promega, Cat. No. E1541), encoding Firefly luciferase (Fluc), with a short 3′UTR. The part of 3′UTR was substituted with a segment of the 3′UTR of GAPDH gene (used in (65)) in a location of 52 nt downstream of the stop codon. The vector was amplified by PCR with Q5 polymerase (NEB) using FlucGib_F and FlucGib_R primers (Supplementary Table S1), and the insert was obtained with FlucGN_F, FlucGN_R primers (Supplementary Table S1). The insert was integrated into the linearized vector using a Gibson Assembly® Master Mix (NEB).

To introduce a premature termination codon (PTC) in Fluc, the 447th TAT codon encoding tyrosine was replaced with one of three stop codons (TAA, TAG or TGA) using the QuikChange Site-Directed Mutagenesis Kit (Agilent) with a pair of specific primers (respectively, Fluc_UAA_F and Fluc_UAA_R, Fluc_UAG_F and Fluc_UAG_R, Fluc_UGA_F and Fluc_UGA_R) (Supplementary Table S1). To obtain template for run-off transcription, we amplified fragments of the pGEM-Fluc-GN plasmid (wild type or with PTC) using Fluc_gl_ART1_koz_F(forward) and Fluc_R primers (Supplementary Table S1). At the 5′ end, the forward primer contained the T7 promoter and a translationally optimized 5′UTR sequence (part of the 5′UTR of the human β-globin gene followed by the best Kozak sequence according to (66)).

Cell free translation

HEK293F lysate was prepared as described in (67). 0.26 pmol of capped or uncapped Fluc mRNA (0.52 pmol for Vaccinia capped mRNA) wild-type (wt) or premature termination codon (PTC)-containing was incubated in 10 μl mixture containing 50% (v/v) HEK293F lysate, 20 mM HEPES–KOH pH 7.5, 2 mM DTT, 0.25 mM spermidine, 0.6 mM Mg (OAc)₂, 16 mM creatine phosphate, 0.06 U/μl creatine kinase (SigmaAldrich), 1 mM ATP, 0.6 mM GTP, 60 mM KOAc, 0.05 mM of each proteinogenic amino acid (Promega), 0.2 U/μl RiboLock (Thermo Scientific), 5 mM d-luciferin. Optionally, 9 pmol of recombinant PABP was included in some reactions. Translation assays in wheat germ extract (WGE) (Promega) was conducted in 10 μl of 50% extract, following to manufacturer's protocol.

Luminescence of Fluc was measured at 30°C using a Tecan Infinite 200 Pro (Tecan, Männedorf, Switzerland) during 120 min. The translation efficiency was calculated as a maximal derivative of the growing linear section of the luminescence curve (v₀, RLU/min).

Western blot analysis of lysates and ribosomal complexes

HEK293 and HeLa lysates were analysed by standard immunostaining with the appropriate antibodies raised against eRF1, eRF3a, eRF3c and PABP (Supplementary materials). Human cell lines PC-3, MG-63, SAOS2, HOS, HEK293, MCF-7 were cultured on 25 cm² Tissue Culture Flasks until reaching 70–90% confluence. Subsequently, cells were washed on ice twice with ice-cold PBS and lysed in 100 μl of O’Farrell lysis buffer (68). The lysis buffer was supplemented with 1 μM PMSF and Protease Inhibitor Cocktail (Promega, G652A). The presence of eRF1, eRF3a, eRF3c and PABP was determined through immunostaining analysis of 6 μl of each lysate with the appropriate antibodies (Supplementary materials).

To obtain detectable amounts of proteins in pretermination complexes (preTCs), 1 femtomole of preTC-Nluc was precipitated with 10% (v/v) trichloroacetic acid in the presence of 2% (v/v) of casamino acids at 4 ºC overnight. Following centrifugation at 16 100 g for 1 h at 4ºC, the precipitate was diluted in 15 μl of standard sample buffer with an additional 0.5 M Tris–HCl pH 7.5. The presence of translation factors and ribosomes was determined by immunostaining with the appropriate antibodies (Supplementary materials).

Purification of ribosomal subunits and proteins

40S and 60S ribosomal subunits, eEF1 and eEF2 were purified from a RRL cell lysate, while eIF2 and eIF3 were purified from a HeLa cell lysate, as described previously (59). Recombinant eRF3a and eIF4G were produced using the baculovirus MultiBac expression system (34). Other human translation factors (eIF1, eIF1A, eIF4A, eIF4B, ΔeIF4G, eIF4E, ΔeIF5B, eIF5, eRF1(AGQ), eRF3c) were produced as recombinant proteins in Escherichia coli strains BL21(DE3) or Rosetta(DE3) (59). Recombinant proteins were purified using Ni-NTA agarose and ion-exchange chromatography, as described previously (34,59).

Human PABP, eRF1 and recombinant nanoluciferase (Nluc) were expressed as fusion proteins with His-SUMO tag. To construct pET-SUMO-PABP plasmid, PABPC1 coding sequence was amplified from the pET3b-PABP plasmid (34) using primers petSUMO_PABP_F and petSUMO_PABP_R (Supplementary Table S1). The petSUMO plasmid (Invitrogen) was linearized using primers petSUMO_F and petSUMO_R (Supplementary Table S1) (64). The pET-SUMO-PABP was then obtained with Gibson Assembly® Master Mix (NEB), following to the manufacturer's protocol. The resulting plasmid encoded an N-terminal His-tagged SUMO fused to PABP (His-SUMO-PABP). To obtain pET-SUMO-eRF1, eRF1 coding sequence was amplified from the plasmid pET23b-eRF1 (69) using primers petSUMO_eRF1_F and petSUMO_eRF1_R (Supplementary Table S1). To obtain pET-SUMO-Nluc, Nluc coding sequence was amplified from the plasmid pNL-globin (64) using primers petSUMO_Nluc_F and petSUMO_Nluc_R (Supplementary Table S1). Recombinant proteins were expressed in E. coli Rosetta(DE3) or BL21(DE3). The pelleted cells were sonicated in the buffer contained 20 mM Tris–HCl pH 7.5, 800 mM KCl, 15% glycerol, 0.1% Triton X-100, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol (DTT). Proteins were purified using Ni-affinity (HisTrap HP, Cytiva) and ion-exchange (HiTrap SP or HiTrap Q Cytiva) chromatography. Finally, The His-SUMO tag was then cleaved from the proteins of interest using His-tagged Ulp1 peptidase in the buffer contained 20 mM Tris–HCl pH 7.5, 250 mM KCl, 15% glycerol, 0.15% Triton X-100 1 mM DTT at 4 ºC overnight, as described in (64). To remove His-SUMO and His-Ulp1 from the samples, Ni-affinity chromatography was performed. Flow-through fractions, contained untagged protein, were collected and dialyzed in the storage buffer (20 mM Tris–HCl pH 7.5, 100 mM KCl, 15% glycerol, 1 mM DTT).

