Introduction

Eukaryotic translation initiation occurs in two different modes-Cap-dependent and Cap-independent. Cap-dependent translation initiation in eukaryotes can be divided into four major steps- (i) formation of 43S ribosomal PIC [40S subunit in complex with eIF1, eIF1A, eIF3, eIF5 and ternary complex (eIF2-GTP-tRNA_i)]; (ii) recruitment of 43S PIC to 5’end of capped mRNA, mediated by eIF4F complex (formed by eIF4E,eIF4G and eIF4A) yielding 48S complex (48S PIC); (iii) scanning of 5’ untranslated region (UTR) of mRNA and start codon recognition; and (iv) joining of large 60S ribosomal subunit to form elongation-competent 80S ribosomal complex. Among these, major rate-limiting steps include- the regulation of available tRNA_i as the ternary complex (eIF2.GTP.Met-tRNA_i) and recruitment of mRNA on 43S PIC (Jackson et al 2010; Brito Querido et al 2024). Cryo-EM studies on eukaryotic canonical 48S PICs from yeast (Hussain et al 2014; Llacer et al 2015) to humans (Eliseev et al 2018; Simonetti et al 2020; Brito Querido et al 2020; Yi et al 2022; Brito Querido et al 2024; Petrychenko et al 2024) have revealed various interactions among the initiation factors, mRNA, and ribosome in scanning (P_OUT or open) state and scanning-arrested (P_IN or closed) states. In mammals, 48S PIC formation is mediated by the interaction of eIF3 and eIF4G (Villa et al 2013) and recent attempts to understand the mammalian canonical 48S PIC could capture interactions of eIF4 proteins (eIF4G and eIF4A) with eIF3 (Brito Querido et al 2020; Brito Querido et al 2024).

Alternatively, several positive strand RNA viruses use Internal Ribosome Entry Sites (IRESs), which are internal cis-acting sequences present in 5’ UTR of mRNA that drive the assembly of translation initiation complex without the requirement of 7-methylguanosine cap (Martínez-Salas et al 2018), initially reported in picornaviruses such as poliovirus (PV) and encephalomyocarditis virus (EMCV) RNA genome (Jang et al 1988; Pelletier and Sonenberg, 1988; Trono et al 1988). As of 2020, the IRESbase database reported 1328 IRESs, out of which 554 were viral IRESs from 198 viruses (Zhao et al 2020). These RNA elements have unique secondary and tertiary structures, which allows them to hijack the host translational machinery and promote translation initiation internally by recruitment of host ribosome, canonical eIFs, and IRES trans acting factors (ITAFs) via multiple RNA–RNA and RNA–protein interactions (Lee et al 2017; Martínez-Salas et al 2018). The IRESs lack sequence homology and enfolds different structural organization, thus requiring different eIFs for the assembly of 48S PICs (Jackson et al 2010). The type 1 IRES (represented by PV) and type 2 IRES (EMCV) requires almost all eIFs, except eIF4E or N terminal of eIF4G for the formation of 48S PIC (Lozano and Martinez-Salas, 2015). However, initiation on the latter does not require scanning for start codon recognition (Kaminski et al 1994; Sweeney et al 2014). The type 3 IRES (Hepatitis C virus-HCV) does not require eIF4 factors or scanning for 48S PIC formation (Niepmann and Gerresheim 2020). The type 4 IRES (Cricket paralysis virus-CrPV intergenic IRES) initiates without the requirement of any eIFs and tRNA_i (Johnson et al 2017). The HCV IRES and CrPV intergenic IRES directly interacts with host 40S ribosome, however they differ in their mechanism of attachment. Where, the CrPV IRES (type 4) mimics anticodon-codon interaction via pseudoknot 1 to bind with ribosome in an elongation-competent state (Petrov et al 2016; Fernandez et al 2014) the HCV IRES (type 3) binds to the solvent side of 40S subunit by replacing eIF3 and directly interacts with expansion segment (ES7) of 18S rRNA and ribosomal proteins near mRNA exit channel, thus placing the initiation codon at the P-site (Hashem et al 2013; Niepmann and Gerresheim 2020; Brown et al 2022). While the structures of ribosome-bound type 4 (Spahn et al 2004; Fernandez et al 2014; Muhs et al 2015; Murray et al 2016; Pisareva et al 2018; Acosta-Reyes et al 2019; Abaena et al 2020) and type 3 IRES (Spahn et al 2004; Boehringer et al 2005; Hashem et al 2013; Yamamoto et al 2014; Quade et al 2015; Yamamoto et al 2015; Yokoyama et al 2019; Brown et al 2022) have been determined, there is no structural information about Type 2 and Type 1 IRES and their mode of recruitment to 48S PIC.

