Translation initiation by cap‐dependent ribosome recruitment: Recent insights and open questions

(a) Schematic of the eukaryotic translation cycle showing a single mRNA translated by multiple ribosomes forming a polysome. Canonical translation initiation (green) begins with the assembly of the small ribosomal subunit (SSU), eukaryotic initiation factors (eIFs), and initiator methionyl–tRNA (Met‐tRNA_i^Met) at the mRNA 5′ end, assisted by 5′ cap to 3′ poly(A) interactions. The SSU complex then moves in a 3′ direction until it recognizes a start codon, rearranges its configuration, and is joined by the large ribosomal subunit (LSU). This commences elongation (blue), whereby ribosomes move through the open reading frame (ORF) and synthesize the polypeptide, aided by elongation factors (eEFs). Termination (red) occurs at nonsense (stop) codons; here, eukaryotic release factors (eRFs) assist in the release of the completed protein as well as the LSU and SSU, which can be recycled for additional rounds of translation. The gradual displacement of eIFs at start codons is depicted in light moss green (). Because SSUs at start codons can belong to both initiation and elongation complexes, they are shown in amazonite (). The sequential nature of ribosome assembly during initiation (LSU attaches last) and disassembly during termination (LSU leaves first) is shown by sky blue () and caramel () colors, respectively. Sequential displacement of eRFs with initially recycling followed by initiation factors upon completion of termination is depicted by citron () SSU and pink‐colored () eRFs. (b)–(g) schematics of polysome topologies as detected by electron microscopy. (b) “circular” polysomes observed in rabbit reticulocyte lysate on native mRNA with 5′ cap and 3′ poly(A) (Warner, Rich, & Hall, ). (c) “circular” and “hairpin” polysomes seen in Tetrahymena pyriformis cells (Yazaki, Yoshida, Wakiyama, & Miura, ). (d) “circular” polysomes formed in a wheat germ cell‐free translation system on capped and polyadenylated (left) and noncapped, nonpolyadenylated (right) synthetic mRNAs (Afonina, Myasnikov, Shirokov, Klaholz, & Spirin, ). (e) “linear” polysome formed in a wheat germ cell‐free translation system on capped but nonpolyadenylated mRNA, featuring interacting SSUs along the central axis of a left‐handed (5′ to 3′) spiral of mRNA (Myasnikov et al., ). (f) Endoplasmic reticulum (ER)‐attached “circular”, “linear”, “double‐row”, and “spiral” polysomes detected in rat hepatocytes (Palade, ). (g) ER‐attached “circular”, “hairpin”, and “spiral” polysomes from rat fibroblasts showing SSU‐inward orientation of ribosomes in the tightly packed curved regions (Christensen & Bourne, )

Possible configurations of the initiating SSU as it progresses through the postulated stages of LSU joining. Figure layout as described for Figure . (a1) represents PRE‐translocation‐like state of the SSU:(GTP:eIF5B:Met‐tRNA_i^Met) complex when eIF5B:GTP first attaches to the late‐P_IN SSUs and reorients the Met‐tRNA_i^Met. This complex may now productively bind LSU. (a2) depicts POST‐translocation‐like state of the SSU:(GTP:eIF5B:Met‐tRNA_i^Met) complex, adopted upon attachment to the LSU and ~6° SSU back‐rolling on it along its head‐to‐body vector, fully slotting the Met‐tRNA_i^Met into the LSU P‐site. This stage may coincide with the hydrolysis of the eIF5B‐bound GTP. (a3) represents the first elongation intermediate, with the complete ribosome protecting ~32 nt over the start codons as determined by the in vivo TCP‐seq footprinting (Archer et al., ; Shirokikh et al., ). (b1) structure of the rabbit SSU complexed with eIF5B:GMP‐PNP in the PRE‐translocation‐like state (as in Figure 9b5). (b2) structure of the rabbit SSU complexed with eIF5B:GMP‐PNP in the POST‐translocation‐like state (PDB entry 4UJC) (Yamamoto et al., ). mRNA and eIF1A modeled as described for Figure 9b5. (b3) hypothetical rabbit SSU structure corresponding to the first elongation intermediate, derived from (b2) by removing the eIF5B‐ and eIF1A‐corresponding densities. (c) Corresponds to (b) but from the head‐to‐body point of view. (d) Cartoon schematics corresponding to (c) depicts as described in Figure