Assembly and analysis of ribosomal complexes on MVHL mRNA

Ribosomal complexes were assembled using the in vitro reconstituted translation system (RETS) with modifications based on previously described protocols (59,70). The assembly involved various stages of translation, including initiation, elongation, and termination. eIF4F assembly was initiated by incubating eIF4G with a 2× excess of eIF4E, eIF4A and eIF4B. The 48S preinitiation ribosomal complex (48S PIC) was assembled at 4°C in 10 μl reaction mixture containing 0.375 pmol MVHL mRNA, 3 pmol Met-tRNA_i^Met, 1 pmol 40S, 1 pmol eIF4F, 7 pmol eIF4A, 5 pmol eIF2, 1 pmol eIF3, 2 pmol eIF1 and eIF1A, and optionally 4.5 pmol PABP in translation buffer composed of 20 mM Tris acetate, pH 7.5, 100 mM KAc, 2.5 mM MgCl₂, 2 mM DTT, 0.4 U/μl Ribolock (Thermofisher), 1 mM ATP, 0.25 mM spermidine and 0.2 mM GTP. The reaction mixture was incubated at 37°C for 15 min. For assembly of 80S initiation complex, 1 pmol 60S, 2 pmol eIF5 and 2 pmol ΔeIF5B were added to 10 μl of 48S PIC and incubated at 37°C for 10 min. Elongation of translation (pre-termination complex formation) was performed for 15 min at 37°C after addition to 80S of 3 pmol aminoacylated Val, His and Leu tRNAs, 30 pmol of aminoacylated total calf tRNA, 3 pmol eEF1, and 0.5 pmol eEF2. Translation termination (post-termination complex formation) was performed at 37°C for 10 min after addition to preTC different concentrations of eRF1 and eRF3a.

For detection of the ribosome position at mRNA, a fluorescent toe-print analysis was performed using AMV reverse transcriptase and FAM-labeled primer FAM-toe (Supplementary Table S1) as described in (70,71). Reverse transcription reaction of 10 μl 48S PIC, preTC or postTC in translation buffer supplemented by 0.5 mM 4NTPs, 8 mM MgCl₂, 0.5 μM FAM-toe primer, and 0.3 U/μl AMV RT (Promega) was incubated for 1 h at 37°C. The samples were diluted with 0.1% SDS, purified by phenol:chloroform (1:1) extraction, precipitated in 70% ethanol, and dissolved in 95% formamide containing SD-450 size standard (Syntol, LLC). Samples were analyzed by the ABI Prism® Genetic Analyser 3100 (Applera) or the Syntol LLC Nanophor 05 Genetic Analyser (Syntol, LLC) in PDMA-6 polymer (Syntol, LLC).

Purification of preTC-Nluc and Termi-Luc assay

Pre-termination ribosomal complexes, containing Nluc, (preTC-Nluc) for the Termi-Luc assay were purified using a modified method described in (64,72). Initially, 100% RRL lysate (Green Hectares) was nuclease treated via incubation with 1 mM CaCl₂ and 3 U/μl Micrococcal nuclease (Fermentas) at 30°C for 10 min, followed by the addition of EGTA to a final concentration of 4 mM. Then, the treated lysate was diluted to 70% (v/v) with a translation buffer. Final concentration of components was 20 mM HEPES–KOH pH 7.5, 80 mM KOAc, 0.5 mM Mg(OAc)₂, 0.3 mM ATP, 0.2 GTP, 0.04 mM each of 20 proteinogenic amino acids (Promega), 0.5 mM spermidine, 0.45 μM aminoacylated total rabbit tRNA, 10 mM creatine phosphate, 0.003 U/μl creatine kinase (Sigma), 2 mM DTT and 0.2 U/μl Ribolock (ThermoFisher) (70% RRL mix).

For preTC-Nluc assembly, 220 μl of the 70% RRL mix were preincubated in the presence of 1.7 μM eRF1(AGQ) at 30°C for 10 min. Subsequently, 10.5 pmol of Nluc mRNA was added, and the mixture was incubated at 30°C for 40 min. Next, KOAc concentration was adjusted to 300 mM, and the mixture was layered onto 5 ml of 10–35% linear sucrose gradient in buffer containing 50 mM HEPES–KOH pH 7.5, 7.5 mM Mg(OAc)₂, 300 mM KOAc, 2 mM DTT. The gradient was centrifuged in a SW55-Ti (Beckman Coulter) rotor at 55 000 rpm (367 598 g_max) for 1 h. The gradient was fractionated by 150 μl from bottom to top for 15 fractions, the rest of the sucrose was removed. The presence of preTC-Nluc in the fractions was determined using the Termi-Luc assay (details provided below). Functionally active preTC-Nluc was identified at the bottom of the gradient. Notably, fractions containing this complex were distinguishable from those associated with the peak detected through the optical absorbance of RNA at 260 nm. preTC-Nluc from the bottom peak were aliquoted, flash-frozen in liquid nitrogen, and stored at −70°C.