The 5’UTR of EMCV, genera Cardiovirus and Picornaviridae family (International Committee on Taxonomy of Viruses Executive Committee, 2020) folds into various stem-loops numbered D-L and, domains H-L, followed by initiation codon at the 834^th residue (∼450 nucleotides in length) makes a functional IRES moiety (Carocci et al 2012). It requires the core of 43S PIC, the central domain of eIF4G, eIF4A, and an essential ITAF-polypyrimidine tract binding protein 1 (PTB1) and the presence of eIF4B enhances 48S formation (Hellen and Wimmer 1995; Martinez-Salaz et al 2018; Sweeney et al 2014). Functional characterization of these domains showed H, and I interact with 40S ribosomal subunit (Chamond et al 2014), and J-K domain recruits eIF4G, enhanced by eIF4A (Pestova et al 1996; Lomakin et al 2000). The structure of J-K domain-eIF4G1(HEAT1)-eIF4A revealed that positively charged patches on the eIF4G1-HEAT1 domain interact with two separated negatively charged clefts (in J and St domain) without perturbing its innate function of recruiting eIF4A (Imai et al 2016; Imai et al 2023). The binding sites for PTB1 were revealed to be dispersed, encompassing H to L domains (Kafasla et al 2009). These domains function as a single entity to form 48S PIC and any mutation in conserved IRES motifs can drastically affect the translation rates (Fernández-Miragall et al 2009; Fernández et al 2011). For example, the RAAA and GNRA loop in the I domain of type 2 IRES are crucial for IRES activity (Robertson et al 1999), and the GNRA loop is also found in other IRES families such as type 1 (poliovirus) and type 5 (Aichi virus A) picornaviruses (Abdullah et al 2023). The inherent flexibility within the IRES domains provide a challenge for structural studies of full length IRES and to capture IRES in the context of 48S PIC. EMCV IRES can independently interact with 40S subunit without any eIFs (Chamond et al 2014) and does not require scanning to recognize the start codon (Pestova et al 1996), unlike type 1 IRESs which scans for the start codon (Sweeney et al 2014). Besides, EMCV IRES can form 48S PIC with HEAT1-eIF4G without the requirement of eIF4G residues 1015-1104 that are known to interact with eIF3 (Lomakin et al 2000). These residues are indispensable in case of canonical initiation on capped mRNAs and for type 1 IRESs (Villa et al 2013; Sweeney et al 2014). However, how EMCV IRES interacts with 40S subunit or what are the molecular interactions essential to form the EMCV IRES-48S PIC remains a question.

In this study we have used pull down assay to isolate EMCV IRES-bound 48S PIC from Rabbit Reticulocyte Lysate (RRL) and subjected the complex to cryo-electron microscopy (Cryo-EM) analysis. The cryo-EM map of EMCV IRES-bound 48S PIC, henceforth mentioned as EMCV IRES-48S PIC, shows densities corresponding to EMCV IRES domains that reveals how EMCV IRES contacts the ternary complex and 40S head at the inter-subunit interface. The structural details presented here accounts for numerous biochemicals studies that have been reported earlier on Type 2 IRES. Furthermore, the structural analysis suggests how Type 2 IRES would interact with 43S PIC to form 48S PIC. Importantly, the study reveals a unique strategy used by viral IRES to capture host translational apparatus for making viral polypeptide.

Results

1. Overview and features of EMCV IRES-48S PIC

To isolate 48S PIC on EMCV IRES, we used a Talon affinity based pull down from nuclease-treated RRL. EMCV IRES RNA used harboured residues from 280 to 905 with AUG (start codon) at 834^th position. PTB1 was recombinantly overexpressed with a N-terminal 6X His tag, followed by a 3C protease cleavage site. PTB1 was incubated with the IRES, followed by RRL addition and 48S PIC was stalled using GMP-PnP. The complex was eluted from Talon matrix employing 3C protease cleavage and pelleted (Sup. Fig. 1.1 A). The EMCV IRES-48S PIC was subsequently analysed by Cryo-EM. The processed data yielded 3 major classes- (i) 40S without any factors-Map A, (ii) 40S-IRES-tRNA_i-Map B, and (iii) 40S-IRES-ternary complex-Map B1, namely, EMCV IRES-48S PIC (Sup. Fig. 1.1 B). Map B and Map B1 has an overall resolution of 4.6Å and 5.0Å, respectively (Sup. Fig. 1.1 B). The core of 40S is at around 4.0Å and the local resolution across the maps were largely in the range of 4.0-8Å (Sup. Figs. 1.2 A and B). Only the extreme tip of beak of 40S in Map B1 (Sup. Fig. 1.2 A) and ends of IRES and eIF2γ is around 12Å in Map B2 (Sup. Figs 1.2 B).