Possible configurations of the initiating SSUs as they progress through the start codon recognition stages (1–5). (a) Schematics depicting the possible changes in composition of initiation complexes and matching in vivo TCP‐seq footprint lengths (Archer et al., ; Shirokikh, Archer, Beilharz, Powell, & Preiss, ) (see text). (a1) represents initial pausing over the start codon, while the SSU remains in its scanning configuration featuring the open latch, the head swiveled to the solvent side, and the Met‐tRNA_i^Met displaced laterally from the P‐site in the scanning‐compatible P_OUT conformation; (a2) depicts an early stage of start codon recognition where the mRNA latch is tightened and the mRNA entry more constricted, with Met‐tRNA_i^Met moving further into the P‐site and partially acquiring P_IN state, which induces delocalization of eIF1 within the complex; (a3) shows the stabilized P_IN with the Met‐tRNA_i^Met strongly bound in the P‐site, resulting in the full arrest of SSU sliding on mRNA as eIF1 leaves the initiation complex; (a4) depicts the initiation complex after phosphate release gated by the eIF1 detachment, also resulting in the release of eIF2:GDP and eIF5 from the SSU; (a5) shows the last stage of the process where the Met‐tRNA_i^Met is rebound by eIF5B:GTP, resulting in a readjustment of its position, making it compatible with productive LSU attachment. (b) Structures of the SSU reorganization viewed from the LSU side and corresponding to the start codon recognition stages as shown in (a), with the focus on the intersubunit surface rearrangements. Densities of only mRNA, Met‐tRNA_i^Met, eIFs 1, 1a, 3b, 3i, 3 g, and 5B are shown for simplicity. SSU depicted as described in Figure a. (b1) Structure of the yeast initiating SSU in the open state (PDB entry 3JAQ) (Llacer et al., ) with eIF3b (PDB entry 5K1H) (Simonetti et al., ) modeled as described in Figure a. mRNA model is aligned from the closed state (PDB entry 3JAP) by the 18S rRNA trace in the body region of the SSU. (b2) Same as (b1) but for the SSU in the closed state (PDB entry 3JAP) (Llacer et al., ). (b3) Same as (b2) but with the densities of eIF1 and eIF3b omitted. (b4) Structure of the rabbit late P_IN SSU (PDB entry 5K0Y) (Simonetti et al., ) with the eIF1A density modeled from the yeast closed state (PDB entry 3JAP), by aligning the 18S rRNA trace in the body region of the SSU. (b5) Structure of the rabbit SSU complexed with eIF5B:GMP‐PNP in the PRE‐translocation‐like state (PDB entry 4UJD) (Yamamoto et al., ) during the late start codon recognition stage (derived from the complete ribosome structure by removing LSU‐related densities). mRNA and eIF1A models aligned from the yeast closed state SSU (PDB entry 3JAP) by the 18S rRNA trace in the body region. (c) Same as (b) but in the head‐to‐body view with the head density shown as semitransparent and with gray outline. (d) Schematics of the key interactions between the LSU‐side factors progressing through the start codon recognition stages corresponding to the models in (a)–(c). Head‐to‐body view as in (c) schematized as in Figure . (d1) Multiple interactions between LSU‐side eIFs guided by the C‐ and N‐terminal parts of eIF1A and eIF5 converge on eIF1 binding and stabilize the scanning configuration. (d2) The eIF1A C‐terminal part is displaced from the P‐site upon early P_IN state, leading to the structuring of its N‐terminal part in the A‐ and P‐sites and stabilization of the codon–anticodon interactions. Interaction of the liberated eIF1A C‐terminal part with the N‐terminal part of eIF5 and the liberated eIF5 C‐terminal part with the eIF2β destabilize eIF1 binding. (d3–d5) The eIF1A‐ and eIF5‐guided cascade of rearrangements leads to the loss of affinity to eIF1 and adoption of the stable scanning‐arrested P_IN configuration and gated release of P_i, followed by the eIF2:GDP and eIF5 dissociation