The Termi-Luc assay, designed to measure the release of Nluc from the ribosome during translation termination, was conducted as described with certain modifications (72). Peptide release performed in 20 μl reaction mixture containing 1.25 μl of the sample (usually 1.5 pM preTC-Nluc), 45 mM HEPES–KOH pH 7.5, 1.4 mM Mg₂OAc, 56 mM KOAc pH 7.0, 1 mM DTT, 177 μM spermidine, 1.5% (ω/ω) sucrose, 0.8 mM MgCl₂, 0.2 mM GTP supplemented with equimolar MgCl₂, 0.5% NanoGlo (Promega) in the presence of release factors. Luminescence was measured at 30°C using a Tecan Infinite 200 Pro (Tecan, Männedorf, Switzerland) during 40 min. Translation efficiency was calculated as the maximal derivative of the growing linear section of the luminescence curve (v₀, RLU/min), with v₀ in the absence of release factors subtracted. Apparent K_M and v_max values were determined from the v₀, determined for various concentrations of release factors, utilizing a non-linear approximation to the Michaelis-Menten function with the dir.MM function from the renz library for R (73).

The concentration of the preTC-Nluc was evaluated by the maximal luminescence in the presence of puromycin. To determine concentration of the preTC-Nluc, 1.25 μl of the sample were incubated in a 20 μl of mixture, finally contained 45 mM HEPES–KOH pH 7.5, 1.4 mM Mg₂OAc, 56 mM KOAc pH 7.0, 1 mM DTT, 177 μM spermidine, 1.5% (ω/ω) sucrose, 0.8 mM MgCl₂, 0.2 mM GTP supplemented with equimolar MgCl₂, 0.5 g/l BSA (Fermentas) and 1 mM puromycin, at 37°C for 15 min. Subsequently, NanoGlo (Promega) was added to achieve a final concentration of 0.5% (v/v). Luminescence was measured at 30°C using a Tecan Infinite 200 Pro (Tecan, Männedorf, Switzerland) during 40 min. Then RLU_max units were converted to moles using a calibration curve for recombinant Nluc in the same buffer without puromycin.

Statistical analysis

All the main experiments were conducted with a minimum of three replicates. The data are presented as mean ± standard error of the mean (SE). Statistical comparisons between mean values of two groups were carried out using an unpaired two-tailed t-test. For multiple comparisons, the Holm correction of P value was applied (74). Asterisks in the results indicate statistically significant differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001). P values were computed using T.TEST function in Microsoft Excel or t_test function (rstatix library) in R. Adjustments were performed using the p.adjust function from the stats library in R.

Results

The translation rate of capped mRNA does not linearly depend on the length of the poly(A) tail

To investigate the translation efficiency of mRNAs with varying poly(A) tail lengths, we constructed a vector encoding firefly luciferase with an optimal Kozak sequence designed in accordance with (66), along with the previously described 3′UTR sequence (65) (Figure 1A). Translation of the equal amounts of model mRNAs (0.26 pmol) was conducted in a cell-free system using HEK293F cell lysate. The equal mRNA capping efficiency was confirmed by an RNAse H-based fluorescent method (Supplementary Figures S1 and S2). The appropriate lengths of poly(A) tails were confirmed by gel electrophoresis (Supplementary Figure S3).

Figure 1.

Effect of poly(A) tail length on translation in HEK293F cell lysate. (A) Schematic representation of the model mRNAs encoding firefly luciferase (Fluc) used in cell-free translation. (B) Initial translation rate v₀ of capped and uncapped Fluc mRNA in the absence and the presence of exogenous PABP (PABP_exo). RLU, relative luminescent units. The graphs show the mean ± standard error, n = 3. Asterisks indicate statistically significant differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001), insignificant differences are not shown. Red asterisks indicate differences between the values of different curves, blue and orange asterisks indicate differences with the A0 value of the corresponding curve. (C) Western blot analysis of the HEK 293F lysate with antibodies specific to human eRF1, eRF3 and PABP.

On the capped mRNA, an increase in the poly(A) tail length to 10 nt led to a 50% enhancement in the translation rate compared to the mRNA without a poly(A) tail (Figures 1B and S4A). The background signal is shown in Supplementary Figure S5. This translation rate remained constant when the poly(A) length was further extended to 50 nt (Figures 1B and S4A). However, the mRNA with a 75 nt poly(A) tail exhibited significantly faster translation (Figures 1B and S4A). The mRNA with the 100 nt poly(A) tail was translated at the same rate as the mRNA with poly(A) tails ranging from 10 to 50 nt (Figures 1B and S4A). As mRNA stability could affect the translation rate, we estimated it using real-time PCR of two amplicons (internal and 3′UTR) (Supplementary Figure S6A). We have shown that the stability of capped mRNA increased to 25 nt poly(A) and stayed constant up to 100 nt poly(A). It should be noted that we determined only 3′ end degradation, because we normalized the mRNA stability on the internal control amplicon. We suppose that interaction of the poly(A)-bound PABP with the cap-bound eIF4F complex prevents degradation of mRNA from the 3′ end as a result of closed-loop structure formation. The constant stability of capped mRNAs at any poly(A) tail length greater than 25 nt also correlates with the rate of translation on these mRNAs. However, the higher translation rate of 75 nt poly(A) mRNA does not correlate with the degradation rate and is likely determined by other factors.

Thus, we observed a translation maximum in the HEK293F lysate with a 75 nt poly(A) tail on the capped mRNA. Strikingly, the observed optimal poly(A) tail length for translation is close to the most common length of poly(A) tails in human mRNAs, which is around 80 nt (5). Given that other species may have different most common poly(A) tail lengths, we hypothesized that they might exhibit optimal translation at those lengths. To test this hypothesis, we conducted translation in a cell-free system based on wheat germ extract (WGE). We found that in WGE, the translation rate was already maximal with the 50 nt poly(A) tail (Supplementary Figure S7), aligning with the most common poly(A) tail length in plants, which is around 50 nt (5). Thus, the optimal for translation length of the poly(A) tail varies among species.

We also investigated the impact of poly(A) tail length on the translation of uncapped mRNA. In this scenario, PABP bound to the poly(A) tail does not interact with eIF4G, a component of the cap-binding complex. This allowed us to assess the effects of poly(A)-bound PABP on translation without considering stimulation of translation initiation, including mRNA stability and stimulation of the release factor eRF3. Uncapped mRNA was translated approximately 10-fold less efficiently than capped mRNA probably due to reduced initiation efficiency (Figures 1B and S4A). We observed that as the poly(A) tail length of uncapped mRNA increased to 100 nt, the translation rate also increased (Figures 1B and S4A). This finding contrasts with the results obtained for capped mRNA, where we identified the optimal poly(A) tail length as 75 nt, and translation efficiency decreased at 100 nt poly(A). The stability of uncapped mRNA also increased linearly with the poly(A) tail length (Supplementary Figure S6B). Therefore, at the uncapped mRNA the linear increase in translation rate with increasing poly(A) tail length is likely determined by the complex influence of mRNA stability and translation termination. Interestingly, capped mRNAs were much more stable from the 3′ end than uncapped (Supplementary Figure S6C) which confirms our hypothesis of preventing 3′ end mRNA degradation by closed-loop structure.