The cryo-EM densities in EMCV IRES-48S PIC corresponds to 40S, tRNA_i, eIF2α, eIF2γ and RNA in the mRNA channel along with an extra density connecting 40S ribosomal head to tRNA_i (Fig. 1A-B). However, it lacks distinct density for PTB1, eIF4G, eIF4A, and eIF3 and hence these factors are not modelled. Since nuclease-treated RRL lacked endogenous RNA, the presence of density for mRNA in the channel indicates trapping of EMCV IRES in 48S PIC. Also, the density connecting 40S head to tRNA_i in Map B and B1 could be assigned to a double stranded RNA structure found in EMCV IRES (Fig. 1A-B).

Besides, an extra density is evident at the mRNA entry site, contacting 40S ribosomal proteins-uS3, eS10, and uS5 and h16 (Helix 16) of 18S rRNA (Sup. Fig. 1.2 C). In the human 48S PIC, this region positions eIF3g-RRM (RNA recognition motif) and downstream ORF (Sup. Fig.1.2 D) (Brito Querido et al 2020; Brito Querido et al 2024). We anticipate this extra density in EMCV IRES-48S PIC could be contributed by downstream ORF, however, it’s difficult to assign it to eIF3g-RRM due to lack of eIF3 (core/peripheral) density. Furthermore, our recent study on interaction of yeast eIF4B with 40S ribosome suggested occupancy of eIF4B (N terminal-RRM) in the same region (Datey et al 2025), which opens up the possibility of mammalian eIF4B-NT-RRM to bind to this region in the absence of eIF3, although in the human 48S PIC eIF4B was tentatively positioned slightly away from this location (Brito Querido et al 2024). An interesting possibility could be positioning one of the RRM domains of PTB1 bound to UCUUU sequence of 18S rRNA (PTB1 can bind to UCUUU sequence- Maris et al 2020) present at the tip of h16 (528-532 nucleotides) of 18S rRNA. An extra density in the same position is also present in Map A (40S without EMCV IRES or factors-Sup. Fig. 1.2 E) as well, and PTB1 was used as a bait for the pull-down, therefore, this density may be contributed by PTB1-RRM interacting with h16 of 18S rRNA.

Figure 1

Table 1:

2. The EMCV IRES-48S PIC is trapped in a closed conformation

In EMCV IRES, A-834-G-836 forms the start codon (Kaminski et al 1994; Hellen and Wimmer 1995; Pestova et al 1996) and it is present in a Kozak context (CRCCaugG; R is a purine) (Kozak 1989), where the -3 position is A-831 and +4 is G-837 (Sup. Fig. 2D). Based on these, we anticipate the 834^th AUG base-paired to the anticodon (Fig. 2A) and the flanking residues were added as per the sequence (Sup. Fig. 2D). The recognition of the start codon at the P-site by tRNA_i leads to accommodation of ternary complex and 48S PIC adopts a P_IN state in contrast to P_OUT state observed during scanning (Llacer et al 2015; Llacer et al 2021; Yi et al 2022; Petrychenko et al 2024). A distinct difference of 7.0Å could be seen by comparing the tRNA_i position of EMCV IRES-48S PIC with that in the open state (Fig. 2B), depicting a P_IN state of tRNA_i. The entrapment of start codon in the P-site also evokes closure of the mRNA latch formed by 18S rRNA helices-h34 of 40S head and h18 of body (Hussain et al 2014; Llacer et al 2015; Hinnebusch 2017). The h34 of 18S rRNA of EMCV IRES-48S PIC is shifted by 9Å as compared to the human canonical open state 48S PIC (PDB-7QP6, Yi et al 2022). This conformation of 18S rRNA corelates well with that of closed 48S PIC (PDB-7QP7, Yi et al 2022), suggesting EMCV IRES-48S PIC was captured in a closed state (Fig. 1C). This closed conformation locks the mRNA in the channel formed within 40S head and body. On the other hand, the conformation of 18S rRNA in Map A (40S ribosome without initiation factors) shows an open state (Sup. Fig. 2E). The transition from open (PDB Id-(7QP6) to closed states (PDB Id-7QP7) structures is also accompanied by an upward shift of N terminal domain of eS17 (connecting the head to body) by 10Å (Yi et al 2022). The EMCV IRES-48S PIC structure corelates with the conformation of eS17 in the human 48S-closed PIC, that is eS17 N-terminal helix associated with the ribosomal head shifts upward by 10Å, keeping C-terminal domain position constant (Fig. 2D; Sup. Fig. 2F).

