Skip to main content
PMC home page
eLife logoLink to eLife
. 2020 Apr 14;9:e54575. doi: 10.7554/eLife.54575

A complex IRES at the 5'-UTR of a viral mRNA assembles a functional 48S complex via an uAUG intermediate

Ritam Neupane 1,2,†, Vera P Pisareva 3,†, Carlos F Rodriguez 4, Andrey V Pisarev 3,✉, Israel S Fernández 2,✉
Editors: Adam Frost5, John Kuriyan6
PMCID: PMC7190351  PMID: 32286223

Abstract

Taking control of the cellular apparatus for protein production is a requirement for virus progression. To ensure this control, diverse strategies of cellular mimicry and/or ribosome hijacking have evolved. The initiation stage of translation is especially targeted as it involves multiple steps and the engagement of numerous initiation factors. The use of structured RNA sequences, called Internal Ribosomal Entry Sites (IRES), in viral RNAs is a widespread strategy for the exploitation of eukaryotic initiation. Using a combination of electron cryo-microscopy (cryo-EM) and reconstituted translation initiation assays with native components, we characterized how a novel IRES at the 5'-UTR of a viral RNA assembles a functional initiation complex via an uAUG intermediate. The IRES features a novel extended, multi-domain architecture, that circles the 40S head. The structures and accompanying functional data illustrate the importance of 5'-UTR regions in translation regulation and underline the relevance of the untapped diversity of viral IRESs.

Research organism: Human

Introduction

Metagenomic studies of environmental samples have uncovered a great diversity of viruses that have a pervasive presence in the biosphere (Zhang et al., 2019; Zhang et al., 2018; Greninger, 2018). This diversity is especially overwhelming in RNA viruses that infect animal hosts (Shi et al., 2016; Dolja and Koonin, 2018). As strict cellular parasites, viruses rely on capturing cellular ribosomes to gain access to the host machinery for protein production (Jan et al., 2016). In eukaryotes, especially in animals, this machinery is complex and sophisticated, involving large, multi-component protein factors that assist in the operation of eukaryotic ribosomes (Hashem and Frank, 2018). Although complex, translation in eukaryotes conserves four main phases that are also found in its prokaryotic counterparts, namely: initiation, elongation, termination and recycling (Schmeing and Ramakrishnan, 2009). Initiation is significantly expanded in eukaryotes, with two GTP-regulated steps required for the correct positioning of the first aminoacyl-tRNA responsible for setting up the correct reading frame on the messenger RNA (mRNA) (Jackson et al., 2010; Aitken and Lorsch, 2012; Myasnikov et al., 2009; Hinnebusch, 2014).

Eukaryotic initiation starts when the 40S subunit together with the initiation factors eIF1, eIF1A, eIF3, eIF5 and the Ternary Complex (TC) (eIF2–Met-tRNAiMet–GTP) form the 43S Pre-Initiation Complex (43S-PIC), which is competent for mRNA recruitment (Jackson et al., 2010). Eukaryotic mRNAs are then docked to the 43S-PIC at their 5' ends, forming the 48S complex (Hinnebusch, 2017). Once the AUG codon is detected, a structural transition in the 48S from an open, scanning-competent conformation to a closed, scanning-arrested conformation occurs (Hussain et al., 2014). This conformational change is accompanied by the release of eIF1, eIF2 and GDP, leaving the Met-tRNAiMet at the P-site of the 40S base paired with the AUG codon (Aitken and Lorsch, 2012). A second GTP-regulated step, catalyzed by initiation factor eIF5B, is then required for the recruitment of the large (60S) ribosomal subunit (Pestova et al., 2000; Lee et al., 2002). A full (80S) ribosome primed with mRNA and Met-tRNAiMet at the P-site then transitions to the elongation phase (Wang et al., 2019; Voorhees and Ramakrishnan, 2013).

The pathway described above is called the canonical, 5'-end and cap-dependent translation route of initiation (Hinnebusch, 2014). The bulk of eukaryotic mRNAs transitions follow this route, but deviations from the canonical route are common, and normally associated with translation under stress conditions (Starck et al., 2016; Shatsky et al., 2010). Non-canonical initiation is also associated with extended 5' UnTranslated Regions (5'-UTRs) on mRNAs (Sendoel et al., 2017; Young and Wek, 2016). In complex eukaryotes, 5'-UTRs can be very long and can harbor short Open Reading Frames (ORFs) designated as upstream ORFs (uORFs) (Young and Wek, 2016; Wethmar, 2014). Well-studied examples of the functional relevance of uORFs at 5'-UTRs can be found in the yeast stress response regulator GCN4 or the mammalian transcription factor ATF4 (Hinnebusch, 1993; Vattem and Wek, 2004). uAUG codons that are immediately followed by a stop codon (designated as ‘start-stop uORFs’) are also found in the 5'-UTRs of mammalian mRNAs (Wethmar, 2014; Gunišová et al., 2018), but little is known about how these ‘start-stop uORFs’ regulate translation.

Viruses exploit the complexity of eukaryotic initiation to gain access to the host machinery for protein production (Jaafar and Kieft, 2019). Strategies such as mimicking the cap structure or transferring caps from cellular mRNAs (‘cap-snatching’) allow viral mRNAs to hijack host ribosomes, redirecting them towards the production of viral proteins (Jan et al., 2016; Jaafar and Kieft, 2019). A more prominent viral strategy for ribosome hijacking is the use of structured RNA sequences in viral mRNAs (Yamamoto et al., 2017). These sequences are called Internal Ribosomal Entry Sites (IRES), and a tentative classification based on their degree of RNA structure and dependency on canonical initiation factors divided them in four main types (Filbin and Kieft, 2009; Johnson et al., 2017).

The Dicistroviridae family of positive single-stranded RNA ((+)-ssRNA) viruses employs two types of IRESs to express the regulatory versus the structural genes differentially (Nakashima and Uchiumi, 2009). The genome architecture of these viruses functionally segregates both kinds of genes in two ORFs (Figure 1A; Hertz and Thompson, 2011). The first ORF is preceded by an approximately 700-nucleotide 5'-UTR, which harbors an IRES assigned to the type III family (Gross et al., 2017). In vitro characterization of the 5'-UTR-IRES of the Cricket Paralysis Virus (CrPV), a prototypical Dicistrovirus, narrowed down the region of the 5'-UTR responsible for the IRES activity and established the strict requirement of eIF3 for this IRES to initiate translation. Interestingly, the AUG codon of the CrPV ORF1 is immediately preceded by a ‘start-stop uORF’ (Gross et al., 2017).

Figure 1. Dicistroviridae genome organization, in vitro complex formation and cryo-EM maps.

(A) Top, schematic representation of the genome organization of Dicistroviruses. The approximate genomic lengths of the different components are indicated by the arrows. Bottom, detailed view of the region described as harboring the IRES activity of the 5'-UTR of the CrPV. On the right, the sequence adjacent to the initiation AUG codon of ORF1, located at nucleotide 709 and preceded by a ‘start-stop uORF’ indicated in red. (B) Sucrose-gradient analysis showed that the 5'-UTR-IRES is dependent on eIF3 in order to form a stable complex with the 40S. 5'-UTR-IRES co-migrates with the 40S only in the presence of eIF3 (top). By contrast, HCV IRES does not require eIF3 for 40S binding (bottom). (C) Representative cryo-EM micrograph of the 40S–5'-UTR-IRES–eIF3 complex. Bottom, representative reference-free 2D class averages used for further image processing. (D, E) After 3D classifications, two classes showing density for 40S (yellow), eIF3 (red) and 5'-UTR-IRES (blue) could be identified in the dataset. Class-1 (D) presents a non-swiveled configuration of the 40S head and the density for eIF3d is absent. Class-2 (E) shows a swiveled configuration of the 40S head (arrows) with eIF3d (indicated) contacting eIF3's core subunits.

Figure 1.

Figure 1—figure supplement 1. Cryo-EM representative images and classification workflow.

Figure 1—figure supplement 1.

(A) Two examples of aligned micrographs used for image processing. Data collection in thick ice was instrumental to avoid complex disassembly and preferential orientation. (B) Representative reference-free 2D averages. (C) The classification scheme followed to identify the two described complex classes.

Figure 1—figure supplement 2. FSC correlation curves, local resolution and model validation.

Figure 1—figure supplement 2.

Top, class-1. (A) Gold-standard Fourier Shell Correlation (FSC) curves between half-maps independently refined in Relion3. Global resolution by the 0.143 cutoff criterium was estimated to be 3.3 Å. (B) FSC between the final refined model and the final unsharpened map (black curve). The absence of model overfitting is demonstrated by the overlapping of FSC curves between half-map 1 (included in the refinement, blue) and the model and half-map 2 (not included in the refinement, red). (C) Unsharpened map colored according to local resolution as computed by ResMap. On the right, detailed views for the three 5'-UTR-IRES domains. (D) Refined model colored according to estimated B-factors computed by Refmac. Bottom, class-2. (E) Gold-standard FSC curves between half-maps independently refined in Relion3. Global resolution by the 0.143 cutoff criterium was estimated to be 3.3 Å. (F) FSC between the final refined model and the final unsharpened map (black curve). The absence of model overfitting is demonstrated by the overlapping of FSC curves between half-map 1 (included in the refinement, blue) and the model and half-map 2 (not included in the refinement, red). (G) Unsharpened map colored according to local resolution as computed by ResMap. On the right, detailed views for the 40S head before and after masked refinements. (H) Refined model colored according to estimated B-factors computed by Refmac.
Open in a new tab

We sought to characterize the structure of the 5'-UTR-IRES of the CrPV in its ribosome-bound configuration, to gain insights relating to the ribosome-binding determinants of this peculiar IRES, and to understand how the delivery of Met-tRNAiMet is accomplished. Two high-resolution cryo-EM reconstructions of 40S–5'-UTR-IRES–eIF3 complexes, combined with biochemical analysis, allowed us to characterize how this IRES uses an extended structure with a modular, multi-domain architecture to bind to and manipulate the 40S.

Results

The 5'-UTR-IRES of the CrPV requires eIF3 for a stable interaction with the 40S

Previous studies of the IRES located at the 5'-UTR of the CrPV (hereafter referred to as 5'-UTR-IRES) precisely defined the region of the 5'-UTR that is responsible for the IRES activity (residues 357 to 709), as well as its dependency on initiation factor eIF3 for efficient translation initiation (Gross et al., 2017). In contrast to the well-characterized type IV family of IRESs found in the InterGenic Region (IGR-IRES) of these viruses, the 5'-UTRs of Dicistroviruses seem to harbor divergent sequences, making structural modelling based on sequence conservation difficult (Kieft, 2009). In order to address this gap in knowledge, we produced a truncated version of the 5'-UTR region of the genomic RNA of the CrPV that contains the IRES (residues 357 to 728, Figure 1A) in order to obtain structural information about its 40S-bound conformation by electron cryo-microscopy (cryo-EM). We initially tested the in vitro dependency of 5'-UTR-IRES on eIF3 when engaging purified 40S ribosomal subunits in a stable interaction. We assayed the ability of the 5'-UTR-IRES to co-migrate with purified 40S in sucrose density gradients as a test for the presence of a stable complex that is suitable for structural studies (Figure 1B). Unexpectedly, the 5'-UTR-IRES does not form a stable complex with the 40S in the absence of eIF3, in contrast to the HCV-IRES, which is able to form stable complexes with the 40S subunit alone and even with full (80S) ribosomes (Figure 1B; Yokoyama et al., 2019). Sucrose density gradients were manually fractionated from the bottom, where density is heavier. This caused small variations in the position of the 5'-UTR-IRES at the top of the gradients among different experiments. We do not believe such shifts have functional implications.

In the presence of eIF3, however, the 5'-UTR-IRES co-migrates with purified 40S subunits (Figure 1B). This complex revealed clear particles in cryo-EM images, rendering detailed two-dimensional class averages in which density for eIF3 could be identified, albeit at lower threshold (Figure 1C and Figure 1—figure supplement 1). The 40S–5'-UTR-IRES–eIF3 complex exhibited a delicate behavior under cryo-EM conditions, with a strong tendency to disassemble in thin ice. Extensive screening for suitable ice areas was essential to obtain particles of the fully assembled complex (Figure 1C and Figure 1—figure supplement 1). The sample also exhibited a high degree of heterogeneity, which could be resolved by image processing in Relion (Scheres, 2012; Scheres, 2016; Figure 1D,E and Figure 1—figure supplement 2).