To ensure complete coating of poly(A) tails in our experiments with PABP, we supplemented the translation reaction with additional PABP. Prior to this, we quantified the levels of endogenous PABP and release factors in the HEK293F lysate through Western blot hybridization with specific antibodies (Figure 1C). Our analysis revealed that the concentration of eRF1 in the lysate was 0.24 ± 0.03 μM, and the concentration of full-length eRF3a (UniProt P15170-2) was 1.15 ± 0.12 μM, which is approximately 5 times higher than the concentration of eRF1 (Table 1). Notably, we did not detect a truncated form of eRF3a, eRF3c (UniProt P15170-1), which is annotated as the canonical isoform of human eRF3. While the eRF3c form was either undetectable or not predominant in HEK293 and HeLa cells, significant amounts of eRF3c were detected in SAOS3, HOS, MCF-7 cells (Supplementary Figure S8). Given that eRF3c lacks a PABP-binding domain, it functions independently of poly(A) tail length, and the activity of eRF3a is directly dependent on its binding to PABP (34). Thus, as we only detected eRF3a in HEK293, translation termination in this cell-free system is likely to be dependent on the concentration of PABP. The determined concentration of PABP in the HEK293F lysate is 1.83 ± 0.12 μM (Table 1), corresponding to approximately 3 × 10⁶ molecules per cell This concentration aligns with the published amount of PABP in HeLa cells, ranging from 2.5 to 8 × 10⁶ molecules per cell (41,42). Therefore, in the HEK293 lysate, the amount of PABP was approximately 2-fold higher than the amount of eRF3a (Table 1).

Table 1.

The concentrations of release factors and PABP in 100% HEK 293F lysate

Translation factor	μМ	Relative amount
eRF1	0.24 ± 0.03	× 1.0
eRF3a	1.15 ± 0.12	× 4.8
PABP	1.83 ± 0.12	× 7.6

Mean ± SE are given, n = 3.

When an equimolar (relative to native) amount of exogenous PABP was added to capped mRNA translation, similar translation stimulation occurred across all tested poly(A) tails (Figures 1B and S4A). This result suggests that the poly(A) tails of actively translated mRNAs were effectively coated with PABP. The addition of exogenous PABP to weakly translated mRNAs, with free poly(A) tails, increased the number of active mRNA molecules and enhanced overall translation efficiency. We hypothesize that complete coverage of the poly(A) tails by PABP is required for the stimulation of cap-dependent translation. In the case of uncapped mRNAs, no stimulation was observed at any poly(A) length when PABP was added. Conversely, a suppression of translation was noted in the absence of a poly(A) tail (Figures 1B and S4A). Apparently, any coating density of the poly(A) tail by PABP seems sufficient to activate translation at uncapped mRNA or to prevent mRNA degradation. The observed translation suppression by PABP on uncapped mRNA lacking a poly(A) tail can be explained by its binding to essential translation proteins in the solution, leading to a decrease in their effective concentrations. In the presence of poly(A), this inhibition disappeared, as the translation efficiency with a poly(A) tail compensated for the inhibitory effect of free PABP.

Hence, we observed distinct dependencies of the translation rate on the length of the poly(A) tail for capped and uncapped mRNA. The former reflects the impact of PABP on cap-dependent translation initiation and the formation of the closed-loop structure, while the latter demonstrates its effect on other processes, including translation termination and mRNA stability. As translation termination occurs not only at natural stop codons but also at premature stop codons (PTCs), we further investigated how the poly(A) tail length could influence PTC readthrough. To assess this, we determined the translation efficiency of capped mRNAs encoding Fluc, containing a PTC in addition to the natural UAA stop codon. Luminescence was observed only if readthrough of the PTC occurred, resulting in the synthesis of full-length luciferase (Supplementary Figure S9). At all three stop codons the poly(A) tail length did not significantly impact translation efficiency. Addition of exogenous PABP suppressed PTC readthrough at UAG and UAA (Supplementary Figure S9, S10). For the UGA PTC, we observed the stimulation of translation by exogenous PABP, but less than the cap-dependent translation of normal Fluc (Supplementary Figures S9, 1B). Consequently, the effect of the poly(A) tail and PABP on PTC readthrough differs at UAG / UAA and UGA codons.

Poly(A) tail regulates cap-independent translation initiation

Given that translation in cell lysate is influenced by numerous factors simultaneously, we aimed to investigate the impact of poly(A) length on individual stages of translation using an in vitro mammalian translation system reconstituted from individual components (mRNA, ribosomal subunits, translation factors, and aminoacylated tRNAs) (59). To examine the effect of poly(A) tail length on translation initiation, we obtained 48S complexes on capped and uncapped mRNAs with different poly(A) tail lengths in the presence and absence of PABP (Supplementary Figure S11). The model mRNAs contained the β-globin 5′UTR and encoded the MVHL tetrapeptide (Figure 2A). Notably, at the 5′ end of these mRNAs, there were four CAA repeats, lacking secondary structure, which facilitated the formation of initiation complexes in the absence of cap. Otherwise, it would have been difficult to observe complexes assembled on uncapped mRNA. Equal mRNA capping efficiency was confirmed by an RNAse H-based fluorescent method (Supplementary Figures S1 and S2). The assembly of the 48S complex was detected using fluorescent toe-print analysis, involving the synthesis of fluorescently labeled cDNA on mRNA associated with the ribosomal complex.

Figure 2.

Effect of poly(A) tail length on 48S preinitiation complex formation. (  A) Schematic representation of the model mRNAs encoding MVHL tetrapeptide used in the reconstitution of ribosomal complexes. (B) Relative amount of the 48S preinitiation complexes formed in the reconstituted mammalian translation system in the presence or absence of PABP on model capped and uncapped mRNAs with different poly(A) tails (0–100). The graphs show the mean ± standard error, n = 4. The amount of the 48S formed on capped mRNA without a poly(A) tail was set as 1. Asterisks indicate statistically significant differences between the values of different curves (*, P < 0.05; **, P < 0.01; ***, P < 0.001), while insignificant differences are not shown.