Figure 2

3. The IRES density connecting 40S head to tRNA_i is contributed by the I domain apical part

In EMCV IRES-48S PIC complex we observed a density, likely for double stranded RNA, connecting 40S head to the tRNA_i elbow region (Fig. 3A). The extra density interacts with the elbow region and acceptor stem of tRNA_i and ribosomal proteins-uS19 (RPS15) and uS13 (RPS18) (Nomenclature-Ban et al 2014). The EMCV IRES contains 4 major stem-loops (H-L) in the functional IRES region (Fig. 3B) (Hellen and Wimmer 1995). Among these, domain H and I have been shown to interact with 40S (Chamond et al 2014) and mutations of conserved residues in these domains severely compromised translation on EMCV IRES (López de Quinto and Martínez-Salas 1997; Hellen and Wimmer 1995). We reasoned that H and I domain may contribute to the double stranded RNA density emanating from 40S head. The density architecture (Fig. 3C) could be interpreted as a long main stem (S1) extending away from 40S where the base is anchored to the ribosome by two branches (B1 and B2), and to tRNA_i by one branch (B3), further divided into two sub-branches (B3a and B3b). Visualizing the IRES RNAfold determined-secondary structure (Fig. 3D-which corelates with the experimental structure proposed in Duke et al 1992), this architecture could be contributed by the apical part of I domain.

To determine the tertiary structure of the I domain apex, the IRES region from nucleotide 507-619 was modelled using Alphafold3 (Sup. Fig. 3A). The model was decomposed and reconstructed based on the best-fit in the obtained density for the IRES using Chimera and Coot (Sup. Fig. 3A). The final model was generated after multiple rounds of geometry correction and real space refinements (Fig. 3E). The final model holds a correlation coefficient of 0.8 with respect to the map (Sup. Fig. 3B), where B1 is AAG loop, B2 is CAAA loop, B3a-GCGA loop, B3b-C rich loop1, and S1 is the major double stranded stem of I domain. The B-factor of the modelled IRES largely ranges from 124 to 200 (Sup. Fig. 3C). The Alphafold3 predicted model of H domain could not fit properly in the density (Sup. Fig. 3D). The J-K domain adopts a Y-shaped structure and its placement of its Cryo-EM (PDB Id-8HUJ) or NMR (PDB Id-2NBX) structure in the density would clash with 40S (Sup. Fig. 3E). Moreover, in context to EMCV IRES-48S PIC, J-K domain binds eIF4G and the location of eIF4G has been mapped close to ES6 of 18S rRNA, located near the left foot 40S ribosome (Yu et al 2011). The I domain apex model in EMCV IRES-48S PIC shows the RAAA and AAG motif contacts uS19 and uS13, and the GNRA loop with tRNA_i (Fig. 4A). These interactions with 40S head and tRNA_i could be facilitated by the long length and flexible nature of I domain.

Figure 3

4. The interaction of I domain of EMCV IRES with ribosomal proteins and initiator tRNA

The I domain is the longest domain in EMCV IRES, which harbours important motifs such as GNRA, RAAA and C rich, crucial for IRES activity (Roberts and Belsham, 1997; Fernández-Miragall and Martínez-Salas 2003) and conserved across all cardioviruses (Hellen and Wimmer 1995). Among these, the CAAA and AAG motifs shares potential contact sites with uS13 and uS19. The alpha helix (100-117 residues) of uS13 contacts the IRES element at the B3 stem (connecting the GNRA loop to RAAA loop) (Fig. 4B). uS19 contacts the IRES majorly at its CAAA motif through multiple sites involving N-terminal residues, residues-67 to 75, C-terminal-102 to 124 (Fig. 4C). These regions of uS13 and uS19 is rich in basic residues, which might interact with the negatively charged backbone of the IRES element (Fig. 4B-C). The role of uS13 and uS19 also involves formation of inter-subunit bridges during 60S joining to form elongation competent 80S complexes. uS13 interacts with uL5 (RPL11) and uS19 with helix 38 or h38 (1748-1778, in humans) in 28S rRNA to form inter-subunit bridges-B1b/c and B1a, respectively (Sup. Fig. 4A) (Ben-Shem et al 2011; Bowen et al 2015; Khatter et al 2015). These interactions are dynamic owing to ribosomal subunit rotation and swivelling during 80S ribosomal translocation states (Khatter et al 2015).

On superimposition of 80S ribosomal structure to EMCV IRES-48S PIC model, the IRES density clashes with the position of uL5 and h38 of 28S rRNA (Sup. Fig. 4B), suggesting repositioning of IRES domain from 40S head during 48S to 80S transition. Interestingly, the h38 residues interacting with uS19 shares considerable similarity in sequence to I domain apex in the EMCV IRES (Fig. 4D), suggesting the I domain apex of EMCV IRES could mimic h38 (60S)-40S interaction (Fig. 4D). The similarity of h38 with the I domain residues provides additional support for annotation of I domain apex in the density.