Two main classes of particles containing density for 5'-UTR-IRES, 40S and eIF3 were found in the dataset (Figure 1D and E). Both classes contain density for the 40S, the IRES and the core subunits of eIF3 (a/c/e/k/l/f/m), and class-2 also presents density for eIF3 subunit d (Figure 1E, eIF3d). Class-2 exhibits a 40S head in a swiveled configuration. Owing to this swiveled configuration of the 40S head, eIF3d establishes interactions with eIF3a, a core subunit of eIF3 (see below).

Robust density ascribable to the 5'-UTR-IRES could be found in both classes of complex (Figure 1D and E, blue). The ribosome-bound conformation of 5'-UTR-IRES shows an extended configuration, almost circling the 40S head (Figure 2A and Figure 2—figure supplement 1). Three domains connected by flexible linkers could be defined: an elongated domain I (DI) at the back of the 40S head contacting ribosomal proteins uS3 and RACK1 (Figure 2), a second domain (DII) formed by a dual hairpin at the back of the 40S body interacting with eIF3 (Figure 3), and a third, large helical domain (DIII) placed at the periphery of the 40S E-site, contacting ribosomal proteins uS7 and uS11 (Figure 4).

Figure 2. The 5'-UTR-IRES domain I engages ribosomal proteins RACK1 and uS3.

(A) Overview of the 40S–5'-UTR-IRES–eIF3 map, with 40S and eIF3 depicted gray and 5'-UTR-IRES in blue. (B) Detailed view of the cryo-EM density of the 40S–5'-UTR-IRES–eIF3 map centered around 5'-UTR-IRES domain I (DI). Ribosomal proteins are colored brown, 18S rRNA yellow and 5'-UTR-IRES blue. Contacts between 5'-UTR-IRES domain I and ribosomal proteins RACK1 (C) and uS3 (D) could be defined thanks to well-resolved local cryo-EM densities.

Figure 2.

Figure 2—figure supplement 1. Structurally derived secondary structure diagram for the 5'-UTR-IRES.

Figure 2—figure supplement 1.

Secondary structure diagram for the 5'-UTR-CrPV IRES derived from the cryo-EM structure.
Open in a new tab

Figure 3. The 5'-UTR-IRES domain II is formed by a dual hairpin that mediates eIF3 recruitment.

Figure 3.

Open in a new tab

(A) Overview of the 40S–5'-UTR-IRES–eIF3 cryo-EM map with 40S colored gray, eIF3 red and 5'-UTR-IRES blue. On the right, a zoomed view centers around 5'-UTR-IRES domain II. (B) Detailed view of the cryo-EM map for the region occupied by 5'-UTR-IRES domain II, with 40S components colored gold, eIF3 red and 5'-UTR-IRES blue. Domain II is sandwiched between ribosomal protein uS17 (located at the back of the 40S body) and eIF3 core subunits a and c. (C) 5'-UTR-IRES domain II is formed by a dual hairpin that establishes interactions with α-helices 8 and 10 from eIF3a. These contacts are mediated mainly by basic residues of eIF3 and the phosphate backbone of the IRES. (D) superposition of the 40S–5'-UTR-IRES–eIF3 complex with the canonical 48S complex (left, PDB ID 6FEC) and with the CSFV-IRES–40S complex (right, PDB ID 4c4q). The 5'-UTR-IRES binds to the 40S with a conformation that is compatible with the canonical position described for eIF3 in the 48S complex.

Figure 4. Non-canonical base pairing in 5'-UTR-IRES domain III assists on P-site access.

Figure 4.

Open in a new tab

A) Overview of the 40S–5'-UTR-IRES–eIF3 cryo-EM map with 40S and eIF3 colored gray and 5'-UTR-IRES blue. On the right, a detailed view of the E-site, where 5'-UTR-IRES domain III is placed, shows the cryo-EM map with the 5'-UTR-IRES colored blue and 40S components brown. On the far right, the final refined model is colored following the same color scheme. Ribosomal proteins uS7 and uS11 as well as several 18S rRNA bases contact 5'-UTR-IRES domain III. (B) Non-canonical base pairs found in 5'-UTR-IRES domain III induce a distortion of the double helix near the E-site. At the top, two examples are shown with the refined model inserted in the experimental cryo-EM density. The corresponding chemical diagrams below show the base edges and the hydrogen bonds involved in interactions. (C) Superpositions of the canonical 48S complex (PDB ID 6FEC, top) with the HCV-IRES/40S complex (PDB ID 5A2Q, bottom) and the 40S/5'-UTR-IRES/eIF3 (middle) models, focused on the tRNA A, P and E binding sites. 5'-UTR-IRES domain III and HCV-IRES occupies a space on the E-site that overlaps with the position described for eIF2 in the canonical 48S complex. In the middle, the last residue of 5'-UTR-IRES is indicated (C695), as is the putative path along the mRNA binding channel followed by the IRES (dashed line).

Domain I of the 5'-UTR-IRES contacts the ribosomal proteins RACK1 and uS3

The 5' proximal segment of the 5'-UTR-IRES (residues 357 to 486) forms domain I, which is characterized by an elongated T-shaped structure anchored to the back of the 40S head (Figure 2A and B). A long helical segment in this domain ‘wraps’ around the apical part of ribosomal protein RACK1. Two bases of this helical segment of domain I, C442 and C444, are extruded from the body of the double helix to establish hydrophobic interactions with tyrosine residue 140 of RACK1 (Figure 2C). These interactions bend the main helical segment of the 5'-UTR-IRES DI, directing the tip of this domain towards ribosomal protein uS3 (Figure 2D). Guanine residue 395 is inserted deep into a hydrophobic pocket of ribosomal protein uS3, establishing contacts with main-chain atoms of this ribosomal protein. In this location, 5'-UTR-IRES DI is found adjacent to the mRNA entry channel of the 40S, overlapping with the space previously described as being occupied by the helicase DHX29 involved in canonical initiation (Figure 2D; Hashem et al., 2013a).

5'-UTR-IRES binding to the 40S is compatible with a canonical configuration of eIF3

The second domain of 5'-UTR-IRES (DII) is connected to domain I by a flexible linker that is poorly defined in our maps as it is exposed to the solvent. This second domain of the 5'-UTR-IRES is formed by a dual hairpin and is wedged between the back of the body of the 40S and eIF3 subunits a and c (Figure 3A and B). Ribosomal protein uS17 peripherally contacts this domain, establishing interactions with the phosphate backbone of the IRES (Figure 3B). A network of interactions involving eIF3 subunits a, c and h anchors DII to this position (Figure 3C). These interactions are also established through contacts between positively charged residues on eIF3 and the phosphate backbone of the IRES. No contacts with specific IRES bases could be observed.

Currently, medium-resolution cryo-EM reconstructions for 40S complexes containing eIF3 and the rest of the components of the canonical 48S complex are available (Eliseev et al., 2018) (PDB ID 6FEC), as are such reconstructions for the 40S in complex with eIF3 and the CSFV-IRES (Hashem et al., 2013b) (PDB ID 4c4q). Comparisons of these structures with our complex reveal a positioning of eIF3 relative to the 40S that is very similar to the canonical 48S complex and different from the position adopted by eIF3 in the CSFV-IRES–40S complex (Figure 3D). In the 48S canonical configuration, eIF3 contacts the 40S through helix 1 of eIF3a and helix 22 of eIF3c, as well as through eIF3d, which is isolated in its 40S interaction, away from the core subunits of eIF3 (Figure 3D, left). The CSFV-IRES engages the 40S, displacing eIF3 from its position in the canonical 48S (Figure 3D, right). In addition, in the canonical 48S complex, eIF3 interacts with the 40S peripherally, allowing the formation of cavities between eIF3 and the back of the 40S. These cavities are exploited by the 5'-UTR-IRES, which inserts its domain II into one of them, adopting a configuration that is compatible with the binding of eIF3 to the 40S in the canonical 48S complex (Figure 3D, middle). No major rearrangement of eIF3 (compared to its position in the canonical 48S complex) is required for the binding of the 5'-UTR-IRES, so there could be an advantage in hijacking preformed cellular 48S complexes that are ready to transit the cap-dependent route of initiation.

Non-canonical base pairing in the 5'-UTR-IRES DIII places the uAUG codon near the P-site

Threading through the 40S channel formed by ribosomal proteins uS7 and uS11, a flexible single stranded linker connects DII with DIII (Figure 4A). DIII forms a prominent, helical mass in the surroundings of the E-site of the small subunit at the inter-subunit face of the 40S. The helical segment is very well defined in our maps because it is stabilized by numerous contacts with ribosomal proteins uS7 and uS11 and with 18S ribosomal RNA (rRNA) bases (Figure 4 and Figure 2—figure supplement 1). However, the distal part of this domain forms two short stem loops that, given their flexibility, could only be modelled at low resolution.

Inspection of the cryo-EM density reveled a distortion in the canonical double helix of the main segment of this domain as it approaches the E-site. The quality of the maps in this area allowed de novo modelling of these residues, revealing a set of non-canonical interactions between the RNA bases (Figure 4A and B). In-plane triple-base interactions involving sugar and the Hoogsteen edges of the bases, as well as purine–purine Hoogsteen base pairs, could be found in this stretch of residues of the helical segment of DIII (Figure 4B; Leontis and Westhof, 1998). Overall, these non-canonical base pairs induce a distortion at the base of DIII that helps to position the single-stranded segment of the 5'-UTR-IRES harboring the uAUG codon at position 701 in the mRNA-binding channel of the 40S (Figure 4C, middle). The 5'-UTR-IRES accesses the 40 S P-site through the E-site, blocking a concurrent recruitment of the TC (eIF2–Met-tRNAiMet–GTP, Figure 4C). Interestingly, a similar strategy is followed by the HCV-IRES. A superposition of the structure of the HCV-IRES in complex with the 40S (Yamamoto et al., 2015; Quade et al., 2015) (PDB ID 5A2Q) with our structure reveals a very similar positioning of the domain II of HCV-IRES, accessing the P-site through the E-site to position the AUG codon in the surroundings of the P-site (Figure 4C, bottom). Even though both IRESs differ markedly in their interaction with the back of the 40S and eIF3, both converge to similar structural solutions for the placement of the AUG initial codon close to the 40 S P-site.

Swiveling of the 40S head locks the 5'-UTR-IRES, inducing a compact conformation of eIF3

Initial processing of the cryo-EM data revealed flexibility of the 40S head. Masked classification and refinement in Relion3 (Zivanov et al., 2018) revealed two major populations of particles, which differ in the degrees of 40S head swiveling (Figure 1). The 40S head is attached to the body by a single RNA helix, making this component of the ribosome extremely flexible (Johnson et al., 2017). Intrinsic and independent movements of the 40S head are instrumental in tRNA translocation and also in canonical initiation (Ratje et al., 2010; Flis et al., 2018). The 5'-UTR-IRES seems to exploit this intrinsic dynamic to bind to the 40S and then to ‘lock’ the IRES in a specific conformation that commits the complex towards viral translation (Figure 5). In class-1 (open conformation), the head of the 40S shows an almost canonical configuration with very little swiveling and no tilt. In this conformation, the latch of the 40S (an early defined contact between the head and the body of the 40S [Frank et al., 1995]) is closed. At the other side of the 40S head, access to the channel formed by ribosomal proteins uS7 and uS11 is exposed and eIF3d density is not well defined, probably because of a high degree of flexibility or low occupancy (Figure 5A, left). In class-2 (closed conformation), the head of the 40S exhibits a medium-range degree of swiveling when compared to the widest displacement reported (Ratje et al., 2010).

Figure 5. A 40S head swiveling movement ‘locks’ 5'-UTR-IRES on the 40S, inducing a compact eIF3 configuration.

Figure 5.

Open in a new tab

(A) Cryo-EM maps obtained for the two classes present in the 40S–5'-UTR-IRES–eIF3 dataset with 40S colored gold, eIF3 red and 5'-UTR-IRES blue. The positions of the latch, eIF3d and the swiveling rotation axis are indicated. A 40S head swiveling movement in class-2 brings eIF3d closer to the core subunit of eIF3 (i.e. eIF3a), establishing interactions that stabilize its conformation. (B) Left, ribbon diagram colored according to pairwise root mean square deviation (r.m.s.d.) displacements for the open-to-closed transition, with displacements scale at the center. On the right, a simplified diagram shows only the 18S rRNA colored according to the used on the left. Two orthogonal views are shown, in which it can be appreciated that the main displacement is localized at the 40S head. (C) Overview of the closed class with 40S colored gray, eIF3 red and 5'-UTR-IRES blue. Inset, detail of the experimental density obtained for the eIF3a–eIF3d interface for this class. Clear information on side chains was present in the maps, allowing proper model building and refinement.