Models of scanning SSU configuration and possible determinants of scanning motion directionality. (a) Combination of the atomic cryo‐EM reconstruction structure (PDB entry 3JAQ) and cartoon schematics of the scanning SSU with the scanning‐conducive, relatively “open” mRNA latch and widened and less‐obstructed mRNA entry marked. Met‐tRNA_i^Met is in the P_OUT conformation, leaning toward the E‐site and with a loose and more lateral positioning in the P‐site. Surfaces of only eIF1, eIF1A (left), and these factors together with eIF2 or eIF3b (right) are shown for simplicity. Atomic ball‐and‐stick model of rRNA is shown in yellow and ribosomal proteins in green. Gray line on the head‐to‐body view outlines the SSU head. The mRNA position was taken from the “closed” SSU structure (PDB entry 3JAP) aligned with 3JAQ by the rRNA trace in the SSU body region. The superimposed eIF3b cartoon was taken from Figure g. (b)–(e) Cartoons illustrating different concepts and models to explain 5′ → 3′ directional motion during scanning. SSU schematics and indications as in Figure . Possible alterations of states of higher (green) and lower (blank) factor affinity to mRNA are shown where applicable. (b) eIF4 group factors perform mRNA “unwinding” in front of the SSU as the structure of the initiation complex directs them to the incoming portion of mRNA (SSU entry) (left plot). The 3′‐ward stretch of single‐stranded mRNA then allows the SSU to spontaneously relocate (diffuse) onto it (middle two plots). Secondary structure can refold upon the mRNA exiting SSU and ensure motion directionality as no helicase activity is present at the SSU exit site (right plot). (c) eIF4 group factors perform “unwinding” in front of the SSU by cycles of clamping to the single‐stranded mRNA. eIF4B has a higher affinity to mRNA (green) while in close contact with eIF4A (left). Upon eIF4A conformational change, eIF4A rebinds to the downstream (incoming) portion of mRNA (green; middle left plot) and then restores its tight binding to the complex, bringing the downstream portion of mRNA along into the SSU and stimulating eIF4B’s high affinity to mRNA (green; middle right plot). The cycle then repeats (right plot). Note that (b) and (c) are most compatible with an entry‐side “threaded” mRNA loading at the beginning of initiation. (d) eIF4 group factors are sequentially lodged behind the SSU as it takes small 3′‐ward advances due to the random (thermal) motion and act as anti‐rollback “pawl” structures. All factors are tightly bound to the SSU, mRNA, and to each other (left plot; green). Due to an induced conformational change, mRNA‐bound factor(s) lose(s) affinity to the rest of the complex but not to mRNA. This allows 3′‐ward SSU sliding along mRNA, with accidental and sequential unwinding of some of the downstream structure (middle left plot). When a sufficient exit portion of the single‐stranded mRNA is exposed, a new molecule of mRNA‐binding factor (X₁) rebinds it, preventing backward SSU sliding (middle right plot). The cycle repeats with the next RNA‐binding molecule (X₂; right plot). (e) Similar to (d), but eIF4‐group factors (eIF4G) and eIF3 also undergo a large‐scale conformational change, possibly induced by the eIF4A enzymatic cycle, and actively pull out a portion of mRNA 3′‐ward, after which they rebind mRNA more proximally to the SSU. Note that (d) and (e) are most compatible with a lateral loading (slotting) of mRNA into the SSU mRNA‐binding cleft, where cap:SSU interactions begin near mRNA exit site

In vivo capture of translation initiation intermediates in yeast cells using translation complex profile sequencing (TCP‐seq). (a) Footprints (FPs) of the SSUs fixed within 5′ UTRs and at start codons of selected mRNAs. FPs are shown near the annotated mRNA 5′ end (‘0’ position on left part of X‐axis before break) or near the start codon in the region corresponding to +/−50 nt from the first nucleotide of the start codon (‘0’ position on right part of X‐axis after break). Metagene modal 5′ (red) and three 3′ (blue) FP extremities (corresponding to the stages of start codon recognition; see Figure and the corresponding text) are marked. (b) FP length characteristics for SSU or complete ribosome (RS) complexes depending on position along mRNAs. TCP‐seq data and some graphic elements were modified from (Archer et al., )

Alternative models for cap‐dependent SSU attachment to mRNA. (a) Schematic of interactions between SSU and eIFs at the cap structure. (b) Model of lateral mRNA attachment to the SSU. eIF4E (red) bound to mRNA cap (purple) first attaches near the mRNA exit site, at the solvent side of the SSU near the head‐to‐body junction. The 3′‐ward portion of mRNA is then “slotted” laterally into the mRNA binding cleft. Once mRNA is stably bound, eIF4E:SSU and likely eIF4E:mRNA interactions are broken as a result of rearrangements upon commencement of scanning, as depicted by possible cap‐bound or “free” eIF4E molecules (red, semitransparent). Solid arrows indicate completed, and dashed – continuing or anticipated, movements or processes. (c) Model of mRNA “extrusion” into the SSU (color scheme and indications as in (b)). Cap‐bound eIF4E first attaches near the SSU mRNA entry. The mRNA then detaches from eIF4E and is sequentially relocated (pushed) into the SSU mRNA‐binding cleft, first reaching the A‐site and continuing to P‐ and E‐sites, and eventually to the mRNA exit site of the SSU. In this case, eIF4E is required to dissociate from the cap structure after mRNA attachment to the SSU. SSU schematics as in Figure