The GCGA (GNRA motif-where N is any nucleotide and R is a purine) is known for long range RNA-RNA interactions, widespread in ribosomal RNA and in some catalytic RNAs. It forms a characteristic U-turn structure (Fiore and Nesbitt 2013) and interacts with minor grooves of helical RNA elements (Sup. Fig. 4C-Reiter et al 2010). A single point mutation within this tetraloop (GCGA to GCGC) severely reduced the IRES activity, suggesting its essential for IRES activity (Roberts and Belsham, 1997; Fernández-Miragall and Martínez-Salas 2003). The density extending from the elbow region of tRNA_i could fit in the characteristic U-turn, adopted by conventional GNRA motifs. In EMCV IRES, the GNRA motif is represented by GCGA loop, preceded by a C rich region and in close contact with the tRNA_i elbow and acceptor stem (Fig. 4E). Thus, we infer that EMCV IRES interacts with tRNA_i by virtue of its GCGA loop.

Figure 4

5. The position of eIF2-ternary complex is shifted towards 40S head in EMCV IRES-48S PIC in contrast to canonical 48S PIC

We could fit eIF2α and eIF2γ in their respective densities in Map B1 (Fig. 5A). Focussed classification or refinement did not yield any distinct density corresponding to the position of eIF2β, probably due to the flexibility associated with repositioning of eIF2β during transition from open to closed complexes (Llacer et al 2015; Llacer et al 2021; Yi et al 2022; Petrychenko et al 2024). Previous reports on EMCV IRES suggested its direct interaction with eIF2 (Scheper et al 1991; Scheper et al 1994) and inactivation of eIF2 compromises EMCV IRES mediated translation (Welnowska et al 2011; Kwon et al 2016), indicating EMCV IRES’s dependence on canonical ternary complex. Here, we observe a direct interaction of EMCV IRES with ternary complex via tRNA_i, a feature not observed in previously determined HCV and CrPV IRESs-bound ribosomal structures.

As found in canonical 48S PICs (Hashem et al 2013; Brito Querido et al 2024; Petrychenko et al 2024), eIF2α-domain 1 is in close contact with ribosomal protein-uS7, and domain 2 with tRNA_i elbow and domain 3 with eIF2γ in EMCV IRES-48S PIC. The position of eIF2γ and eIF2α-domain3 (D3) is distinct from mammalian 48S-closed PICs (PDB Id-6YAN and 7QP7) as we could observe a shift by ∼10Å towards the head of 40S on superimposing the 18S rRNA from EMCV IRES 48S PIC and mammalian late-stage 48S PIC (Fig. 5B). The ternary complex is flexible, and it moves away from ribosomal head towards the body on recognition of authentic start codon as studied in yeast (Villamayor-Belinchón et al 2024) and human (Petrychenko et al 2024). This opposite directional shift of the ternary complex in the EMCV IRES-48S PIC is evident in tRNA_i acceptor stem as well (Fig. 5C). This shift could be due to the rigid stem B3 (consisting G-C base pairs), connecting 40S head to GNRA loop, which interacts with the tRNA_i at its elbow and acceptor arm, and the association of eIF2α-D3 and eIF2γ with the acceptor arm of tRNA_i orchestrated with the conformational change.

Figure 5

Discussion

The mechanism of IRES recruitment on 40S ribosome varies considerably among different types of IRESs. CrPV intergenic IRES binds to 40S ribosome by mimicking tRNA-mRNA interaction with the help of pseudoknot 1 (Petrov et al 2016; Fernandez et al 2014; Murray et al 2016; Acosta-Reyes et al 2019), whereas HCV IRESs associates with the solvent side of 40S body by replacing eIF3 with its domain 3 (Spahn et al. 2004; Hashem et al 2013; Niepmann and Gerresheim 2020; Brown et al 2022). In this study, we capture EMCV IRES in 48S PIC context, interacting with 40S ribosomal head and tRNA_i. Model building suggests that I domain of EMCV IRES interacts with 40S ribosome head and tRNA_i elbow stem (Fig. 6A). The similarity in the sequence of h38 of 28S rRNA with the I domain and its ability to interact with uS19-N terminal via RAAA motif (Khatter et al 2015) suggests a mimicry mechanism adopted by EMCV IRES for its recruitment to the ribosome. Moreover, the conservation of I domain apex sequence and motifs (RAAA, AAG loop, C-rich and GNRA) across Cardioviruses-Theiler’s murine Encephalitis virus (TMEV), Vilyuisk human encephalomyelitis virus (VHEV), Theiler’s-like rat virus (TRV) and Saffold viruses 1 and 2 (SAFV-1 and SAFV-2) and Apthoviruses (AAG loop is replaced by ACG loop)-Foot and mouth disease virus (Hellen and Wimmer 1995; Liang et al 2008) suggests that these IRESs might adopt similar strategies for 48S formation (Fig. 6B-C). EMCV IRES can directly interact with 40S unlike type 1 IRESs. The I domain of EMCV IRES is similar to domain IV of polioviral IRES (or other type 1 IRESs) in terms of length, secondary structure and conserved motif’s (GNRA, C-rich) positioning (Fig. 6C), therefore, anticipating a similar interaction with tRNA_i, highlighting a sequestering tendency by competing with cellular mRNAs. However, the lack of RAAA motif in type 1 provides a plausible reason for no direct interaction between type 1 IRES and 40S (Lozano and Martinez-Salas 2015).