In the open and closed classes, the positions of the 5'-UTR-IRES relative to the 40S head are very similar (Figure 5A, right and B). In the swiveled conformation, the latch is open, and the channel formed by ribosomal proteins uS7 and uS11 is plugged by eIF3d, which in this class presents robust density (Figure 5C). The main subunits of eIF3 (a/c/e/k/l/f/m) show a similar conformation in both classes, having a similar orientation with respect to the 40S body (Figure 5B). In the swiveled configuration (class-2), the 40S head brings eIF3d close to eIF3a, one of the core subunits of eIF3 (Figure 5C). Well-defined density in this area could be observed for the eIF3a–eIF3d interface (Figure 5C, right). This compact state of eIF3 represent a hitherto unknown conformation (Lee et al., 2016).

TC delivers Met-tRNAiMet to uAUG at position 701, and initiation factors eIF1 and eIF1A assist in AUG location

Our structures of the 40S–5'-UTR-IRES–eIF3 complex revealed a positioning of the DIII of the IRES that overlaps with the position that the TC populates at the E-site in canonical initiation (Figure 4C; Hussain et al., 2014; Eliseev et al., 2018). In addition, in our maps, we could only confidently identify density for the single-stranded segment of RNA of the IRES placed close to the P-site until residue 695, whereas the canonical AUG of ORF1 is found at nucleotide 709. These facts prompted us to wonder how the delivery of Met-tRNAiMet to the AUG is accomplished. Making use of an in vitro reconstituted mammalian initiation assay with native components and toe-printing analysis (Kolupaeva et al., 2007), we analyzed the different steps followed by the 5'-UTR-IRES in order to place Met-tRNAiMet based paired with the AUG codon at the P-site (Figure 6A). Translation initiation factors in mammals and insects are highly homologous. In particular, eIF2-alpha shares 57% identity and 74% similarity between human and Drosophila, eIF2-beta 74% identity and 83% similarity, eIF2-gamma 82% identity and 88% similarity, eIF5B 71% identity and 85% similarity, eIF3a 46% identity and 63% similarity, and eIF3c 51% identity and 66% similarity. This high level of homology justifies the utilization of mammalian initiation factors for CrPV analysis, as has been done before for the CrPV IGR-IRES.

Figure 6. The 5'-UTR-IRES requires TC, eIF1 or eIF1A to assemble a functional initiation complex via an uAUG intermediate.

(A) Toe-print analysis of 48S initiation complexes assembled on 5'-UTR-IRES in an in vitro reconstituted system. eIF2 delivers Met-tRNAiMet to the uAUG (lane 2) and requires the presence of eIF1 or eIF1A to transition to the bona fide AUG codon of ORF1 (lane 3). Under some conditions, eIF5B can substitute for eIF2 in Met-tRNAiMet delivery (Terenin et al., 2008). In the absence of eIF1 or eIF1A, a robust toe-print signal is detected in the presence of eIF5B (lane 4), however, Met-tRNAiMet is delivered to the uAUG, and thus eIF5B is unable to find the annotated AUG even in the presence of eIF1 or eIF1A (lane 5). (B) A model for 5'-UTR-IRES-mediated translation initiation. From the bottom right: injection of the genomic (+)-ssRNA of the CrPV into the cytosol allows the 5'-UTR-IRES to capture free 40S and eIF3, which is recruited to the complex with an initial canonical conformation of the 40S head without tilt or swivel. Insertion of 5'-UTR-IRES DIII in the vicinity of the E-site and a swiveling movement of the 40S head induces a ‘locked’ conformation of the complex, with the uAUG at 701 in the vicinity of the P-site. Delivery of Met-tRNAiMet and location of ORF1 AUG is achieved by the concerted action of eIF2, eIF1 and eIF1A. Large subunit recruitment mediated by eIF5B will allow transitioning into elongation.

Figure 6.

Figure 6—figure supplement 1. Control toe-print experiments.

Figure 6—figure supplement 1.

(A) Toe-print analysis for 5'-UTR-IRES, 5'-UTR-IRES–eIF3 and 40S/5'-UTR-IRES–eIF3 complexes (lanes 1, 2 and 3, respectively). No toe-print signal was detected that is ascribable to uAUG or annotated AUG in any of these reactions. Robust toe-print signal could be detected 15–17 nucleotides downstream of uAUG and AUG only for 40S–5'-UTR-IRES–eIF3, and in the presence of TC (eIF2–Met-tRNAiMet–GTP) and eIF1 or eIF1A. (B) The uAUG toe-print signal can be abolished by mutation of the uAUG to ACG, redirecting the Met-tRNAiMet loading event exclusively to the annotated AUG in the presence of TC and eIF1 or eIF1A.
Open in a new tab

Toe-print assay permits identification of the location of functional ribosomal complexes assembled on mRNAs by reverse transcription of a primer annealed to the mRNA. The length of the resulting extended DNA fragment provides information about the position of the ribosome on the mRNA. Due to its large size, the paused ribosome protects a segment of the mRNA, precluding further primer extension and generating toe-print signals approximately 15–17 nucleotides downstream of the P-site of 40S. Cognate aminoacyl or peptidyl-tRNAs in the P-site or post-termination complexes with eRF1 in the A-site yield robust toe-print signals (Skabkin et al., 2013). The 40S–5'-UTR-IRES–eIF3 complex produces signal that is ascribable to the secondary structure of the 5'-UTR-IRES (Figure 6—figure supplement 1, lanes 1–3), indicating no measurable pausing of the ribosome on the mRNA around any of the AUG codons. In isolation, the TC (eIF2–Met-tRNAiMet–GTP) is able to load Met-tRNAiMet onto the P-site of the 40S–5'-UTR-IRES–eIF3 complex, producing a robust toe-print (Figure 6A, lane 2, label 48S–uAUG) 15–17 nucleotides away from the uAUG located at 701. A similar uAUG delivery of Met-tRNAiMet can be accomplished by eIF5B which, under stress conditions, has been described as substituting for eIF2 in Met-tRNAiMet delivery (Terenin et al., 2008; Pestova et al., 2008; Yamamoto et al., 2014; Kenner et al., 2019), with eukaryotic initiation then following a ‘bacterial-like’ mode (Figure 6A, lane 4). Transitioning to the correct AUG could only be detected in the presence of eIF1and eIF1A, and only when the TC was present, and not for eIF5B (Figure 6A, lanes 3 and 5). Notably, the presence of eIF1and eIF1A seems to be detrimental to uAUG Met-tRNAiMet loading by eIF5B, as their presence significantly reduces the toe-print signal that can be observed for eIF5B in isolation. However, no concomitant increase in toe-print signal for the canonical AUG could be observed for the eIF1/eIF1A/eIF5B reaction.

Only eIF2 as part of the TC and assisted by eIF1 and eIF1A can properly locate the bona fide AUG of ORF1. The role of the uAUG located at nucleotide 701 is not clear, but the fact that eIF5B can deliver Met-tRNAiMet only to this codon points towards an important role for this uAUG in initiation when eIF2 is unavailable.

Discussion

Ribosome-profiling datasets have revealed the presence of translating ribosomes paused on 5'-UTRs, implying a decisive role of these sequences in regulating translation, especially under stress conditions (Sendoel et al., 2017; Archer et al., 2016; Andreev et al., 2015; Ingolia et al., 2009; Brar and Weissman, 2015; Resch et al., 2009).

The 5'-UTR of the (+)-ssRNA of the Dicistovirus CrPV harbors an IRES that is able to direct initiation towards ORF1 in the early phase of infection (Hertz and Thompson, 2011; Garrey et al., 2010). Expression of ORF1 is instrumental for virus replication because the RNA-dependent RNA polymerase (RdRp), and the protease responsible for the proteolytic digestion of the polyprotein containing the structural proteins, are encoded in ORF1 (Jan et al., 2003).

The 5'-UTR-IRES features a novel multi-domain, extended architecture that encircles three quarters of the 40S head, exploiting binding sites not previously described for any IRESs (Figures 2, 3 and 4). Ribosomal proteins uS3 and RACK1 are used by the IRES to anchor its DI to the back of the 40S head (Figure 2). The structure thus rationalizes previous data showing a preeminent role of RACK1 in CrPV and related viruses that infect Drosophila (Majzoub et al., 2014). The interaction of DI with RACK1 is also instrumental in positioning DII at the back of the 40S body, sandwiched in between ribosomal protein uS17 and eIF3 (Figure 3). Interestingly and in contrast with the HCV-IRES, the conformation observed for eIF3 in the complex with 5'-UTR-IRES is very similar to that observed for eIF3 in the 48S complex, with the IRES ‘filling up’ cavities that are present between the 40S and eIF3 in this canonical complex (Eliseev et al., 2018). The HCV-IRES and related IRESs, such as the CSFV-IRES, displace eIF3 from its canonical location using a very different mechanism for IRES docking to the 40S (Hashem et al., 2013b).

In order to place the AUG of ORF1 in the surroundings of the P-site, the 5'-UTR-IRES accesses the P-site through the E-site, in a manner similar to that of the HCV-IRES (Figure 4C; Yamamoto et al., 2015). In this aspect, the 5'-UTR-IRES recapitulates binding strategies that are known for other IRESs, such as the IGR-CrPV-IRES that also makes use of ribosomal protein uS7 for its binding to the ribosome or the HCV-IRES that places its domains II and IV in the surroundings of the P-site, sliding the elongated DII from the back of the 40S to the P-site through the E-site (Pisareva et al., 2018).

The placement of the AUG of ORF1 in the surroundings of the P-site seems to be exerted by a mechanism involving the intrinsic dynamics of the 40S head (Johnson et al., 2017; Figure 5). The 5'-UTR-IRES exploits the characteristic swiveling movement of the 40S head to bind and progress towards a conformation that ‘locks’ the IRES onto the 40S, and at the same time, induces a compact conformation of eIF3 that has subunit eIF3d in close contact with the core subunits of eIF3 (Lee et al., 2016). These dynamics are probably instrumental for the ability of the 5'-UTR-IRES–40S complex to localize the annotated AUG, in a genomic context where a uAUG-stop is physically close. The capacity of the 40S to scan an mRNA bidirectionally upon termination on a stop codon has been previously reported (Skabkin et al., 2013). It is thus plausible that the peculiar genetic configuration of the CrPV around the annotated AUG of ORF1 (Figure 1A) evolved to leverage these re-initiation mechanisms already present in the translation of cellular messengers. However, these considerations are highly speculative, as the particular role that the uAUG exerts in Met-tRNA-iMet recruitment, or more generally its involvement in initiation of viral messengers, remains enigmatic. A comprehensive understanding of the role of uAUG and the start-stop configuration will demand further studies, ideally in vivo.

We propose the following model for how the 5'-UTR-IRES of the CrPV operates: immediately after the (+)-ssRNA genomic molecule of the CrPV is injected into the cytoplasm of the host cell, the IRES harbored at the 5'-UTR captures 40S subunits (Figure 6B, bottom). Recruitment of eIF3 is mediated by DII, allowing the sliding of the flexible linker connecting DII and DIII between the head and the platform of the 40S to place DIII in the surroundings of the E-site (Figure 6B, bottom and left). A swivel movement of the 40S head closes the channel between the head and the platform of the 40S, effectively ‘locking’ the 5'-UTR-IRES into the 40S and inducing a compact conformation of eIF3 with the eIF3d subunit in interacting distance with eIF3's core subunit a (Figure 6B, left top). With this configuration, eIF2 as part of the TC can deliver Met-tRNAiMet to the uAUG located at nucleotide 701, and further assistance by initiation factors eIF1 and eIF1A allows for a downstream location of the AUG codon of ORF1 at nucleotide 709. Large subunit recruitment grants transitioning towards elongation, committing the ribosome to the production of viral proteins (Figure 6B, right top).

In summary, we have structurally characterized the 5'-UTR-IRES of the CrPV in its ribosome-bound state and have characterized the delivery of Met-tRNAiMet by eIF2 and eIF5B. Given the rich diversity of viral sequences in the animal virome, new IRESs exploiting different aspects of animal translation will probably be discovered.

Materials and methods

5'-UTR-IRES and HCV IRES production

For cryo-EM analysis, a transcription vector for 5'-UTR-IRES (nucleotides 357–728) was constructed by inserting a T7 promoter sequence upstream of the 5'-UTR-IRES sequence followed by a BamHI restriction site, using pUC19 as a scaffold vector. For toe-print assays, 5'-UTR-IRES with the extended ORF part for primer annealing was cloned by a similar strategy. The uACG-AGA mutant was obtained by site-directed mutagenesis of 5'-UTR-IRES. T7 RNA polymerase in vitro transcription and purification on Spin-50 mini-column (USA Scientific) were used to obtain highly purified RNAs.