In the absence of PABP, we observed a slight decrease in the formation of the 48S complex with increasing poly(A) tail length for both capped and uncapped mRNAs (Figure 2B). The addition of PABP eliminated this dependence and stimulated the formation of the 48S complex. We hypothesized that, in the absence of PABP, certain RNA-binding initiation factors might nonspecifically bind to the long poly(A) tail, reducing their local concentration near the 5′ end of the mRNA and, consequently, decreasing the efficiency of translation initiation. Upon the addition of PABP, it specifically bound to the poly(A) tail and displaced nonspecifically bound proteins. This observation aligns with previous studies demonstrating that poly(A) sequences bind to PABP and displace it from random RNA sequences (30,42), free dimers (47) and the ribosomes (31). Similarly, we proposed that PABP could displace nonspecifically bound proteins from the poly(A) tail.

Interestingly, in the case of capped mRNA, PABP stimulated translation initiation regardless of the presence and length of poly(A); at lengths A0 and A100, the efficiencies of 48S assembly were similar (Figure 2B). This implies that 48S complex formation was stimulated both by the free and poly(A)-bound forms of PABP. However, with uncapped mRNA, which contains an unstructured leader sequence, the stimulation of 48S complex formation by PABP was observed only in the presence of poly(A) tail. Even the shortest A10 tail was sufficient to twofold stimulate translation initiation (Figure 2B). Consequently, in the absence of cap, PABP stimulates 48S formation only in its poly(A)-bound form.

Bound to the poly(A) PABP stimulates termination complex formation more strongly than free PABP

To investigate the impact of PABP on translation elongation, we proceeded to assemble the 80S initiation and elongation complexes on mRNA with a 75-nt poly(A) tail, adding the necessary translation components to the 48S complex. These elongation complexes, containing the UAA stop codon at the A site, are referred to as pretermination complexes (preTC). Our results demonstrated that PABP stimulated the formation of the 80S initiation complex (Figure 3A), probably indirectly through its earlier observed stimulation of the 48S complex formation (Figure 2B). However, PABP did not affect preTC formation, indicating no discernible impact on translation elongation (Figure 3A).

Figure 3.

Effect of poly(A) tail length on elongation and post-termination complexes formation. (A) Toe-printing analysis of the 80S initiation complex and preTC formation in the reconstituted mammalian translation system in the presence or absence of PABP on the model mRNA with 75 nt poly(A) tail (MVHL A75). The graph shows the mean ± standard error, n = 4. Asterisks indicate statistically significant differences (**, P < 0.01), while insignificant differences are not shown. (B) Toe-printing analysis of the postTC formation in the reconstituted mammalian translation system in the presence of PABP and different amounts of release factors on the model mRNAs without poly(A) tail (MVHL A0) and with 75 nt poly(A) tail (MVHL A75). r.u., relative units. The graph shows the height of peaks, corresponding to the postTC, relative to the peaks of the preTC.

In our previous work, we demonstrated that PABP facilitates the loading of the eRF1–eRF3–GTP release factors complex into the ribosome (34). To explore the impact of the poly(A) tail on this process, we assembled the preTC at the UAA stop codon in the presence of PABP on two model mRNAs — one containing a 75-nt poly(A) tail and another without a poly(A) tail. The amount of the preTCs in both complexes was comparable (Figure 3B) since the poly(A) tail does not affect translation elongation (Figure 3A). Subsequent addition of the eRF1–eRF3a termination complex, which recognizes stop codons and triggers peptidyl-tRNA hydrolysis, led to the transition of the preTC into a post-termination complex (postTC). Both complexes can be identified through toe-print analysis, as the cDNA synthesized at postTC and preTC have different lengths due to the distinct conformation of the complexes (Figure 3B) (70). In the presence of the poly(A) tail and 0.08–0.1 pmol of release factors, the postTC formed more efficiently than in the absence of the poly(A). This result suggests a more effective loading of release factors into the ribosome facilitated by poly(A)-bound PABP.

Length of the poly(A) tail increases the efficiency of release factors binding to the ribosome

To investigate the impact of the poly(A) tail length on translation termination efficiency in detail, we employed the Termi-Luc method (64,72). This approach allows the determination of the amount of nanoluciferase (NLuc) released from pretermination ribosomal complexes (preTC-Nluc) during translation termination. NLuc folds into a catalytically active form and induces the luminescence of the substrate only after being released from the ribosome. To obtain preTC-Nluc, containing UAA stop codon in the A site, capped and uncapped Nluc mRNAs were translated in RRL in the presence of an excess of a human mutant eRF1 (eRF1(AGQ)), which is able to bind to the stop codon but is unable to induce peptidyl-tRNA hydrolysis (75). Thus, eRF1(AGQ) suppressed translation termination, leading to the accumulation of preTC-Nluc in the lysate. Subsequently, preTC-Nluc was purified from eRF1(AGQ) and other RRL components by high-salt sucrose density gradient centrifugation. In contrast to previous studies (64,72), we relied on the amount of released Nluc after the addition of excess of release factors rather than on the OD₂₆₀ to select preTC-enriched fractions. In this case, the peak of preTC-Nluc was shifted to the bottom of the gradient compared to the peak of the main fraction of the ribosomes at OD₂₆₀. Fractions from preTC-Nluc peak were pooled and frozen (Supplementary Figure S12). We propose that the peak corresponding to OD₂₆₀ contains ribosomal elongation and initiation complexes, as well as empty 80S ribosomes, explaining the high optical density compared to preTC peak. Equal mRNA capping efficiency was confirmed by an RNAse H-based fluorescent method (Supplementary Figures S1 and S2).

We observed no discernible dependence of the efficiency of preTC-Nluc assembly on the length of the poly(A) tail for both capped and uncapped mRNAs (Supplementary Figures S12 and S13). In contrast to our findings, a prior study (14) demonstrated that the length of the poly(A) tail on translated mRNA in RRL enhances the efficiency of polysome assembly. The absence of a similar effect in our study might be attributed to the additional enrichment of the lysate with components essential for translation. This optimization of reaction conditions could have equalized the efficiency of translation for mRNAs with different poly(A) tails.