Structures and schematics of ribosomal localization of some of the additional initiation components, including an additional configuration of eIF3, with eIF3j in the entry of the mRNA‐binding cleft of the mRNA‐free SSUs. Designations and SSU schematics are as in Figure . (a) Cryo‐EM structure of human eIF2D bound to the human SSU (left) (Weisser et al., ). Outline of the human eIF2D on yeast SSU (SSU outline as described in Figure ); alignment of structures performed by the 18S rRNA trace (right). (b) Crystal structures of human density regulated reinitiation and release factor (DENR; C‐terminal half, amino acids 110–198) and multiple copies in T‐cell lymphoma‐1 (MCT‐1) bound to the human SSU (left) (Lomakin et al., ) and the outline of these factors shown on yeast SSU aligned to the human SSU as in (a) (right). (c) Cartoon of the cryo‐EM density localization of rabbit DHX29 in complex with SSU, eIFs 1, 1A, GMP‐PNP:2:Met‐tRNA_i^Met, and 3 modified from the original publication (Hashem et al., 2013a) and shown on the yeast SSU outline. (d) Cartoon of eIF3j, eIF3i, and eIF3b positions on the solvent SSU side near mRNA entry as observed in the mRNA‐free “43S” complexes (Erzberger et al., ). A tentative localization of eIF3d shown as a pink circle. Modified from the original publication (Erzberger et al., ) and shown on the yeast SSU outline and the other eIF3 subunits as in Figure g

Structures and schematics of possible ribosomal localization of the main components of cap‐dependent translation initiation for which no precise SSU position is known. SSU depiction as in Figure . The likely SSU localization is shown as a white semitransparent circle (unless stated otherwise), with a wider region of high probability of localization indicated by a dashed black line (right; on the SSU outline). (a) Domain structures of yeast eIF4A and human eIF4AIII (eIF4AI homologue) (left). Crystal structure of yeast eIF4A in the open conformation (Caruthers, Johnson, & McKay, ) and human eIF4AIII in the closed conformation bound to single‐stranded poly(U) RNA and AMP‐PNP (ANP) (Buchwald et al., ) (middle). (b) Domain structures of human eIF4B and its partial homologue eIF4H (aligned by the region of homology; left). Solution structure of the human eIF4B amino acids 96–176 (shown by the gray bar in the human eIF4B domain scheme on the left) (middle) (Fleming et al., ). Position of eIF4B on the SSU outline accounting for directed hydroxyl radical mapping data (Walker et al., ). (c) Domain structures of human eIF4G and DAP5 (eIF4G2, p97, NAT1, AAG1) and yeast eIF4G (TIF4631) (left). Outline of the small angle X‐ray scattering (SAXS) envelope of human eIF4G amino acids 712–1,451 (shown by the gray bar in the human eIF4G domain scheme on the left) (Nielsen et al., ) with the eIF4A‐corresponding density cut out manually (middle top). Crystal structure of the N‐terminal part of human DAP5 (Fan, Jia, & Gong, ) (shown by the gray bar in the human DAP5 domain scheme on the left) (middle bottom). eIF4G SAXS envelope (minus eIF4A) shown in‐scale with the SSU and positioned as originally proposed (Nielsen et al., ). (d) Crystal structure of yeast eIF4G eIF4A‐binding domain amino acids 571–853 (shown by the gray bar in the eIF4G domain scheme on the left) bound to yeast eIF4A:AMP complex (middle) (Schutz et al., ). Outline of the human eIF4A:eIF4G SAXS envelope (see (c)) (Nielsen et al., ) shown in scale with the SSU and oriented by eIF4G eIF3 binding sites to the locations of eIF3c, d, e subunits and the position of the functionally analogous HCV IRES (see text) (right) (Yamamoto, Unbehaun, & Spahn, ). (e) Domain scheme (left) and crystal structure of human eIF4E bound to human eIF4G amino acids 608–642 (middle) (Gruner et al., ). (f) Domain scheme (left) and combined crystal structures of the human PABP RNA recognition motif (RRM) 1–2 domains bound to (A)₉ RNA and amino acids 179–198 of human eIF4G (Safaee et al., ), together with PABP C‐terminal (PABC) domain bound to amino acids 59–73 of human eRF3B (middle) (Kozlov, Menade, Rosenauer, Nguyen, & Gehring, ). (g) Domain scheme (left) and combined solution structure of the N‐terminal GTPase‐activating protein (GAP) domain (Conte et al., ), together with crystal structure of the C‐terminal Pfam W2 domain (middle) (Bieniossek et al., ), of human eIF5. Possible localization on the ribosome taking into account cryo‐EM data (right) (Hussain et al., )