Purification of translation components and ribosomal subunits

Native 40S subunits, eIF2, eIF3, eIF5B and rabbit aminoacyl-tRNA synthetases were prepared as previously described (Pestova and Hellen, 2005). Recombinant eIF1 and eIF1A were purified according to a previously described protocol (Kolupaeva et al., 2007). In vitro transcribed Met-tRNAiMet was aminoacylated with methionine in the presence of rabbit aminoacyl-tRNA synthetases as previously described (Pisarev et al., 2010).

Assembly of ribosomal complexes

To reconstitute different ribosomal complexes for toe-print assays, we incubated 0.3 pmol 5'-UTR-IRES RNA with 1.8 pmol 40S subunits, 10 pmol eIF1, 10 pmol eIF1A, 10 pmol eIF2, 5 pmol eIF3, 5 pmol eIF5B, and 5 pmol Met-tRNAiMet, as indicated, in a 20 μL reaction mixture containing buffer A (20 mM Tris-HCl [pH 7.5], 100 mM KCl, 2.5 mM MgCl2 and 1 mM DTT) with 0.4 mM GTP and 0.4 mM ATP for 10 min at 37◦C. We analyzed the assembled ribosomal complexes via a toe-print assay, essentially as described by Pestova and Hellen (2005). For the sucrose density gradient experiment, we incubated [32P]-labelled 5'-UTR-IRES or HCV-IRES RNAs co-transcriptionally with 3.7 pmol 40S subunits and 11 pmol eIF3, as indicated, in a 60 µL reaction mixture containing buffer A for 10 min at 37°C, subjected the samples to a 10–30% sucrose density gradient centrifugation, and analyzed the gradient fractions by radioactivity counting.

Cryo-EM sample preparation and data acquisition

Aliquots of 3 μl of assembled ribosome complexes at a concentration range of 250–350 nM were incubated for 30 s on glow-discharged holey gold grids (Russo and Passmore, 2014) (UltrAuFoil R1.2/1.3). Grids were blotted for 2.5 s and flash cooled in liquid ethane using a FEI Vitrobot. Grids were transferred to a FEI Titan Krios microscope equipped with an energy filter (slits aperture 20 eV) and a Gatan K2 detector operated at 300 kV. Data were recorded in counting mode at a magnification of 130,000, corresponding to a calibrated pixel size of 1.08 Å. Defocus values ranged from 1 μm to 3.6 μm. Images were recorded in automatic mode using the Leginon (Carragher et al., 2000) and APPION (Lander et al., 2009) software and frames were aligned using the Relion3 (Zivanov et al., 2018) implementation of the Motioncor2 algorithm (Zheng et al., 2017).

Image processing and structure determination

Contrast transfer function parameters were estimated using GCTF (Zhang, 2016), and particle picking was performed using GAUTOMACH without the use of templates and with a diameter value of 260 pixels. All 2D and 3D classifications and refinements were performed using RELION. An initial 2D classification with a 4 times binned dataset identified all ribosome particles. A consensus reconstruction with all 40S particles was computed using the AutoRefine tool of RELION. Next, 3D classification without alignment (four classes, T parameter 4) identified a class with unambiguous density for eIF3. This class was independently refined, and further masked classification allowed the identification of two subclasses that are distinguishable by a different degree of 40S head swiveling and by the presence or absence of eIF3d density. Final refinements with unbinned data for the selected classes yielded high-resolution maps with density features in agreement with the reported resolution. Local resolution was computed with RESMAP (Kucukelbir et al., 2014).

Model building and refinement

Models for the mammalian 40S and eIF3 docked into the maps using CHIMERA (Pettersen et al., 2004) and COOT (Emsley et al., 2010) were used to adjust these initial models manually. 5'-UTR-IRES was built manually using COOT. An initial round of refinement was performed in Phenix using real-space refinement (Afonine et al., 2018) with secondary structure restraints and a final step of reciprocal-space refinement with REFMAC (Murshudov et al., 1997). The fit of the model to the map density was quantified using FSCaverage and Cref and model-to-maps over-fitting tests were performed following standard protocols in the field (Brown et al., 2015; Amunts et al., 2014).

Cryo-EM data collection, refinement and validation statistics
Class-1 (open)
(EMDB-21529) (PDB 6W2S)		 Class-2 (closed)
(EMDB-21530) (PDB 6W2T)
Data collection and processing
Magnification
Voltage (kV)
Electron exposure (e–/Å2)
Defocus range (μm)
Pixel size (Å)
Symmetry imposed
Initial particle images (no.)		 130,000
300
59.55
−1 /– 3
1.06
C1
915,647
Final particle images (no.) 14,257 23,444
Map resolution (Å)
FSC threshold		 3.3
0.143		 3.3
0.143
Map resolution range (Å) 3–8 3–8
Refinement
Initial model used (PDB code) 5A2Q 5A2Q
Model resolution (Å)
FSC threshold		 3.6
0.5		 3.6
0.5
Model resolution range (Å) 3.3–8 3.3–8
Map sharpening B factor (Å2) −31.94 −43.41
Model composition
Non-hydrogen atoms
Ligands		 106,817
-		 109,684
-
B factors (Å2)
Protein
RNA		 92.47
114.1		 96.5
117.4
R.m.s. deviations
Bond lengths (Å)
Bond angles (°)		 0.014
1.77		 0.014
1.78
Validation
MolProbity score
Clashscore
Poor rotamers (%)		 2.12
6.13
1.62		 1.99
4.92
1.39
Ramachandran plot
Favored (%)
Allowed (%)
Disallowed (%)
RNA validation
Angles outliers (%)
Sugar puckers outliers (%)
Average suit		 88.92
98.37
1.63
0.18
2.35
0.442		 90.25
98.50
1.50
0.17
2.25
0.428
Open in a new tab

Acknowledgements

We are grateful to Dr Jean-Luc Imler for a generous donation of a CrPV-5'-UTR-IRES plasmid. We are thankful to Prof. Jennifer Doudna for a generous donation of an HCV-IRES transcription vector. We are thankful to Prof. Kathrin Lang for the identification of an error in Figure 4 in the pre-print version of this manuscript. We are thankful to Bob Grassucci and Zhening Zhang for assistance in cryo-EM data acquisition. This work was supported by the NIH National Institute of General Medical Sciences (GM097014 to AVP). Part of this work was performed at the Simons Electron Microscopy Center and National Resource for Automated Molecular Microscopy located at the New York Structural Biology Center, supported by grants from the Simons Foundation (SF349247), NYSTAR, and the NIH National Institute of General Medical Sciences (GM103310).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Andrey V Pisarev, Email: andrey.pisarev@downstate.edu.

Israel S Fernández, Email: isf2106@cumc.columbia.edu.

Adam Frost, University of California, San Francisco, United States.

John Kuriyan, University of California, Berkeley, United States.

Funding Information

This paper was supported by the following grants:

  • Columbia University Start package to Israel S Fernández.

  • National Institute of General Medical Sciences GM097014 to Andrey V Pisarev.

Additional information

Competing interests

No competing interests declared.

Author contributions

Software, Validation, Investigation, Visualization, Writing - review and editing.

Investigation, Writing - review and editing.

Investigation, carried out experiments.

Conceptualization, Investigation, Writing - original draft, Writing - review and editing.

Conceptualization, Resources, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Project administration.

Additional files

Transparent reporting form

Data availability

Atomic coordinates have been deposited in the PDB with accession numbers and 6W2S and 6W2T for the open and closed classes , respectively . CryoEM maps have been deposited at the EMDB with accession numbers EMDB 21529 and 21530 for the open and closed classes respectively.

The following datasets were generated:

Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map open class. Electron Microscopy Data Bank. EMD-21529

Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map closed class. Electron Microscopy Data Bank. EMD-21530

Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the open state (Class 1) RCSB Protein Data Bank. 6W2S

Neupane R, Pisareva V, Rodriguez CF, Pisarev A, Fernandez IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the closed state (Class 2) RCSB Protein Data Bank. 6W2T