To determine the content of translation factors in preTCs-Nluc assembled on mRNAs with different poly(A) tails, we analyzed equal amounts of each complex through Western blot hybridization with antibodies targeting translation factors and ribosomal proteins. The eIF4F cap-binding complex consists of eIF4E, eIF4A, and eIF4G, which interacts with PABP, forming the closed-loop structure (see reviews (15,49)). Given that eIF4B and eIF3 can also bind to eIF4F, we analyzed preTCs-Nluc assembled on capped mRNAs for the presence of these proteins, along with eRF1 and eRF3a. Notably, eIF4B can bind to the initiation complex independently of eIF4F. For preTCs-Nluc assembled on uncapped mRNAs, we focused on the presence of PABP, eIF4B, eRF1 and eRF3. Our observations indicate that on both uncapped and capped mRNAs, the amount of PABP increased with the length of the poly(A) tails (Figure 4A). In the complex assembled on capped mRNA without poly(A) tail, eRF1 and eIF4B were found. The remaining complexes contained, in addition to ribosomal proteins, only various amounts of PABP depending on the poly(A) length (Figure 4A). We hypothesize that initiation factors dissociate from the ribosomal complexes during centrifugation in high-salt sucrose gradient. In contrast, PABP binds to the poly(A) tail of mRNA much more tightly and does not dissociate under these conditions. Thus, as a result, we obtained preTCs-Nluc with an identical protein composition on capped and uncapped mRNA, without any translation factors aside from PABP. Therefore, it is suitable for studying the effect of the amount of poly(A) bound-PABP on peptidyl-tRNA hydrolysis in a pure system.

Figure 4.

Effect of poly(A) tail length on peptide release. (A) Western blot analysis of preTCs-Nluc obtained at capped and uncapped mRNA containing poly(A) tails of different length with antibodies specific to PABP, rpL9, eRF1, eRF3a, eIF4A, eIF4B, eIF4E, eIF4G, eIF3a and eIF3b. As a control, 0.1 pmol of the corresponding purified protein was used. α means antibodies used to detect translation factors. (B) Rate of peptide release v₀ on uncapped preTC-Nluc in the absence and presence of exogenous PABP. RLU, relative luminescent units. The graphs show the mean ± standard error, n = 3. Asterisks indicate statistically significant differences between the values of different curves (*, P < 0.05; **, P < 0.01; ***, P < 0.001), while insignificant differences are not shown.

To the purified preTCs-Nluc, we added eRF1 and eRF3 at various concentrations, and the initial rate of Nluc release was determined from the resulting kinetic curves (Supplementary Figure S14). We observed a weak dependence of the initial rate of peptidyl-tRNA hydrolysis on the length of the poly(A) tail (Figure 4B). This aligns with the increased amounts of PABP with poly(A) tail lengths. Addition of exogenous PABP in an equal amount to release factors significantly increased the rate of peptidyl-tRNA hydrolysis and eliminated the dependence on poly(A) length (Figure 4B).

Based on the dependence of the initial rate of peptidyl–tRNA hydrolysis on the concentration of release factors, we calculated the apparent Michaelis constant (K_M) and the maximum reaction rate (v_max) (Figure 5, S14). Two forms of eRF3 were used in the experiments: a full-length eRF3a, which can be activated by PABP, and eRF3c, lacking the N-terminus and unable to bind to PABP (34,59). We observed that translation termination parameters did not differ between capped and uncapped mRNAs (Figure 5). As mentioned above, these preTCs did not differ from each other in protein composition, as they did not contain initiation factors, with the exception of eIF4B and eRF1 in one complex assembled on the capped mRNA without poly(A) tail. Since the amount of endogenous eIF4B and eRF1 in this complex was insignificant, they did not affect the peptidyl-tRNA hydrolysis.

Figure 5.

Effect of poly(A) tail length on the kinetic constants of peptide release. (A) Schematic representation of the model mRNAs encoding nanoluciferase used in Nluc-preTC assembly. (B) Dependence of the kinetic constants apparent K_M, v_max and k_cat/K_M on the length of poly(A) tail. The graphs show the mean ± standard error, n = 3. Asterisks indicate statistically significant differences (*, P < 0.05; **, P < 0.01; ***, P < 0.001), while insignificant differences are not shown. Red asterisks indicate differences between the values of different curves, blue and orange asterisks indicate differences with the A0 value of the corresponding curve.

Our findings suggest that K_M decreased with poly(A) length in the presence of eRF3a, reaching a minimum at the 100 nt poly(A) tail, and remained relatively constant in the presence of eRF3c (Figure 5). Moreover, the minimal K_M value of eRF3a was close to the value of eRF3c. Thus, we assumed that poly(A) tail extension increases the affinity of eRF3a to preTC and does not affect the already maximal affinity of eRF3c to preTC. This confirms the involvement of PABP in the loading of eRF3a into the ribosome. Additionally, it indicates that the N domain of eRF3a has a role in reducing its affinity to the ribosome in the absence of poly(A)-bound PABP. Interestingly, v_max, representing the fraction of preTCs capable of termination, did not depend on the poly(A) length and was slightly higher in the presence of eRF3a compared to eRF3c (Figure 5).Thus, we showed that poly(A) tail affects the affinity of eRF3a to preTC and does not affect the proportion of preTCs involved in termination. Furthermore, when considering the overall catalytic efficiency (specificity constant) k_cat/K_M, we observed its increase with the poly(A) length in the presence of eRF3a, while it remained constant in the presence of eRF3c (Figure 5). The higher k_cat/K_M for the 100 nt poly(A) tail in the presence of eRF3a compared to eRF3c indicates a positive effect of PABP associated with the long poly(A) tail on translation termination mediated by the eRF1-eRF3a complex, compared to termination mediated by the eRF1-eRF3c complex.