Structures and ribosomal localization (directly available from structural data) of the main components of cap‐dependent translation initiation. Cartoons on the top indicate the point of view relative to the SSU oriented head up, intersubunit side front, and mRNA entry side (opposite to the 5′ cap) to the left. Surfaces are calculated and outlines are derived from published structural data. Rendering was performed with the YASARA (Krieger & Vriend, ) Model. (a) Arrangement of the SSU rRNA (green), P‐site Met‐tRNA_i^Met (purple), and mRNA (red) (Lomakin & Steitz, ). (b) Views of Met‐tRNA_i^Met–mRNA interactions at the P‐site of the initiating SSU (left) (Lomakin & Steitz, ) and schematics of the mRNA and Met‐tRNA_i^Met location on the initiating SSU with key structural features of the SSUs indicated in outline (right) (Llacer et al., ). In the head‐to‐body SSU view, the head is rendered semitransparent to allow observation of the body outline, and the head outline is augmented by a gray bezel line. The darker green areas in all three views (LSU side, head‐to‐body, solvent side) indicate SSU portions that are located to the rear of the main outline plane to improve depth perception. (c) Secondary structure and surface of the globular ribosome‐bound eIF1 core (Lomakin & Steitz, ) with its N‐terminal unstructured extension indicated by the dotted line (left) and outline of its location on the SSU (right) (Llacer et al., ). (d) Same as (c) but for eIF1A (Lomakin & Steitz, ). (e) Secondary structure and surface of yeast eIF2 as part of the SSU‐bound TC (left) and its outline on the SSU (right) (Llacer et al., ). (f) Same as (e) but for yeast eIF3 (Llacer et al., ). (g) Structure of mammalian eIF3 from the eIF3:SSU mRNA‐free complex also containing DHX29 (des Georges et al., ) shown with yeast SSU outline and aligned by its eIF3c subunit to the yeast eIF3c in the eIF3:SSU complex) (Llacer et al., ) using MUSTANG (Konagurthu, Whisstock, Stuckey, & Lesk, ). eIF3a and eIF3c intersubunit exit‐side extensions repeated from yeast structure in (f). eIF3b manually docked from the mammalian eIF3 as part of the mRNA‐bound SSU complex (Simonetti et al., ). (h) Secondary structure and surface of rabbit eIF5B as part of the eIF5B:GMP‐PNP:Met‐tRNA_i^Met complex bound to the complete ribosomes in complex with the hepatitis C virus (HCV) internal ribosome entry site (IRES) (left) (Yamamoto et al., ) and its outline on the yeast SSU (right) (Llacer et al., ). Alignment of the yeast and rabbit structures was performed by the 18S rRNA trace in the YASARA Model

Eukaryotic translation initiation via cap‐dependent scanning of mRNA. Note that, as initiation is highly dynamic, there is some uncertainty regarding the exact temporal sequence of factor interactions and subcomplex assembly. The schematic presents one possible route of initiation, but other routes involving the same factors might well exist, also depending on type of mRNA and experimental system. For example, instead of parallel but separate formations of “multifactor” and “43S” complexes, the involved factors have also been proposed to assemble with each other in a different order (Aylett & Ban, ; Hinnebusch, ; Jackson et al., ). Late stages of eukaryotic initiation are depicted next to the homologous stages of the much simpler prokaryotic initiation process. Key cellular control inputs are further indicated in bold font. Eukaryotic translation initiation factors (eIFs), whose dysregulation prominently associates with the onset or maintaining of malignant phenotype, are shown in red (Bitterman & Polunovsky, ; Chu, Cargnello, Topisirovic, & Pelletier, ; Bramham, Jensen, & Proud, 2016; Wortham & Proud, )