References

  1. Afonine PV, Poon BK, Read RJ, Sobolev OV, Terwilliger TC, Urzhumtsev A, Adams PD. Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallographica Section D Structural Biology. 2018;74:531–544. doi: 10.1107/S2059798318006551. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Aitken CE, Lorsch JR. A mechanistic overview of translation initiation in eukaryotes. Nature Structural & Molecular Biology. 2012;19:568–576. doi: 10.1038/nsmb.2303. [DOI] [PubMed] [Google Scholar]
  3. Amunts A, Brown A, Bai XC, Llácer JL, Hussain T, Emsley P, Long F, Murshudov G, Scheres SHW, Ramakrishnan V. Structure of the yeast mitochondrial large ribosomal subunit. Science. 2014;343:1485–1489. doi: 10.1126/science.1249410. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Andreev DE, O'Connor PB, Fahey C, Kenny EM, Terenin IM, Dmitriev SE, Cormican P, Morris DW, Shatsky IN, Baranov PV. Translation of 5' leaders is pervasive in genes resistant to eIF2 repression. eLife. 2015;4:e03971. doi: 10.7554/eLife.03971. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Archer SK, Shirokikh NE, Beilharz TH, Preiss T. Dynamics of ribosome scanning and recycling revealed by translation complex profiling. Nature. 2016;535:570–574. doi: 10.1038/nature18647. [DOI] [PubMed] [Google Scholar]
  6. Brar GA, Weissman JS. Ribosome profiling reveals the what, when, where and how of protein synthesis. Nature Reviews Molecular Cell Biology. 2015;16:651–664. doi: 10.1038/nrm4069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brown A, Long F, Nicholls RA, Toots J, Emsley P, Murshudov G. Tools for macromolecular model building and refinement into electron cryo-microscopy reconstructions. Acta Crystallographica Section D Biological Crystallography. 2015;71:136–153. doi: 10.1107/S1399004714021683. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Carragher B, Kisseberth N, Kriegman D, Milligan RA, Potter CS, Pulokas J, Reilein A. Leginon: an automated system for acquisition of images from vitreous ice specimens. Journal of Structural Biology. 2000;132:33–45. doi: 10.1006/jsbi.2000.4314. [DOI] [PubMed] [Google Scholar]
  9. Dolja VV, Koonin EV. Metagenomics reshapes the concepts of RNA virus evolution by revealing extensive horizontal virus transfer. Virus Research. 2018;244:36–52. doi: 10.1016/j.virusres.2017.10.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Eliseev B, Yeramala L, Leitner A, Karuppasamy M, Raimondeau E, Huard K, Alkalaeva E, Aebersold R, Schaffitzel C. Structure of a human cap-dependent 48S translation pre-initiation complex. Nucleic Acids Research. 2018;46:2678–2689. doi: 10.1093/nar/gky054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of coot. Acta Crystallographica. Section D, Biological Crystallography. 2010;66:486–501. doi: 10.1107/S0907444910007493. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Filbin ME, Kieft JS. Toward a structural understanding of IRES RNA function. Current Opinion in Structural Biology. 2009;19:267–276. doi: 10.1016/j.sbi.2009.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Flis J, Holm M, Rundlet EJ, Loerke J, Hilal T, Dabrowski M, Bürger J, Mielke T, Blanchard SC, Spahn CMT, Budkevich TV. tRNA translocation by the eukaryotic 80S ribosome and the impact of GTP hydrolysis. Cell Reports. 2018;25:2676–2688. doi: 10.1016/j.celrep.2018.11.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Frank J, Zhu J, Penczek P, Li Y, Srivastava S, Verschoor A, Radermacher M, Grassucci R, Lata RK, Agrawal RK. A model of protein synthesis based on cryo-electron microscopy of the E. coli ribosome. Nature. 1995;376:441–444. doi: 10.1038/376441a0. [DOI] [PubMed] [Google Scholar]
  15. Garrey JL, Lee YY, Au HH, Bushell M, Jan E. Host and viral translational mechanisms during cricket paralysis virus infection. Journal of Virology. 2010;84:1124–1138. doi: 10.1128/JVI.02006-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Greninger AL. A decade of RNA virus metagenomics is (not) enough. Virus Research. 2018;244:218–229. doi: 10.1016/j.virusres.2017.10.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gross L, Vicens Q, Einhorn E, Noireterre A, Schaeffer L, Kuhn L, Imler J-L, Eriani G, Meignin C, Martin F. The IRES5′UTR of the dicistrovirus cricket paralysis virus is a type III IRES containing an essential pseudoknot structure. Nucleic Acids Research. 2017;45:8993–9004. doi: 10.1093/nar/gkx622. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gunišová S, Hronová V, Mohammad MP, Hinnebusch AG, Valášek LS. Please do not recycle! translation reinitiation in microbes and higher eukaryotes. FEMS Microbiology Reviews. 2018;42:165–192. doi: 10.1093/femsre/fux059. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hashem Y, des Georges A, Dhote V, Langlois R, Liao HY, Grassucci RA, Hellen CU, Pestova TV, Frank J. Structure of the mammalian ribosomal 43S preinitiation complex bound to the scanning factor DHX29. Cell. 2013a;153:1108–1119. doi: 10.1016/j.cell.2013.04.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hashem Y, des Georges A, Dhote V, Langlois R, Liao HY, Grassucci RA, Pestova TV, Hellen CU, Frank J. Hepatitis-C-virus-like internal ribosome entry sites displace eIF3 to gain access to the 40S subunit. Nature. 2013b;503:539–543. doi: 10.1038/nature12658. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hashem Y, Frank J. The jigsaw puzzle of mRNA translation initiation in eukaryotes: a decade of structures unraveling the mechanics of the process. Annual Review of Biophysics. 2018:125–151. doi: 10.1146/annurev-biophys-070816-034034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hertz MI, Thompson SR. Mechanism of translation initiation by Dicistroviridae IGR IRESs. Virology. 2011;411:355–361. doi: 10.1016/j.virol.2011.01.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hinnebusch AG. Gene-specific translational control of the yeast GCN4 gene by phosphorylation of eukaryotic initiation factor 2. Molecular Microbiology. 1993;10:215–223. doi: 10.1111/j.1365-2958.1993.tb01947.x. [DOI] [PubMed] [Google Scholar]
  24. Hinnebusch AG. The scanning mechanism of eukaryotic translation initiation. Annual Review of Biochemistry. 2014;83:779–812. doi: 10.1146/annurev-biochem-060713-035802. [DOI] [PubMed] [Google Scholar]
  25. Hinnebusch AG. Structural insights into the mechanism of scanning and start Codon recognition in eukaryotic translation initiation. Trends in Biochemical Sciences. 2017;42:589–611. doi: 10.1016/j.tibs.2017.03.004. [DOI] [PubMed] [Google Scholar]
  26. Hussain T, Llácer JL, Fernández IS, Munoz A, Martin-Marcos P, Savva CG, Lorsch JR, Hinnebusch AG, Ramakrishnan V. Structural changes enable start Codon recognition by the eukaryotic translation initiation complex. Cell. 2014;159:597–607. doi: 10.1016/j.cell.2014.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ingolia NT, Ghaemmaghami S, Newman JR, Weissman JS. Genome-wide analysis in vivo of translation with nucleotide resolution using ribosome profiling. Science. 2009;324:218–223. doi: 10.1126/science.1168978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Jaafar ZA, Kieft JS. Viral RNA structure-based strategies to manipulate translation. Nature Reviews Microbiology. 2019;17:110–123. doi: 10.1038/s41579-018-0117-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Jackson RJ, Hellen CU, Pestova TV. The mechanism of eukaryotic translation initiation and principles of its regulation. Nature Reviews Molecular Cell Biology. 2010;11:113–127. doi: 10.1038/nrm2838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Jan E, Kinzy TG, Sarnow P. Divergent tRNA-like element supports initiation, elongation, and termination of protein biosynthesis. PNAS. 2003;100:15410–15415. doi: 10.1073/pnas.2535183100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Jan E, Mohr I, Walsh D. A Cap-to-Tail guide to mRNA translation strategies in Virus-Infected cells. Annual Review of Virology. 2016;3:283–307. doi: 10.1146/annurev-virology-100114-055014. [DOI] [PubMed] [Google Scholar]
  32. Johnson AG, Grosely R, Petrov AN, Puglisi JD. Dynamics of IRES-mediated translation. Philosophical Transactions of the Royal Society B: Biological Sciences. 2017;372:20160177. doi: 10.1098/rstb.2016.0177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kenner LR, Anand AA, Nguyen HC, Myasnikov AG, Klose CJ, McGeever LA, Tsai JC, Miller-Vedam LE, Walter P, Frost A. eIF2B-catalyzed nucleotide exchange and phosphoregulation by the integrated stress response. Science. 2019;364:491–495. doi: 10.1126/science.aaw2922. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kieft JS. Comparing the three-dimensional structures of Dicistroviridae IGR IRES RNAs with other viral RNA structures. Virus Research. 2009;139:148–156. doi: 10.1016/j.virusres.2008.07.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kolupaeva VG, de Breyne S, Pestova TV, Hellen CU. In vitro reconstitution and biochemical characterization of translation initiation by internal ribosomal entry. Methods in Enzymology. 2007;430:409–439. doi: 10.1016/S0076-6879(07)30016-5. [DOI] [PubMed] [Google Scholar]
  36. Kucukelbir A, Sigworth FJ, Tagare HD. Quantifying the local resolution of cryo-EM density maps. Nature Methods. 2014;11:63–65. doi: 10.1038/nmeth.2727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Lander GC, Stagg SM, Voss NR, Cheng A, Fellmann D, Pulokas J, Yoshioka C, Irving C, Mulder A, Lau PW, Lyumkis D, Potter CS, Carragher B. Appion: an integrated, database-driven pipeline to facilitate EM image processing. Journal of Structural Biology. 2009;166:95–102. doi: 10.1016/j.jsb.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lee JH, Pestova TV, Shin BS, Cao C, Choi SK, Dever TE. Initiation factor eIF5B catalyzes second GTP-dependent step in eukaryotic translation initiation. PNAS. 2002;99:16689–16694. doi: 10.1073/pnas.262569399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Lee AS, Kranzusch PJ, Doudna JA, Cate JH. eIF3d is an mRNA cap-binding protein that is required for specialized translation initiation. Nature. 2016;536:96–99. doi: 10.1038/nature18954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Leontis NB, Westhof E. Conserved geometrical base-pairing patterns in RNA. Quarterly Reviews of Biophysics. 1998;31:399–455. doi: 10.1017/S0033583599003479. [DOI] [PubMed] [Google Scholar]
  41. Majzoub K, Hafirassou ML, Meignin C, Goto A, Marzi S, Fedorova A, Verdier Y, Vinh J, Hoffmann JA, Martin F, Baumert TF, Schuster C, Imler JL. RACK1 controls IRES-mediated translation of viruses. Cell. 2014;159:1086–1095. doi: 10.1016/j.cell.2014.10.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallographica Section D Biological Crystallography. 1997;53:240–255. doi: 10.1107/S0907444996012255. [DOI] [PubMed] [Google Scholar]
  43. Myasnikov AG, Simonetti A, Marzi S, Klaholz BP. Structure-function insights into prokaryotic and eukaryotic translation initiation. Current Opinion in Structural Biology. 2009;19:300–309. doi: 10.1016/j.sbi.2009.04.010. [DOI] [PubMed] [Google Scholar]
  44. Nakashima N, Uchiumi T. Functional analysis of structural motifs in dicistroviruses. Virus Research. 2009;139:137–147. doi: 10.1016/j.virusres.2008.06.006. [DOI] [PubMed] [Google Scholar]
  45. Pestova TV, Lomakin IB, Lee JH, Choi SK, Dever TE, Hellen CU. The joining of ribosomal subunits in eukaryotes requires eIF5B. Nature. 2000;403:332–335. doi: 10.1038/35002118. [DOI] [PubMed] [Google Scholar]
  46. Pestova TV, de Breyne S, Pisarev AV, Abaeva IS, Hellen CU. eIF2-dependent and eIF2-independent modes of initiation on the CSFV IRES: a common role of domain II. The EMBO Journal. 2008;27:1060–1072. doi: 10.1038/emboj.2008.49. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Pestova TV, Hellen CU. Reconstitution of eukaryotic translation elongation in vitro following initiation by internal ribosomal entry. Methods. 2005;36:261–269. doi: 10.1016/j.ymeth.2005.04.004. [DOI] [PubMed] [Google Scholar]
  48. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. UCSF chimera--a visualization system for exploratory research and analysis. Journal of Computational Chemistry. 2004;25:1605–1612. doi: 10.1002/jcc.20084. [DOI] [PubMed] [Google Scholar]
  49. Pisarev AV, Skabkin MA, Pisareva VP, Skabkina OV, Rakotondrafara AM, Hentze MW, Hellen CU, Pestova TV. The role of ABCE1 in eukaryotic posttermination ribosomal recycling. Molecular Cell. 2010;37:196–210. doi: 10.1016/j.molcel.2009.12.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Pisareva VP, Pisarev AV, Fernández IS. Dual tRNA mimicry in the cricket paralysis virus IRES uncovers an unexpected similarity with the hepatitis C virus IRES. eLife. 2018;7:e34062. doi: 10.7554/eLife.34062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Quade N, Boehringer D, Leibundgut M, van den Heuvel J, Ban N. Cryo-EM structure of hepatitis C virus IRES bound to the human ribosome at 3.9-Å resolution. Nature Communications. 2015;6:7646. doi: 10.1038/ncomms8646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Ratje AH, Loerke J, Mikolajka A, Brünner M, Hildebrand PW, Starosta AL, Dönhöfer A, Connell SR, Fucini P, Mielke T, Whitford PC, Onuchic JN, Yu Y, Sanbonmatsu KY, Hartmann RK, Penczek PA, Wilson DN, Spahn CM. Head swivel on the ribosome facilitates translocation by means of intra-subunit tRNA hybrid sites. Nature. 2010;468:713–716. doi: 10.1038/nature09547. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Resch AM, Ogurtsov AY, Rogozin IB, Shabalina SA, Koonin EV. Evolution of alternative and constitutive regions of mammalian 5'UTRs. BMC Genomics. 2009;10:162. doi: 10.1186/1471-2164-10-162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Russo CJ, Passmore LA. Electron microscopy: ultrastable gold substrates for electron cryomicroscopy. Science. 2014;346:1377–1380. doi: 10.1126/science.1259530. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Scheres SH. RELION: implementation of a bayesian approach to cryo-EM structure determination. Journal of Structural Biology. 2012;180:519–530. doi: 10.1016/j.jsb.2012.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Scheres SH. Processing of structurally heterogeneous Cryo-EM data in RELION. Methods in Enzymology. 2016;579:125–157. doi: 10.1016/bs.mie.2016.04.012. [DOI] [PubMed] [Google Scholar]
  57. Schmeing TM, Ramakrishnan V. What recent ribosome structures have revealed about the mechanism of translation. Nature. 2009;461:1234–1242. doi: 10.1038/nature08403. [DOI] [PubMed] [Google Scholar]
  58. Sendoel A, Dunn JG, Rodriguez EH, Naik S, Gomez NC, Hurwitz B, Levorse J, Dill BD, Schramek D, Molina H, Weissman JS, Fuchs E. Translation from unconventional 5' start sites drives tumour initiation. Nature. 2017;541:494–499. doi: 10.1038/nature21036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Shatsky IN, Dmitriev SE, Terenin IM, Andreev DE. Cap- and IRES-independent scanning mechanism of translation initiation as an alternative to the concept of cellular IRESs. Molecules and Cells. 2010;30:285–293. doi: 10.1007/s10059-010-0149-1. [DOI] [PubMed] [Google Scholar]
  60. Shi M, Lin XD, Tian JH, Chen LJ, Chen X, Li CX, Qin XC, Li J, Cao JP, Eden JS, Buchmann J, Wang W, Xu J, Holmes EC, Zhang YZ. Redefining the invertebrate RNA virosphere. Nature. 2016;540:539–543. doi: 10.1038/nature20167. [DOI] [PubMed] [Google Scholar]
  61. Skabkin MA, Skabkina OV, Hellen CU, Pestova TV. Reinitiation and other unconventional posttermination events during eukaryotic translation. Molecular Cell. 2013;51:249–264. doi: 10.1016/j.molcel.2013.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Starck SR, Tsai JC, Chen K, Shodiya M, Wang L, Yahiro K, Martins-Green M, Shastri N, Walter P. Translation from the 5' untranslated region shapes the integrated stress response. Science. 2016;351:aad3867. doi: 10.1126/science.aad3867. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Terenin IM, Dmitriev SE, Andreev DE, Shatsky IN. Eukaryotic translation initiation machinery can operate in a bacterial-like mode without eIF2. Nature Structural & Molecular Biology. 2008;15:836–841. doi: 10.1038/nsmb.1445. [DOI] [PubMed] [Google Scholar]
  64. Vattem KM, Wek RC. Reinitiation involving upstream ORFs regulates ATF4 mRNA translation in mammalian cells. PNAS. 2004;101:11269–11274. doi: 10.1073/pnas.0400541101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Voorhees RM, Ramakrishnan V. Structural basis of the translational elongation cycle. Annual Review of Biochemistry. 2013;82:203–236. doi: 10.1146/annurev-biochem-113009-092313. [DOI] [PubMed] [Google Scholar]
  66. Wang J, Johnson AG, Lapointe CP, Choi J, Prabhakar A, Chen DH, Petrov AN, Puglisi JD. eIF5B gates the transition from translation initiation to elongation. Nature. 2019;573:605–608. doi: 10.1038/s41586-019-1561-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Wethmar K. The regulatory potential of upstream open reading frames in eukaryotic gene expression. Wiley Interdisciplinary Reviews: RNA. 2014;5:765–768. doi: 10.1002/wrna.1245. [DOI] [PubMed] [Google Scholar]
  68. Yamamoto H, Unbehaun A, Loerke J, Behrmann E, Collier M, Bürger J, Mielke T, Spahn CM. Structure of the mammalian 80S initiation complex with initiation factor 5B on HCV-IRES RNA. Nature Structural & Molecular Biology. 2014;21:721–727. doi: 10.1038/nsmb.2859. [DOI] [PubMed] [Google Scholar]
  69. Yamamoto H, Collier M, Loerke J, Ismer J, Schmidt A, Hilal T, Sprink T, Yamamoto K, Mielke T, Bürger J, Shaikh TR, Dabrowski M, Hildebrand PW, Scheerer P, Spahn CM. Molecular architecture of the ribosome-bound hepatitis C virus internal ribosomal entry site RNA. The EMBO Journal. 2015;34:3042–3058. doi: 10.15252/embj.201592469. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Yamamoto H, Unbehaun A, Spahn CMT. Ribosomal chamber music: toward an understanding of IRES mechanisms. Trends in Biochemical Sciences. 2017;42:655–668. doi: 10.1016/j.tibs.2017.06.002. [DOI] [PubMed] [Google Scholar]
  71. Yokoyama T, Machida K, Iwasaki W, Shigeta T, Nishimoto M, Takahashi M, Sakamoto A, Yonemochi M, Harada Y, Shigematsu H, Shirouzu M, Tadakuma H, Imataka H, Ito T. HCV IRES captures an actively translating 80S ribosome. Molecular Cell. 2019;74:1205–1214. doi: 10.1016/j.molcel.2019.04.022. [DOI] [PubMed] [Google Scholar]
  72. Young SK, Wek RC. Upstream open reading frames differentially regulate Gene-specific translation in the integrated stress response. Journal of Biological Chemistry. 2016;291:16927–16935. doi: 10.1074/jbc.R116.733899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Zhang K. Gctf: real-time CTF determination and correction. Journal of Structural Biology. 2016;193:1–12. doi: 10.1016/j.jsb.2015.11.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Zhang YZ, Wu WC, Shi M, Holmes EC. The diversity, evolution and origins of vertebrate RNA viruses. Current Opinion in Virology. 2018;31:9–16. doi: 10.1016/j.coviro.2018.07.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Zhang Y-Z, Chen Y-M, Wang W, Qin X-C, Holmes EC. Expanding the RNA Virosphere by Unbiased Metagenomics. Annual Review of Virology. 2019;6:119–139. doi: 10.1146/annurev-virology-092818-015851. [DOI] [PubMed] [Google Scholar]
  76. Zheng SQ, Palovcak E, Armache JP, Verba KA, Cheng Y, Agard DA. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nature Methods. 2017;14:331–332. doi: 10.1038/nmeth.4193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Zivanov J, Nakane T, Forsberg BO, Kimanius D, Hagen WJ, Lindahl E, Scheres SH. New tools for automated high-resolution cryo-EM structure determination in RELION-3. eLife. 2018;7:e42166. doi: 10.7554/eLife.42166. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: Adam Frost1
Reviewed by: Andrei A Korostelev2