The EMCV IRES does not require scanning, and the start codon (A-834) is directly placed in the P-site, which would eventually place the L domain at the mRNA exit site, preceded by J-K domain that interacts with eIF4G-eIF4A (Fig. 6A). Earlier, biochemical studies suggested eIF4G to be positioned close to ES6 of 18S rRNA in EMCV IRES-bound 48S PIC (Yu et al 2011). The human 48S PIC with a 5’ capped mRNA showed a similar location for eIF4G, that is at the mRNA exit site contacting eIF3 (Brito Querido et al 2020; Brito Querido et al 2024). Locating eIF4F have been challenging due to inherent flexibility associated with the eIF4F complex on mRNA and requires association with eIF3 in canonical 48S context (Brito Querido et al 2020; Brito Querido et al 2024). However, eIF3-eIF4G interaction is dispensable for EMCV IRES-48S PIC formation, so we do not rule out the possibility that EMCV IRES may dislodge eIF3 from its position on the solvent surface as observed in case of HCV IRES (Hashem et al 2013). Although, PTB1 serves as an essential ITAF for 48S PIC formation on EMCV IRES, the obtained map shows no distinct density to PTB1. PTB1 binds to H domain and I domain base, and K domain loop (Kafasla et al 2009) and the flexibility associated with these domains might have hindered capturing of PTB1 in the reported 48S complex (Fig. 6A). However, the unassigned extra density at the mRNA entry site could be contributed by PTB1-RRM interacting with 18S rRNA as discussed previously.

Figure 6

The structural studies on type 2 IRESs have been limited be due to their flexible nature. In the cryo-EM map and structural analysis presented here, we could capture a part of EMCV IRES in 48S context (Fig. 6A) and fetch a significant understanding of ribosomal recruitment by EMCV IRES and 48S PIC formation. The entrapment of additional factors and a higher resolution would significantly provide more insights into EMCV IRES 48S PIC. The conservation of secondary structures and motifs such as GNRA within the picornaviruses (type 1, 2, and Aichiviral IRESs-Abdullah et al 2023) provokes common strategies of interaction in the context of 48S PICs in the Picornaviridae family.

Materials and methods

1. Plasmid constructs and Molecular Cloning

EMCV IRES 905 was obtained from EMCV-L plasmid-EMCV IRES (nt 280–905) into pCR2.1 and inserted in pcDNA3.1 using BamH1 and Xba1 restriction sites. Polypyrimidine tract binding protein 1 was cloned from HEK293 cDNA and 3C protease site (LEVLFQGP) was inserted at the N-terminal and inserted in pET28a in between BamH1 and HindIII restriction sites, retaining the N-terminal 6X Histidine tag. Sequence of oligos or primers used is listed in Sup. Table 1.

2. Protein overexpression and purification

PTB1 (Histidine tag-3Cprotease site-PTB1 gene) was overexpressed using 0.5mM IPTG in E. coli BL21 cells at 30⁰C, 120 rpm for 4 hours in Luria broth. 2l harvested culture was lysed using Buffer N1 (20 mM HEPES pH-7.4, 300 mM KCl, 2mM MgCl₂, 10 % Glycerol, 10mM Imidazole, 5mM β-Mercaptoethanol, 0.05% Triton-X-100, 2mM PMSF) and sonicated at 18% amplitude, 10s ON and 20s OFF pulses, 30 cycles and centrifuges at 20000 rpm for 20 minutes and the supernatant was loaded on Ni-NTA column and eluted using a gradient of N1 to N2 buffer (20 mM HEPES pH-7.4, 300 mM KCl, 2mM MgCl₂, 10 % Glycerol, 500mM Imidazole, 5mM β-Mercaptoethanol). PTB1 was eluted at 250 mM Imidazole concentration. The eluant fractions (diluted to 100 mM KCl) were further loaded on Heparin column and eluted by applying a gradient of 100 mM to 1000 mM KCl. The protein was eluted at 250 mM KCl concentration. The eluant fractions were stored at -80⁰C after Size Exclusion Chromatography in Buffer S (20 mM HEPES pH-7.4, 200 mM KCl, 2mM MgCl₂, 5 % Glycerol, 1mM DTT) at 2.5 mg/ml concentration.