Discussion

In this work, we explored the impact of varying poly(A) tail lengths on diverse stages of eukaryotic translation. During the translation of polyadenylated mRNA, three potential closed-loop mRNA structures can be formed: (i) an initiation loop involving eIF4F-cap and PABP-poly(A), (ii) a termination loop involving eRF3a-eRF1-stop and PABP-poly(A) and (iii) a double loop encompassing both initiation and termination loops with the interaction of initiation and release factors with PABP-poly(A) (15). Through a detailed examination of translation efficiency in all stages using mRNA with different poly(A) tails, we elucidated the optimal poly(A) lengths for each structural formation.

Features of the formation of the initiation closed-loop

The data obtained from the cell-free translation system indicate that the presence of even a short poly(A) tail enhances translation initiation (Figure 1B). To delve deeper into the role of poly(A) in translation initiation, we conducted a detailed examination by assembling the 48S preinitiation complexes in a reconstituted translation system with different poly(A) tails (Figure 2B). Our findings reveal that PABP stimulates both cap-dependent and cap-independent initiation, suggesting a more intricate mechanism for its involvement in translation initiation than previously thought. It is crucial to note that we employed the β-globin leader, which carried an additional unstructured sequence of CAA repeats at the 5′ end to facilitate the assembly of 48S complexes on uncapped mRNA. This addition was necessary to assemble detectable amounts of the 48S complex in the absence of cap. Therefore, since the efficiency of 48S assembly on uncapped mRNA has been artificially heightened, this system may not fully replicate naturally occurring translation initiation in the absence of cap.

The acquired data suggest that PABP exhibits at least two ways of activity in translation initiation. On capped mRNA, PABP can stimulate initiation in both its free and poly(A)-bound forms (Figure 2B). However, data obtained in the cell lysate indicate that in the cellular environment, poly(A)-bound PABP is probably more effective in translation initiation on capped mRNA than free PABP as at the mRNA without poly(A) tail we observed twofold lower translation rate (Figure 1B). On uncapped mRNA containing an unstructured sequence in the 5′ UTR, PABP can stimulate 48S formation only in the poly(A)-bound form (Figure 2B). A recent study demonstrated that eIF4G is incapable of binding to PABP in solution in the absence of the poly(A) tail (47). This observation might elucidate the necessity of at least a 10-nt poly(A) to stimulate initiation (Figure 2B). Presumably, the binding of eIF4G with eIF4A and eIF4E on the cap induces a conformational change in the protein, facilitating eIF4G’s binding to PABP from solution in a reconstituted in vitro translation system. The involvement of the PABP molecule, bound to the first 10 adenines of the poly(A) tail, in translation initiation underscores the requirement for a specific conformation of the mRNA–ribosomal complex and a specific network of ribosome–protein interactions to stimulate 48S complex formation.

In summary, we can conclude the presence of the poly(A) tail is necessary for the stimulation of cap-dependent initiation by PABP. However, in the in vitro system 48S complex formation can be equally effectively stimulated by both free PABP and PABP, bound to the poly(A) tail.

Features of the formation of the termination closed-loop

We hypothesized that during the translation of uncapped mRNA, the poly(A) mainly affects translation termination and mRNA stability, as the formation of a closed-loop structure between eIF4F-cap and PABP-poly(A) is impossible in this case. We propose that, in this scenario, a termination closed-loop is formed between PABP-poly(A) and eRF3a-eRF1, bound to the stop codon on the ribosome. Our findings reveal that the translation rate of uncapped mRNA increases with the length of the poly(A) tail (Figure 1B). However, our experiments showed a linear dependence of mRNA stability on the length of the poly(A) tail. So, what determines the dependence of the translation rate of uncapped mRNA on the length of the poly(A) tail? As we demonstrated in the purified in vitro system, translation termination is also involved into linear dependence of uncapped mRNA translation on the poly(A) length. Consequently, poly(A) length affects both mRNA stability and translation termination.

During the assembly of termination complexes on mRNAs with different poly(A) tail lengths in a reconstituted translation system in the presence of PABP, a dependence of the efficiency of the postTC formation on the poly(A) length was observed (Figure 3B). Subsequent experiments with the purified preTCs revealed that the rate of the peptidyl-tRNA hydrolysis reaction slightly increases with the length of poly(A) tail (Figure 4B). Moreover, we found that increasing the amount of poly(A)-bound PABP decreases the Michaelis-Menten constant, indicating an increase in the affinity of release factors to the stop codon in preTC (Figure 5B). The eRFs affinity increased up to 100 nt poly(A), which is probably the maximum possible value, judging by the activity of eRF3c, independent of PABP. The overall efficiency of the enzymatic reaction increased due to a decrease in K_M. Thus, we concluded that the optimal length of the poly(A) tail for efficient translation termination is about 100 nt. This suggests that at least four PABP molecules, bound to the poly(A) tail, ensure efficient translation termination. Such a cumulative effect cannot be explained by the interaction of only one PABP molecule with release factors during their loading into the ribosome. We assume that in the case of translation termination, PABP also acts as a depot for eRF3a, increasing the local concentration of release factor near the ribosome. Since more PABP molecules are associated with the poly(A) tail near the stop codon, a greater amount of eRF3a is also associated with it and is localized near the stop codon. The effective poly(A) tail length limit of 100 nucleotides is likely determined by the maximum distance from the ribosome for this mechanism to operate.

The data obtained on the purified preTCs, in addition to revealing the positive effect of PABP on the binding of eRF3a to the ribosome, clearly demonstrate the role of the N domain of human eRF3a in the selection of natural stop codons. eRF3c, which lacks this domain, demonstrated high termination efficiency at any length of poly(A) (Figure 5). eRF3a bound to the ribosome weaker than eRF3c in the absence of PABP, and the strength of its binding increased in proportion to the poly(A) bound PABP (Figure 5). Therefore, we assumed that the N domain of eRF3a prevents its spontaneous binding to the ribosomal complex and could provide control of translation termination at premature stop codons that may arise during spontaneous nonsense mutations and are located at a considerable distance from the poly(A) tail. We suppose that the presence of the N domain of eRF3 could prevent translation termination at premature stop codons, reducing the generation of toxic truncated peptides. Interestingly, a truncated form eRF3c was not detected in most of the tested human cell lines (Figures 1 and S8), suggesting that the absence of this form of eRF3 is advantageous for the cell. This observation aligns with the result obtained for yeast eRF3, where N-terminal shortening of eRF3 led to suppression of readthrough of premature stop codons, irrespective of the presence of full-length PABP (76,77).