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

We anticipate that your novel and exciting structures of a viral IRES will be of broad interest because they show how this virus binds initiation factors and wraps around the head of the 40S ribosome to position AUG codons in or near the P-site. Your structures resolved three domains of the IRES making specific contacts with the eIF3 complex and the 40S ribosome; of theses, domain II of the IRES, in particular, engages eIF3 and the 40S subunit RACK1 – explaining the atomic basis of this virus' dependence on RACK1.

Decision letter after peer review:

Thank you for submitting your article "A complex IRES at the 5'-UTR of a viral mRNA assembles a functional 48S complex via an uAUG intermediate" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by John Kuriyan as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: and Andrei A Korostelev (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision as an integration of everyone's feedback to help you prepare a thoroughly revised submission.

Summary:

Neupane and Pisareva et al. report novel and exciting structural insights into vRNA IRES structures from the cricket paralysis virus in different stages of hijacking mammalian ribosomes and initiation factors – and doing so in unexpected ways given prior publications (Gross et al., 2017). These authors utilize a biochemical reconstitution system based on rabbit reticulocyte lysates and purified components with multi-class cryo-EM to determine structures that suggest the CrPV 5'-UTR-IRES is more intricate than the simple type III classification proposed earlier. Specifically, these authors determined two IRES-engaged 48S complex structures in both an "open" state (Class-1) and second "closed" state (Class-2). Three domains of the IRES (Domain I, II and III) were resolved making specific contacts with the 40S and with the multi-subunit factor eIF3. Of particular interest, the authors reveal domain II of the IRES engages eIF3 as well as critical contacts between the IRES and RACK1-explaining the atomic basis of this IRES's dependence on RACK1. They also discuss how this IRES from CrPV requires eIF1/eIF1A and ternary complex-mediated Met-tRNAi delivery to transition from an upstream AUG (uAUG) to a translation-ready assembly on the start codon-including a plausible reaction coordinate for 40S conformational remodeling, eIF3(d) recruitment, and TC-Met-tRNAi delivery.

The use of mammalian components to study a cricket virus, however, raises some questions and concerns. 5'UTR IRESes are not conserved but rather highly divergent and can be quite specific for each virus. This IRES in the 5' UTR may have play a role in defining, at least partially, host tropism. Ideally, an IRES should be studied in a biologically relevant context using the translational machinery and ribosomes from a natural host – this is not the case with rabbit extracts and this discrepancy should be addressed prior to publication because it could explain a potential disagreement between the structural models and the toe-printing data.

Essential revisions:

1) The complex was found to be assembled on an out-of-frame upstream AUG (in the P-site of the 40S, so the authors propose a novel IRES-mediated mechanism using an uAUG as an intermediate complex that is required to further assemble the ribosome on the genuine AUG start codon. In order to strengthen their model, the authors performed toe-printing assays that allowed the detection of a toe-print on the uAUG and another toe-print on the genuine AUG. However, the confusing presentation of this data suggests there may be a contradiction with their model that warrants further exploration. Specifically, in Figure 6A it is unclear if the toeprint corresponding to uAUG (lanes 2-5) reflects tRNA bound to uAUG? It is possible that this band corresponds to the 40S-IRES-eIF3 complex (featured in cryo-EM structures), while tRNA loading does not occur despite the presence of the ternary complex? A control lane (with 40S, IRES and eIF3, without TC), should help to test this point and clarify the mechanism. In the current figure, there appear to be two toe-prints at +17/18 from the uAUG and the genuine AUG, respectively. Are these the expected distances if the uAUG is located in the P-site of the ribosome? If the uAUG is located in the A-site, shouldn't the corresponding toe-print be detected at +13/14? Please clarify. Moreover, the appearance of the genuine AUG toe-print is not correlated with the disappearance of the uAUG toe-print, which seems to indicate that the two toe-prints are independent. The authors propose that the uAUG toe-print corresponds to an intermediate state prior genuine AUG recognition. According to the toe-printing data, this does not seem to be the case and suggest that the structures may not correspond with the functional data.

2) To our knowledge, there is are no evidence in the literature that the uAUG is indeed required for this IRES to function, so the authors should mutate the uAUG and measure whether (i) the IRES is no longer active and (ii) whether both toe-prints disappear.

3) A biochemical test for the proposed roles of the DIII non-canonical base pairs would also strengthen the manuscript. Do mutants in this region of the IRES alter the positioning of the single-stranded segment of the IRES harboring the uAUG-701, and thus alter the outcomes of the toe-printing assay?

4) These authors show the 5'UTR IRES interacting directly with eIF3. Since there are significant differences between rabbit eIF3 and eIF3 from insects, this could explain why eIF3 is positioned differently on the 40S when compared canonical 48S structures. Please address.

5) The FSC 0.5 cut-off for the map-to-model correspondence appears to be notably worse than the half map 0.143 cut-off. The curve also has a very sharp drop-off toward zero, suggesting that the analysis was done with a low-pass filtered map or model? Also, please add a color legend for the refined B-factors.

6) The structural models have high clash scores, many poor rotamers, and far too many disallowed Ramachandran angles. Is this evidence of overfitting to density or other forms inattention to model geometry? Are there regions of the model that should not have been modeled at this level of detail?

7) The authors wrote "Further assistance by eIF1 and eIF1A…". But are there data suggesting such a step-wise IRES-dependent sequence of events? What data suggest that binding of eEF1 and eEF1A is excluded prior to binding of eIF3, IRES and/or delivery of tRNA? Perhaps eIF1 and/or eIF1A can be present prior to IRES and tRNA binding? Please discuss.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for submitting your article "A complex IRES at the 5'-UTR of a viral mRNA assembles a functional 48S complex via an uAUG intermediate" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Adam Frost as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by John Kuriyan as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Andrei A Korostelev (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data. Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

Summary:

The novel and exciting structural insights into an IRES from the cricket paralysis virus reported in the revised study by Neupane and Pisareva et al. shows this viral RNA hijacking mammalian ribosomes and initiation factors-and doing so in unforeseen ways given prior publications (Gross et al., 2017). The authors utilized a biochemical reconstitution system based on rabbit reticulocyte lysates and purified components, with multi-class cryo-EM, to determine structures of an "open" and "closed" state of the 40S ribosome bound by the IRES and eIF3. Three domains of the IRES (Domain I, II and III) were resolved making specific contacts with the eIF3+40S assembly. Domain II of the IRES engages eIF3 and makes contacts between the IRES and RACK1-explaining the atomic basis of this IRES's dependence on RACK1.

Each reviewer was intrigued by these novel structures because they show the viral RNA wrapping around the head of the 40S ribosome to position AUG codons in or near the P-site. Each reviewer, however, also expressed concerns about the mechanistic interpretation of viral translation initiation suggested by these structures-especially the unclear role of the upstream start codon or uAUG. After extensive cross-review, the reviewers consolidated the following requests to address before acceptance.

Essential revisions:

1) Interpretation of the uAUG's role and the toe-printing data

1a) The use of rabbit ribosomes and initiation factors to study an insect virus raised questions and concerns during the first round of review that the authors only partially addressed. One primary concern was that the possibility that the upstream AUG, or uAUG, may not be required for IRES function. The authors did not include a translational control-like reporter expression downstream of the IRES-which would enable functional evaluation of the role of the uAUG. This omission leaves open the possibility that the structures are of non-functional complexes that form in the absence of the ternary complex. The newly added controls show that mutation of the uAUG in the presence of the ternary complex, eIF2GTP+tRNAMeti, still leads to a robust toe-print on the bonafide AUG codon. All of the initiation factors under investigation here are present in cells-including eIF1, eIF1A, eIF3, the ternary complex, and eIF5B. The uAUG of the IRES, therefore, may not be not required for initiation in vivo even though in this system a stable toe-print forms in vitro. The authors should rewrite the description of their model, Discussion, sixth paragraph, to acknowledge these unresolved caveats and state clearly that the mechanism of initiation remains incompletely understood. Fully understanding initiation in this unusual setting (the staged roles of each initiation factor acting on the uAUG start-stop, followed by a genuine AUG start) will require further functional studies. Given the mismatch between host and virus, moreover (mammal versus insect), it is not clear that additional mechanistic experiments in the rabbit reticulocyte lysate context are worthwhile. To help explain this latter point, please include the percent similarity/identity between the rabbit and insect factors.

1b) The author's description and annotation of the location of uAUG and DIII in their structures (subsection “Non-canonical base pairing in the 5'-UTR-IRES DIII places the uAUG codon near the P site”) remain confusing. Please label the putative position of the uAUG in the middle panel of Figure 4C, along with the numbering for DIII, so it is clear to the reader where residue 701 could be. Since the uAUG bases themselves are not resolved in the map, the authors could highlight and label the final residue resolved in the P site or use a dotted line to schematically represented unresolved IRES bases to denote their putative positions.

1c) While the authors show that this IRES does not associate with the 40S without eIF3, the Figure 6 model figure shows "40S capture" rather than 40S+eIF3 capture-please fix.