3C protease was overexpressed using 0.5mM IPTG in E. coli BL21 cells at 30⁰C, 120 rpm in Luria broth. 2l harvested culture was lysed using Buffer N11 (20 mM HEPES pH-7.4, 200 mM KCl, 2mM MgCl₂, 10 % Glycerol, 10mM Imidazole, 5mM β-Mercaptoethanol, 0.05% Triton-X-100, 2mM PMSF) and sonicated at 18% amplitude, 10s ON and 20s OFF pulses, 30 cycles and centrifuges at 20000 rpm for 20 minutes and the supernatant was loaded on Ni-NTA column and eluted using a gradient of N11 to N22 buffer (20 mM HEPES pH-7.4, 200 mM KCl, 2mM MgCl₂, 10 % Glycerol, 500mM Imidazole, 5mM β-Mercaptoethanol). 3C protease was eluted at 250 mM Imidazole concentration. The eluant fractions were subjected to Size Exclusion Chromatography in Buffer S (20 mM HEPES pH-7.4, 100 mM KOAc, 2mM MgCl₂, 10 % Glycerol, 1mM DTT) and peak fractions were concentrated using 3kDa cutoff centricon and stored at 2.5 mg/ml concentration.

3. In vitro transcription of EMCV IRES

EMCV IRES 905-pcDNA3.1 was linearized using Xba1 restriction enzyme and transcribed using Promega-RiboMAX™ Large Scale RNA Production Systems as per manufacturer’s protocol. 1 µg of linearized plasmid yielded 87 µg of RNA after DNase treatment and RNA cleanup (RNA clean up kit-NEB).

4. Assembly of EMCV IRES 48S PIC using Talon Affinity Chromatography

12 µg of IRES was heated at 95⁰C to dissolve any secondary structures acquired and refolded using Buffer R (20 mM HEPES pH-7.4, 150 mM KOAc, 2mM MgCl₂, 2mM β-Mercaptoethanol, 0.5mM Spermidine) at 37⁰C for 5 mins and PTB1 (1:2 = IRES: PTB1) is added and further incubated at 30⁰C for 5 mins. Simultaneously, Rabbit Reticulocyte Lysate (RRL) (Promega) was incubated at 30⁰C with ATP, Amino acid mix minus Leucine, Murine RNase Inhibitor for 5 minutes and then mixed with IRES-PTB1 vial with instant addition of 6mM GMP-PnP per reaction (100 µl) and incubated at 30⁰C for 8 minutes, followed by ice incubation. The reaction was loaded on Talon beads equilibrated with Buffer A (20 mM HEPES pH-7.4, 150 mM KOAc, 2mM MgCl₂, 4 % Glycerol, 5mM Imidazole, 2mM β-Mercaptoethanol). After recommended passes, reloading and incubation, the flow through was collected and the beads were washed with Buffer A until A260 attains a baseline value (∼0), following which 3C protease in Buffer A was added to the beads and left for overnight incubation. Fractions were eluted using Buffer A (400 µl each in 10 vials) and subjected to analysis such as A260 measurements, and agarose gel electrophoresis. Samples having the RNA bands were pelleted using an 1ml Sucrose cushion (20 mM HEPES pH-7.4, 150 mM KOAc, 2mM MgCl₂, 30% Sucrose, 1mM DTT) in SW60 tubes and centrifuged at 50,000 rpm for 10 hours at 4⁰C. The pellet was resuspended in 20µl buffer R (+0.2 mM Spermidine) and used for Cryo-EM grid preparation (No crosslinker was used to avoid artifacts).

5. Negative stain analysis and Mass Spectrometry

Final sample was diluted 10 times using Buffer R and applied on 400-mesh Cu TEM grids which were freshly glow-discharged (negative polarity) for 30 s in GloQube glow-discharge system, stained using 1% Uranyl acetate solution and analysed using microscope at 57000 X magnification. Furthermore, the sample was digested by trypsin and subjected to NanoOrbitrap analysis for identification of proteins in the complex.

6. Cryo-EM sample preparation

3 µl of resuspended pellet (2.96 A260) was applied on glow discharged Quantifoil R 1.2/1.3 300 mesh 2nm Carbon Coated Grid and Blotted using 8s and 8.5s Blot time, zero Blot force at 16⁰C and 100% Humidity and plunged into liquid ethane. Cryo-EM data were collected on Talos Arctica transmission electron microscope equipped with a FEG at 200 kV (Thermo Fisher Scientific). All data were collected using a Gatan K2 Summit Direct Detector at a nominal magnification of 36,000 X and a pixel size of 1.17 Å with a total electron dose of 55 e-/Å2 fractionated over 20 frame movies with a dose rate of ∼2.5 e-/Ų /frame.