Features of double closed-loop formation

We identified an optimal poly(A) tail length of 75 nt during the translation of capped mRNA (Figure 1B). Given that translation initiation is independent of the poly(A) tail length (Figure 2), and the efficiency of translation termination increases linearly with the poly(A) length (Figure 5), we propose a hypothesis that a closed-loop structure with double loops could be formed at the 75 nt poly(A) tail. In this structure, initiation and termination complexes are positioned in close proximity to each other, potentially inducing mutual stimulation of termination and initiation factors. The 75 nt poly(A) tail is capable of binding at least three PABP molecules. We suggest that at 100 nt poly(A) bound with 4 PABP molecules, this structure opens, leading to the uncoupling of initiation and termination complexes.

Model of the influence of the poly(A) tail on translation

Based on the obtained results, we propose the following model for the poly(A) tail's effect on translation and the formation of closed-loop structures (Figure 6). (1) In the absence of ribosomes, the 5′ and 3′ ends of mRNA are in close proximity, as supported by studies on mRNA secondary structure (51,65). The first PABP molecule, bound near the 3′UTR and close to the 5′ end of mRNA, interacts with the cap-bound eIF4F. This interaction promotes the landing of the 43S preinitiation complex (PIC) on the mRNA, leading to the formation of the 48S initiation complex. (2) The 48S initiation complex scans the 5′ UTR, and upon reaching the start codon, the 60S ribosomal subunit joins to form the 80S ribosome. During ribosomal subunits joining, eIF4F dissociates from the ribosome but remains bound with the PABP on the poly(A) tail, potentially also with the cap at the 5′ end of mRNA. This protein complex binds with the new 43S PIC and facilitates its landing on mRNA. (3) As the first and subsequent 80S ribosomes move along the mRNA, they unwind its secondary structure and maintain it in the unstructured state. However, upon reaching the stop codon, the ribosome encounters the eRF1-eRF3a complex, bound with PABP on the poly(A) tail. The proximity of the poly(A) tail to the stop codon is preserved by the secondary structure of the 3′UTR, similarly to the mRNA structuring (51,65). Release factors are loaded into the ribosome by any of the four PABP molecules, associated with the poly(A) tail, inducing the hydrolysis of peptidyl-tRNA. Therefore, different PABP molecules located on the same poly(A) tail can simultaneously interact with both the initiation and termination complexes. Since the ribosome dissociates from the mRNA during recycling after translation termination, further melting of the 3′UTR does not occur, and this structure remains stable. (4) The optimal poly(A) tail length for human translation is 75 nt, bound with the three PABP molecules. In this case, a double closed-loop structure is formed where the first PABP molecule interacts with the 48S initiation complex, and the third PABP molecule interacts with the terminating 80S ribosome. 75 nt long poly(A) RNA, bound with PABP, is approximately 330 Å (24), and the human 80S ribosome's linear size is about 250–300 Å (78). Therefore, 75 nt is the minimum poly(A) length to which the ribosome and the initiation complex can be associated simultaneously. In such a structure, the ribosomal complexes should be tightly adjacent to each other. We propose that such close proximity of ribosomal complexes contributes to the mutual activation of initiation and termination stages, enhancing translation. When the poly(A) tail increases to 100 nt, its size expands to 440 Å, which moves the ribosomal complexes away from each other (Figure 6). We hypothesize that the formation of the double closed-loop structure is determined by the physical possibility of ribosomes approaching each other due to their sizes. A longer poly(A) tail interferes with the physical proximity of ribosomes, preventing synergy of initiation and termination.

Figure 6.

Model of the influence of the poly(A) tail length on the translation and closed-loop structure formation. (1) The first PABP molecule binds with the cap-bound eIF4F, promoting 43S PIC loading on the mRNA and 48S initiation complex formation. (2) The 48S initiation complex joins the 60S ribosomal subunit on the start codon, forming the 80S ribosome. eIF4F dissociates from the ribosome but remains bound to PABP on the poly(A) tail. (3) 80S ribosomes, while moving along the mRNA, unwind its secondary structure and maintain it in the untwisted state. Upon reaching the stop codon, the ribosome interacts with the eRF1-eRF3a complex, bound with PABP on the poly(A) tail. Release factors induce the hydrolysis of peptidyl-tRNA. (4) The 75 nt poly(A) tail is bound with three PABP molecules, forming a double closed-loop structure where the first PABP molecule interacts with the 48S initiation complex, and the third PABP molecule interacts with the terminating 80S ribosome. In such a structure, the ribosomal complexes are adjacent to each other, enhancing translation. Created with BioRender.com.

The maximum translation rate observed at 75 nt poly(A) aligns with the peak of distribution of poly(A) mRNA tail lengths in HEK293 (5). The smaller optimal size of the poly(A) tail, found in WGE (Supplementary Figure S1A), can be attributed to the distinct linear size of plant ribosomes, necessitating a different distance for the formation of the double closed-loop structure. Previous studies using electron and cryoelectron microscopy have demonstrated the physical connection of ribosomes located at different ends of the coding sequence during intensive translation (79,80).

In summary, we have elucidated the nuances of translation stimulation by PABP at various stages, particularly its association with the poly(A) tail and identified the dependency of this process on poly(A) length. Our findings bear practical implications for research. Firstly, we suppose that an optimal poly(A) tail of around 75 nt is preferred for efficient mRNA translation in human cells or cell lysates. Secondly, to investigate the synergistic effects of closed-loop structure formation, it is necessary to find the optimal length of the poly(A) tail for each translation system. Our study has demonstrated the relatively narrow range of poly(A) lengths conducive to closed-loop formation, underscoring the importance of exploring this phenomenon across diverse species, including yeast, plants, and mammals, at varying poly(A) tail lengths.

Data availability

The data that support the findings of this study are contained within the article and the supporting information. All source data generated for this study are available from the corresponding author (Dr Elena Alkalaeva; alkalaeva@eimb.ru) upon reasonable request.

Supplementary data

Supplementary Data are available at NAR Online.

Funding

Russian Science Foundation [23-24-00382].

Conflict of interest statement. None declared.