1d) Also, in Figure 6, the 40S orientation changes too drastically between steps. It appears that the IRES and eIF3 jump between binding sites. Please fix the view of the 40S subunit view to help visualize conformational differences between the proposed initiation steps?

2) Quality and reliability of the atomic models: Concerns about overfitting remain due to the map-to-model FSC curves, the number of poor rotamers, and disallowed Ramachandran angles. The authors make salient points about the challenges of model building in this resolution regime and with this degree of heterogeneity, but they could be more conservative. Since their cryoEM density is not at sufficient resolution to justify clashes or bad geometry, we recommend "stumping" the residues/bases in question to the backbone to remove unsupported model features. In addition, the authors built their models using uniformly sharpened maps, despite the heterogeneous resolution. The models may improve, and their correspondence with the EM density may also improve, if the maps are first filtered and sharpened according to local resolution estimates. Locally filtered and sharpened maps may be better for modeling both high-resolution and low-resolution regions. Also, map-to-model FSC tests can certainly employ masks, but not low-pass filtered maps or models. Relion's default mode postprocessed maps are sharpened and low-pass filtered according to the global FSC. Please repeat map-model correspondence tests with unsharpened, unfiltered maps. There are many ways of generating such plots. Phenix.real_space_refine tools, for example, employ the unfiltered and unsharpened half-maps during model refinement and map-to-model correspondence tests. Finally, the table should include protein and RNA statistics-RNA statistics are currently missing.

eLife. 2020 Apr 14;9:e54575. doi: 10.7554/eLife.54575.sa2

Author response

Summary:

Neupane and Pisareva et al. report novel and exciting structural insights into vRNA IRES structures from the cricket paralysis virus in different stages of hijacking mammalian ribosomes and initiation factors – and doing so in unexpected ways given prior publications (Gross et al., 2017). These authors utilize a biochemical reconstitution system based on rabbit reticulocyte lysates and purified components with multi-class cryo-EM to determine structures that suggest the CrPV 5'-UTR-IRES is more intricate than the simple type III classification proposed earlier. Specifically, these authors determined two IRES-engaged 48S complex structures in both an "open" state (Class-1) and second "closed" state (Class-2). Three domains of the IRES (Domain I, II and III) were resolved making specific contacts with the 40S and with the multi-subunit factor eIF3. Of particular interest, the authors reveal domain II of the IRES engages eIF3 as well as critical contacts between the IRES and RACK1-explaining the atomic basis of this IRES's dependence on RACK1. They also discuss how this IRES from CrPV requires eIF1/eIF1A and ternary complex-mediated Met-tRNAi delivery to transition from an upstream AUG (uAUG) to a translation-ready assembly on the start codon-including a plausible reaction coordinate for 40S conformational remodeling, eIF3(d) recruitment, and TC-Met-tRNAi delivery.

The use of mammalian components to study a cricket virus, however, raises some questions and concerns. 5'UTR IRESes are not conserved but rather highly divergent and can be quite specific for each virus. This IRES in the 5' UTR may have play a role in defining, at least partially, host tropism. Ideally, an IRES should be studied in a biologically relevant context using the translational machinery and ribosomes from a natural host – this is not the case with rabbit extracts and this discrepancy should be addressed prior to publication because it could explain a potential disagreement between the structural models and the toe-printing data.

We certainly understand this concern and from a purely biologically point of view, yes, we agree the best scenario would be to study an IRES in the context of ribosomes from their own host. However, one of the aspects that makes the biology of IRESs so interesting is their ability to operate in a wide range of hosts and indeed such capability has been exploited for biotechnological applications. Two prototypical, widely studied IRESs, the IRES of the intergenic region of Dicistroviruses as well as the EMCV IRES, can manipulate ribosomes from different hosts and this capability has been exploited for gene expression regulation in multiple applications. With that in mind, we decided to explore the possibility the 5'-UTR-IRES of the CrPV would indeed function with mammalian components. Such strategy offers two advantages: on one side we will prove this IRES does not exclusively work in insects and second, and perhaps more importantly, we benefit from the solid knowledge both structural and biochemical, available for the mammalian system. There are dozens of high-resolution structures of mammalian ribosomes in several, biologically relevant configuration as well as a wealth of biochemistry and molecular biology data. All this knowledge is not available for the insect system.

Additionally, the ribosomal and eIF3 components the 5'-UTR-IRES interact with are highly conserved. Ribosomal protein uS17 or the core subunits of eIF3 are almost identical between insects and mammals, what argues against a "species" specific IRES. The final argument to prove the validity of our approach is the fact that toe-print experiments with mammalian components showed a very robust and selective delivery of initiator tRNA to 40S/5'-UTR-IRES/eIF3 complexes. This very specific and precise experiment will never work if the 5'-UTR-IRES do not manipulate mammalian 40S and eIF3 in a very specific way.

Essential revisions:

1) The complex was found to be assembled on an out-of-frame upstream AUG (in the P-site of the 40S, so the authors propose a novel IRES-mediated mechanism using an uAUG as an intermediate complex that is required to further assemble the ribosome on the genuine AUG start codon. In order to strengthen their model, the authors performed toe-printing assays that allowed the detection of a toe-print on the uAUG and another toe-print on the genuine AUG. However, the confusing presentation of this data suggests there may be a contradiction with their model that warrants further exploration. Specifically, in Figure 6A it is unclear if the toeprint corresponding to uAUG (lanes 2-5) reflects tRNA bound to uAUG? It is possible that this band corresponds to the 40S-IRES-eIF3 complex (featured in cryo-EM structures), while tRNA loading does not occur despite the presence of the ternary complex? A control lane (with 40S, IRES and eIF3, without TC), should help to test this point and clarify the mechanism. In the current figure, there appear to be two toe-prints at +17/18 from the uAUG and the genuine AUG, respectively. Are these the expected distances if the uAUG is located in the P-site of the ribosome? If the uAUG is located in the A-site, shouldn't the corresponding toe-print be detected at +13/14? Please clarify. Moreover, the appearance of the genuine AUG toe-print is not correlated with the disappearance of the uAUG toe-print, which seems to indicate that the two toe-prints are independent. The authors propose that the uAUG toe-print corresponds to an intermediate state prior genuine AUG recognition. According to the toe-printing data, this does not seem to be the case and suggest that the structures may not correspond with the functional data.

We appreciate the reviewer's concern about the explanation of these results. Toe-printing experiments are difficult to understand, and a better explanation is needed here. We completely rewrote the section where these experiments are described, citing Skabkin et al.paper. We hope the experiments, which show robust, unambiguous results, could be interpreted better now, especially by the reader not familiarized with this kind of data.

Toe-print signals can only be identified if the ribosome is solidly anchored to the mRNA. This will stop the reverse polymerase. No toe-print signal around any of the AUG codons could be observed for the 5'-UTR-IRES alone, 40S/5'-UTR-IRES complex or the 40S/5'-UTR-IRES/eIF3 complex which was characterized structurally by cryoEM. These controls were very clear and are now included in a supplementary figure (Figure 6—figure supplement 1). Second, there are two clear toe-print signals arising from the delivery of initiator Met-tRNAiMet to the P site of the ribosome. Both eIF2 and eIF5B can only deliver Met-tRNAiMet to the P site of the 40S, where they establish specific interaction and are able to stabilize the Met-tRNAiMet in a solid conformation able to generate a toe-print. If this tRNA/mRNA/ribosome interaction is not solid, the reverse polymerase is able to disassemble the weak complex, keeping elongating and no toe-print signal could be observed. That is the case for Figure 6A lane 1: Met-tRNAiMet by itself is unable to find the right ribosomal context so it can't establish interactions with the mRNA/40S stable enough to generate a toe-print around the AUGs. It is well known that P site toe-prints generate signals 15-17 nucleotides downstream due to that is the length of mRNA covered by the 40S. In our experiments, we see two toe-prints exactly 15/17 nucleotides away from the uAUG or the annotated AUG, what makes the interpretation of the results unambiguous. No Met-tRNAiMet has ever been described to be able to generate a toe-print at 13/14 due to it is pausing the ribosome in the A site.

Band intensity in toe-print experiments are not quantitative, they do not reflect a total amount of radioactivity present in the assay, specially, when we compare experiments with different components like lanes 2-5 in Figure 6A. The main point we want to clarify with these toe-print experiments is how the 5'-UTR-IRES is able to direct by itself Met-tRNAiMet to the P site of the 40S. This delivery event is very specific and only takes place in very determined cellular conditions, normally associated with covalently modified messengers ("capped mRNAs") and the initiator factors of the eIF4 family, absent here. The fact that 5'-UTR-IRES, in conjunction only with eIF2 or eIF5B can generate, solid, robust toe-print signals on AUGs is remarkable, and can only happen due to the ability of this RNA sequence to specifically manipulate 40S and initiation factors. Importantly, this IRES operate in a cellular environment where general translation has been shut off due to phosphorylation of eIF2alpha. In this context, eIF5B can substitute eIF2 in Met-tRNAiMet delivery, however, how can that happen is not fully understood. In our experiments, we can only see delivery of Met-tRNAiMet by eIF5B to the uAUG (Figure 6A lane 4). This delivery is less efficient when eIF1/eIF1A are present (Figure 6A lane 5). eIF2 also delivers Met-tRNAiMet to the uAUG in isolation but, in contrast to eIF5B, in the presence of eIF1/eIF1A it can find both, the uAUG and the annotated AUG, both with more efficiency as judged by the toe-print intensity than the recognition of uAUG in isolation. This represent only part of the truth as all these factors act in vivo in conjunction and with others. We believe these results indicate a dynamic interplay between different initiation factors able to deliver the initiator tRNA to different but physically close AUG codons, and such dynamisms could be essential for an efficient initiation in a cellular context where cellular and viral messages compete for ribosomal access.

We do feel that the fact that two different initiation factors can deliver Met-tRNAiMet to this mRNA is remarkable but clearly, more experiments are needed to fully understand this mechanism.

2) To our knowledge, there is are no evidence in the literature that the uAUG is indeed required for this IRES to function, so the authors should mutate the uAUG and measure whether (i) the IRES is no longer active and (ii) whether both toe-prints disappear.

As we discussed above, there are several factors to have in mind regarding the importance of the uAUG and start-stop configuration that precedes the annotated AUG. We indeed mutate this codon to ACG (new supplementary Figure 6—figure supplement 1B) and, as expected, the delivery of Met-tRNAiMet to uAUG by eIF2 and eIF1/eIF1A is abolished, while the delivery to the annotated AUG is maintained (Figure 6—figure supplement 1B, right lane). However, we are skeptical regarding the importance of this. As noted above, in a cellular environment where eIF2 is not available due to phosphorylation of its alpha subunit, only eIF5B has been reported to substitute eIF2 in Met-tRNAiMet delivery. Thus, in a biologically relevant situation this mutation will probably abolish initiation mediated by the 5'-UTR-IRES.

3) A biochemical test for the proposed roles of the DIII non-canonical base pairs would also strengthen the manuscript. Do mutants in this region of the IRES alter the positioning of the single-stranded segment of the IRES harboring the uAUG-701, and thus alter the outcomes of the toe-printing assay?

Unfortunately, we are unable to generate these mutants and test them by toe-print in the time frame expected for the return of this reviewed version. We strongly believe this mutation will be deleterious for the activity of the IRES as they would radically change the local architecture of the IRES at a critical region.

4) These authors show the 5'UTR IRES interacting directly with eIF3. Since there are significant differences between rabbit eIF3 and eIF3 from insects, this could explain why eIF3 is positioned differently on the 40S when compared canonical 48S structures. Please address.

In our maps and derived models, eIF3 exhibit a positioning respect the 40S very similar to the position described for the canonical 48S complex (Figure 3D). The 5'-UTR-IRES seems to exploit cavities present in the 48S canonical complex to interact with the 40S-facing area of eIF3. In any case, there are two arguments that convince us the interaction of the 5'-UTR-IRES with the mammalian versus insect eIF3 should be similar: on one side, the interaction of DII of 5'-UTR-IRES with the back of the 40S and eIF3 is peripherical, with very limited contacts between DII and eIF3, mediated by contacts with the phosphate backbone of the IRES and not with specific bases. On the other hand, and perhaps more importantly, eIF3 core subunits a and c (that form the area hosting the bulk of interactions with 5'-UTR-IRES DII) are highly conserved, as it is ribosomal protein uS17 at the back of the 40S.

5) The FSC 0.5 cut-off for the map-to-model correspondence appears to be notably worse than the half map 0.143 cut-off. The curve also has a very sharp drop-off toward zero, suggesting that the analysis was done with a low-pass filtered map or model? Also, please add a color legend for the refined B-factors.