7. Data Processing

Micrographs were collected and processed using CryoSPARC v3.3 (Punjani et al 2017). The micrographs were patch motion corrected and CTF was estimated. Using a blob picker with an average diameter of 300 Å particles were picked and extracted. The extracted particles were subjected to multiple rounds of 2D classification. Final 2D classes showing promising ribosome 2D features were selected for ab initio reconstruction. The junk was discarded, and the 237054 good particles were classified into 2 classes using a mask around tRNA_i. We obtained 2 classes-Empty 40S (1.25L particles-Class 1 or Map A), and 40S with tRNA bound (1.11L particles-Class 2). Class 2 was further classified into 2 classes using a 3D mask around the IRES density and subjected to non-uniform refinement with global CTF refinement to yield Map B, having tRNA_i and IRES density (55k particles). This was subjected to Non-uniform Refinement (Punjani et al 2020) with global CTF refinement. Map B class was further classified using a mask around eIF2α and eIF2γ to yield 2 classes-40S-tRNA_i-IRES-eIF2 (28k particles-Map B1) and 40S-tRNA_i-IRES (26k particles). The obtained maps were then subjected to model building and refinement. The local resolution for the obtained maps were estimated using Phenix (Liebschner et al 2019).

8. Map analysis and Model building

To the obtained maps, 40S ribosome (PDB Id-6YAN) was fitted using UCSF Chimera. The 40S head (Head proteins + 18S rRNA (1197-1688) + eS17) and 40S body (Body proteins + 18S rRNA (1-1196; 1689-1870)) was fitted to the obtained maps separately and subjected to Rigid Body Fit and Real Space Refinement using PHENIX (Liebschner et al 2019). The models were merged using Coot (Emsley et al 2010) and subjected to Real Space Refinement. tRNA_i and mRNA models were taken from PDB Id-8OZ0 and fitted to the maps-B and B1 in Chimera. eIF2α and eIF2γ from PDB Id-8OZ0 were rigid-body fitted to Map B1 and then mutated as per Rabbit eIF2 protein sequence (NCBI Reference Sequence-XP_002719561.1; XP_051683593.1) and subjected to a final Real Space Refinement. The IRES I domain apex model was predicted from Alphafold3 (Abramson et al 2024) and the helical sub-domains were dismantled and fitted according to best fit using Chimera, followed by chain joining in Coot with manual Real Space Refinement (Afonine et al 2018). The geometry was corrected using Geometry minimization tool in Phenix with rounds of Real Space Refinement. H domain model was predicted using Alphafold3. The final model yielded 40S-EMCV IRES-tRNA_i (Map B) and 40S-EMCV IRES-tRNA_i-eIF2αϒ (Map B1). All the figures were made using ChimeraX (Pettersen et al 2021). The models were real space refined using Phenix and the Fourier Shell Correlation for ‘Map to Model’ for each was determined at 0.5 FSC (Sup. Fig. 2A-C; Table 1).

9. Secondary Structure Determination and Multiple sequence alignments

The secondary structures for I domain apex was obtained from RNAfold (Gruber et al 2008). Multiple sequence alignments were performed using Clustal Omega (Sievers et al 2021). Sequence accession number for various sequence used-EMCV (NC_001479.1), FMDV (NC_039210.1), TMEV (DQ401688.1), TRV (AB090161.1), VHEV (M80888.1), SAFV (FM207487.1), Poliovirus (NC_002058.3).

Data availability

Maps and atomic coordinates of the 40S ribosome (Map A), 40S ribosome-EMCV IRES-tRNA_i (Map B) and 40S ribosome-EMCV IRES-ternary complex (Map B1) have been deposited in the EMDB database and in the PDB database.

Supplementary figures and table

Supplementary Figure 1.1

Supplementary Figure 1.2

Supplementary Figure 2

Supplementary Figure 3

Supplementary Figure 4

Supplementary Table 1:

Acknowledgements

EMCV IRES (nt 280–905) into pCR2.1 was a kind gift from Dr. Bruno Sargueil, CNRS UMR8015, Universit’e Paris Descartes, France and 3C protease-pET24 was a kind gift from Prof. Raghavan Varadarajan, Indian Institute of Science (IISc), India. We thank various central facilities, namely Cryo-EM, Mass Spectrometry and Computational cluster at the Division of Biological Sciences, IISc for support. DD acknowledges the DBT-JRF Programme for fellowship. The authors acknowledge the DST-FIST support to the department. This work was supported by the Intermediate Fellowship from DBT-Welcome Trust India Alliance to TH (IA/I/17/2/503313).