We understand the concern regarding the discrepancy between the model-vs.-map FSC 0.5 cut-off and half-map 0.143 cut-off as they are indeed not as close as they should be for this resolution. The explanation is related to the wide heterogeneity of the map, even after classification. The map exhibits very good quality at the 40S body and some areas of the 40S head, some areas of the IRES and the regions of the eIF3 in contact with the 40S. Poor density could however be observed in the regions of eIF3 away from the 40S and some segments of the IRES exposed to the solvent. In such scenario, current protocols of model refinement in cryoEM maps are far from ideal. We monitored the weight applied to the geometry term in the final step of the model refinement (using REFMAC) in order to apply as less as possible provided the model was not overfitted. It turned out that due to the areas of the map with severe flexibility, we had to keep the geometry term quite low in order to avoid overfitting, in our opinion a more severe problem that a discrepancy between the model-vs.-map FSC 0.5 cut-off and half-map 0.143 cut-off.

In the model validation FSC (Figure 1—figure supplement 2B and F, black curve) it was used the final, post-processed masked, map which was the same one used for model refinement. The masking artifact explain the drastic fall to zero. Due to this is the final map used for model building and refinement, we feel this is the one that should be used for the FSC calculation.

We appreciate the suggestion of adding labels for the model B-factor scale. These labels have been added.

6) The structural models have high clash scores, many poor rotamers, and far too many disallowed Ramachandran angles. Is this evidence of overfitting to density or other forms inattention to model geometry? Are there regions of the model that should not have been modeled at this level of detail?

As described in the previous point, the heterogeneity of the map imposed strict limitations on the model refinement strategies. We opted for a conservative approach, considering the avoiding of overfitting of the model to the map the most important objective. We performed 10 different model refinement runs for the two maps, adjusting the geometry-versus-map-fit (matrix keyword in REFMAC) and subsequently calculating a model "shake" and half map 1 refinement overfitting test. The final matrix value chose was the highest one that exhibited no overfitting. As shown by the overlapping of the red and blue curves of Figure 1—figure supplement 2B and F, both models are not overfitted.

Regarding the Ramachandran outliers, we consider these values reasonable given the resolution and when compared with other ribosome models at similar resolution both by cryoEM or X-ray crystallography. REFMAC does not implement Ramachandran restrains in its reciprocal-space refinement protocol, due to that could lead to artificially low Ramachandran outliers percentage. We believe this approach is best suited for the current state of the field instead of trying to obtain a 0% Ramachandran outliers score, unrealistic given this resolution and map heterogeneity.

7) The authors wrote "Further assistance by eIF1 and eIF1A…". But are there data suggesting such a step-wise IRES-dependent sequence of events? What data suggest that binding of eEF1 and eEF1A is excluded prior to binding of eIF3, IRES and/or delivery of tRNA? Perhaps eIF1 and/or eIF1A can be present prior to IRES and tRNA binding? Please discuss.

We have indeed no evidence for the sequential recruitment of eIF1/eIF1A after TC recruitment. We have modified our proposed model to reflect this fact, including Figure 6.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Essential revisions:

1) Interpretation of the uAUG's role and the toe-printing data

1a) The use of rabbit ribosomes and initiation factors to study an insect virus raised questions and concerns during the first round of review that the authors only partially addressed. One primary concern was that the possibility that the upstream AUG, or uAUG, may not be required for IRES function. The authors did not include a translational control-like reporter expression downstream of the IRES-which would enable functional evaluation of the role of the uAUG. This omission leaves open the possibility that the structures are of non-functional complexes that form in the absence of the ternary complex.

We respectfully disagree with this concern. Every IRES need to first and foremost interact with the ribosomal subunits before any aminoacyl-tRNA is delivered. One of the most essential features of any IRES is its ability to capture ribosomal subunits in isolation or in combination with initiation factors. The best characterized IRESs (the IGR-IRES of the CrPV and HCV IRES) are both able to interact with the 40S and/or 80S in the absence of the TC. This is so because it is compulsory for these RNAs to first, bind and hijack the ribosomal subunits and only afterwards, recruit a tRNA delivery system either from initiation (eIF2/eIF5B) or from elongation (eEF2/eEF2A). Thus, the sentence:

“This omission leaves open the possibility that the structures are of non-functional complexes that form in the absence of the ternary complex.” in our opinion, is highly misleading, especially for the non-specialized reader, as it suggests that the interaction of the IRES with the 40S in the absence of the ternary complex could be interpreted as an “artifact” when in fact is an essential, biologically relevant step in IRES biology.

The newly added controls show that mutation of the uAUG in the presence of the ternary complex, eIF2GTP+tRNAMeti, still leads to a robust toe-print on the bonafide AUG codon. All of the initiation factors under investigation here are present in cells-including eIF1, eIF1A, eIF3, the ternary complex, and eIF5B. The uAUG of the IRES, therefore, may not be not required for initiation in vivo even though in this system a stable toe-print forms in vitro.

We consider this statement as an oversimplification given two facts: first, we disagree with the conclusion that due to the mutation of uAUG does not abolish the delivery of Met-tRNAiMet to the bona-fide AUG codon by eIF2, uAUG is not required for initiation in vivo. In physiological conditions, if a cell is infected by a virus, it is well established that the availability of eIF2 is highly diminished. Phosphorylation of eIF2-alpha, caused by the antiviral response of the cell, blocks the delivery of Met-tRNAiMet by eIF2. Then, how viral proteins initiate under these conditions? The most probable candidate to substitute eIF2 in Met-tRNAiMet delivery is eIF5B and eIF5B, as our toe-prints clearly show, ONLY delivers Met-tRNAiMet to the uAUG. Secondly, the uAUG is part of a start-stop configuration which means that perhaps re-initiation events mediated by release factors could be critical for sustained levels of translation initiation in infected cell. We agree with reviewers that more experiments, probably in vivo, are required to understand how this peculiar initiation happens, but we believe that ruling out a significant role for the uAUG, just because its mutation does not abolish the recognition of the bone-fide AUG by eIF2, seems premature.

The authors should rewrite the description of their model, Discussion, sixth paragraph, to acknowledge these unresolved caveats and state clearly that the mechanism of initiation remains incompletely understood. Fully understanding initiation in this unusual setting (the staged roles of each initiation factor acting on the uAUG start-stop, followed by a genuine AUG start) will require further functional studies.

We totally agree with this conclusion and we hope this study will inspire experiments directed towards the understanding of the role of uAUGs, eIF5B and the start-stop configuration in cellular and viral messengers. In order to reflect this, we have modified the text with the following sentence:

“However, these considerations are highly speculative, as the particular role the uAUG exerts in Met-tRNAiMet recruitment or more generally, its involvement in initiation of viral messengers remains enigmatic. A comprehensive understanding of the role of uAUG and the start-stop configuration will demand further studies, ideally in vivo.”

Given the mismatch between host and virus, moreover (mammal versus insect), it is not clear that additional mechanistic experiments in the rabbit reticulocyte lysate context are worthwhile. To help explain this latter point, please include the percent similarity/identity between the rabbit and insect factors.

identity/similarity (%)
(Drosophila/mammals)
eIF2
alpha: 57/74
beta: 74/83
gamma: 82/88
eIF5B: 71/85
eIF3a: 46/63
eIF3c: 51/66
Open in a new tab

We have supplemented the manuscript with this comparative analysis.

1b) The author's description and annotation of the location of uAUG and DIII in their structures (subsection “Non-canonical base pairing in the 5'-UTR-IRES DIII places the uAUG codon near the P site”) remain confusing. Please label the putative position of the uAUG in the middle panel of Figure 4C, along with the numbering for DIII, so it is clear to the reader where residue 701 could be. Since the uAUG bases themselves are not resolved in the map, the authors could highlight and label the final residue resolved in the P site or use a dotted line to schematically represented unresolved IRES bases to denote their putative positions.

We followed the reviewer’s advice in modifying Figure 4C to include labels for the last residue of the IRES seen in density (C695) as well as, using as guide the mRNA trajectory from known structures, we now indicate with a dotted line the putative placement of the segment of viral mRNA harboring the uAUG.

1c) While the authors show that this IRES does not associate with the 40S without eIF3, the Figure 6 model figure shows "40S capture" rather than 40S+eIF3 capture-please fix.

We have eliminated this step from the figure and modified the orientation of the complexes to facilitate its viewing. We have also added labels for the IRES domains to allow a better perception of the transitions.

1d) Also, in Figure 6, the 40S orientation changes too drastically between steps. It appears that the IRES and eIF3 jump between binding sites. Please fix the view of the 40S subunit view to help visualize conformational differences between the proposed initiation steps?

These modifications have been introduced. See above.

2) Quality and reliability of the atomic models: Concerns about overfitting remain due to the map-to-model FSC curves, the number of poor rotamers, and disallowed Ramachandran angles. The authors make salient points about the challenges of model building in this resolution regime and with this degree of heterogeneity, but they could be more conservative. Since their cryoEM density is not at sufficient resolution to justify clashes or bad geometry, we recommend "stumping" the residues/bases in question to the backbone to remove unsupported model features. In addition, the authors built their models using uniformly sharpened maps, despite the heterogeneous resolution. The models may improve, and their correspondence with the EM density may also improve, if the maps are first filtered and sharpened according to local resolution estimates. Locally filtered and sharpened maps may be better for modeling both high-resolution and low-resolution regions.

We greatly value reviewers' input in this area. After, careful reading of previous concerns we have realized there were some mistakes in the values reported in the table summarizing cryoEM data. We have updated these values which are now correct, reflecting better statistics.

We started the model building and initial refinement in real space with locally refined maps using masks for the 40S-head-IRES-DI and 40S-head-IRES-DII as the reviewer suggests. These masked maps, as shown in Figure 1—figure supplement 2, locally improved the density allowing a more uniform post-processing. These local masked maps are now uploaded in the EMDB deposition.

However, we believe that the “real” final model should be refined against the final, whole map, especially in the reciprocal-space step when we use Refmac. This step allows for a realistic computation of B-factors in reciprocal space which, as shown in Figure 1—figure supplement 2, reflect accurately the variability of the maps seen by local resolution calculations. We believe this information is valuable for the specialized reader, who can quickly from these diagrams identify which regions of the maps/model are more trustable and which ones should be considered cautiously.

Our models are clearly not over-fitted. We performed a rigorous over-fitting test by refining against half-map-1 a “shake” model. The FSC curves between this re-refined model against half-map-1 and against half-map-2 (not included in the refinement) clearly overlap for both classes, what unambiguously demonstrates that our models are not over-fitted (over-fitting test described in: Amunts et al., 2014 and Brown, et al., 2015).

Also, map-to-model FSC tests can certainly employ masks, but not low-pass filtered maps or models. Relion's default mode postprocessed maps are sharpened and low-pass filtered according to the global FSC. Please repeat map-model correspondence tests with unsharpened, unfiltered maps. There are many ways of generating such plots. Phenix.real_space_refine tools, for example, employ the unfiltered and unsharpened half-maps during model refinement and map-to-model correspondence tests. Finally, the table should include protein and RNA statistics-RNA statistics are currently missing.

We have update the FSC for these graphs using non-filtered unsharpened maps in Figure 1—figure supplement 2.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map open class. Electron Microscopy Data Bank. EMD-21529
    2. Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map closed class. Electron Microscopy Data Bank. EMD-21530
    3. Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the open state (Class 1) RCSB Protein Data Bank. 6W2S
    4. Neupane R, Pisareva V, Rodriguez CF, Pisarev A, Fernandez IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the closed state (Class 2) RCSB Protein Data Bank. 6W2T

    Supplementary Materials

    Transparent reporting form
    elife-54575-transrepform.docx (245.4KB, docx)

    Data Availability Statement

    Atomic coordinates have been deposited in the PDB with accession numbers and 6W2S and 6W2T for the open and closed classes , respectively . CryoEM maps have been deposited at the EMDB with accession numbers EMDB 21529 and 21530 for the open and closed classes respectively.

    The following datasets were generated:

    Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map open class. Electron Microscopy Data Bank. EMD-21529

    Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. CryoEM map closed class. Electron Microscopy Data Bank. EMD-21530

    Neupane R, Pisareva VP, Rodriguez CF, Pisarev AV, Fernández IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the open state (Class 1) RCSB Protein Data Bank. 6W2S

    Neupane R, Pisareva V, Rodriguez CF, Pisarev A, Fernandez IS. 2020. Structure of the Cricket Paralysis Virus 5-UTR IRES (CrPV 5-UTR-IRES) bound to the small ribosomal subunit in the closed state (Class 2) RCSB Protein Data Bank. 6W2T

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES