Topology and regulation of the human eIF4A/4G/4H helicase complex in translation initiation

Abstract

The RNA helicase eIF4A plays a key role in unwinding of mRNA and scanning during translation initiation. Free eIF4A is a poor helicase and requires the accessory proteins eIF4G and eIF4H. However, the structure of the helicase complex and the mechanisms of stimulation of eIF4A activity have remained elusive. Here we report the topology of the eIF4A/4G/4H helicase complex, which is built from multiple experimentally observed domain-domain contacts. Remarkably, some of the interactions are continuously rearranged during the ATP binding/hydrolysis cycle of the helicase. We show that the accessory proteins modulate the affinity of eIF4A for ATP by interacting simultaneously with both helicase domains and promoting either the closed, ATP-bound conformation or the open, nucleotide-free conformation. The topology of the complex and the spatial arrangement of the RNA-binding surfaces offer insights into their roles in stimulation of helicase activity and the mechanisms of mRNA unwinding and scanning.

Figure 1. Domain organization of eIF4G, eIF4A and eIF4H

(A-C) Domain organization of eIF4A (A), eIF4H and eIF4B (B), and eIF4G (C) and constructs used in this work. In panel C, conserved sequence segments in eIF4G are marked above the diagram. Sites of interaction with other proteins and RNA are marked with arrows below the diagram. Abbreviations: NTD, N-terminal domain; CTD, C-terminal domain; RRM, RNA recognition motif domain; 4H-CT-long and 4H-CT-short, long and short C-terminal fragments of eIF4H; PAM-1, PABP-binding motif-1; E, predicted extended eIF4E-binding region; S, RNA-binding region important for scanning, aa. 682–721 (Prevot et al., 2003); N, H1-NT motif, aa. 722–741; C, H1-CT motif, aa. 994–1027; Y, y4G-CT motif, aa. 1118–1136 (Marintchev and Wagner, 2005). (D) Starting model for the eIF4A/4G/4H interaction network. eIF4A-CTD is shown in surface representation and the eIF4G HEAT-1 contact surface (Oberer et al., 2005) is yellow. The interdomain orientation between eIF4A-CTD and eIF4A-NTD (navy ribbon), and the RNA (red wire) are modeled based on the structure of eIF4A3 (Andersen et al., 2006). The interdomain orientation of the HEAT domains of eIF4G (Marintchev and Wagner, 2005) is modeled based on the interdomain orientation in CBP80 (Mazza et al., 2002). The HEAT domains are color-coded as in panel C. Sites of mutations in eIF4G reported to affect eIF4A binding are shown as light blue (top cluster) and navy (bottom cluster) wires. Residues relevant for this work are labeled. The predicted binding sites of eIF4H, eIF4A-NTD and eIF4A-CTD are shown with arrows.

Figure 2. The F978A mutation in eIF4G HEAT-1 affects binding to eIF4A, but not to eIF4A-CTD

(A, B) Overlay of SPR sensorgrams showing binding of eIF4A to immobilized WT eIF4G HEAT-1 (A) and eIF4G HEAT-1 F978A (B). (C, D) SPR graphs showing binding of full-length eIF4A (C) and eIF4A-CTD (D) to immobilized WT eIF4G HEAT-1 (black) and HEAT-1 F978A (red). K_D values marked with a star should be considered only estimates, since concentrations higher than the K_D could not be reached in the titration due to limited solubility.

Figure 3. eIF4G HEAT-2 binds to both domains of eIF4A

(A) Overlay of ¹⁵N TROSY-HSQC spectra of 0.4 mM ¹⁵N-labeled eIF4G HEAT-2 alone (black) and in the presence of 0.15 mM (blue) or 0.4 mM (red) unlabeled eIF4A-NTD. (B) Overlay of ¹⁵N HSQC spectra of 0.15 mM ¹⁵N-labeled eIF4G HEAT-2 alone (black) and in the presence of 0.15 mM unlabeled eIF4A-CTD (red). (C) eIF4A binding surfaces of eIF4G HEAT-2. The orientation of HEAT-2 in the left panel is the same as that in Figure 1D, whereas in the middle panel it is rotated 180° along the Y-axis. The right panel shows the HEAT-2 domain in ribbon, in the same orientation as in the middle panel. Residues affected by eIF4A-NTD binding are painted in dark blue; residues affected by eIF4A-CTD binding are painted in light blue.

Figure 4. Topology of the eIF4A/4G/4H helicase complex

(A) eIF4G-binding surfaces of eIF4A-CTD. The eIF4A structure is as in Figure 1D. Residues of eIF4A-CTD affected by binding of eIF4G are painted in yellow (for eIF4G HEAT-1), light orange (the linker between the eIF4G HEAT-1 and HEAT-2 domains) and dark orange (eIF4G HEAT-2). Residues in eIF4A-CTD not affected by any of the interactions are in blue; residues that could not be used for mapping (e.g., due to spectral overlap) are in grey. (B) eIF4H-binding surface of eIF4A-CTD. eIF4A is shown in the same orientation as in the right panel in A, above. Residues in eIF4A-CTD affected by eIF4H binding are in magenta. (C) Overlap between the eIF4H- and the eIF4 G linker-binding surfaces of eIF4A-CTD. Coloring of residues affected by eIF4H binding is as in panel B, and coloring of residues affected by eIF4G binding is as in panel A, except the residues affected by both eIF4H and eIF4G linker binding, which are painted in purple. The overlap is shown schematically by superimposing cartoons representing eIF4H-CTD (magenta oval) and the eIF4G interdomain linker (orange line). (D)Topology of the eIF4A/4G/4H helicase complex. The mutual orientation of the eIF4G domains and eIF4H is modeled after the structure of the nuclear CBP80/CBP20 complex (Mazza et al., 2002). The orientation of eIF4A with respect to eIF4G is based on NMR chemical shift mapping (this work and (Oberer et al., 2005)). Domains, whose structures are known or could be modeled, are displayed as solid bodies with size and shape corresponding to their structures, providing the overall topology of the complex. eIF4H-CTD, whose structure is not known, is shown as a circle. The eIF4G interdomain linker (not shown) is expected to be wrapping around the eIF4G HEAT-1 domain (as does the corresponding linker in CBP80 (Mazza et al., 2002), see Figure S3A), and around eIF4A-CTD (as shown in this work). The position of the RNA on eIF4A is modeled based on the structure of eIF4A3 (Andersen et al., 2006)). The RNA-binding site of the eIF4H RRM domain is based on that of its homolog CBP20 in the CBP80/CBP20 complex (Mazza et al., 2002). The RNA-binding site of the eIF4G HEAT-1 domain (marked with a star) is based on mutation data (Marcotrigiano et al., 2001).

Figure 5. Effects of ATP and ADP on the eIF4A/eIF4G and eIF4A/eIF4H interactions

(A, B) SPR graphs of eIF4A binding to immobilized eIF4G HEAT-1 (A) or eIF4G HEAT-2 (B) in the absence of nucleotide (black) and in the presence of 1 mM ATP (red) or 1 mM ADP (blue). (C) Overlay of SPR sensorgrams showing binding of 10 μM eIF4A to immobilized eIF4H in the absence of nucleotide (black), in the presence of 1 mM ATP (red) and 1 mM ADP (blue). (D) Fluorescence anisotropy graphs of eIF4A binding to FITC-labeled U₄₀ RNA oligonucleotide in the absence of nucleotide (black) and in the presence of 1 mM ATP (red) or 1 mM ADP (blue). The K_D values in panel D, marked with a star, should be considered estimates because of the tendency of eIF4A to aggregate at higher concentrations. Binding curves at lower eIF4A concentrations are shown since those are least affected by eIF4A aggregation. (E) Summary of the effects of eIF4G, eIF4H and RNA on the affinity of eIF4A for nucleotides. RNA, eIF4H, and eIF4G HEAT-1 (left) favor the closed ATP-bound conformation, whereas eIF4G HEAT-2 (right) prefers the open nucleotide-free conformation of eIF4A.

Figure 6. Opposing effects of eIF4G SY and eIF4G HEAT-2 on RNA binding to eIF4A

(A) Fluorescence anisotropy graphs showing binding of FITC-labeled U₄₀ RNA oligonucleotide to eIF4A in the presence of 1 mM ATP (red); eIF4G SY (cyan); and the eIF4A/eIF4G SY complex (purple). The eIF4G SY fragment (see Figure 1C) consists of the HEAT-1 domain and additional RNA- and eIF4A-binding segments. (B) Fluorescence anisotropy graphs showing inhibition by eIF4G HEAT-2 of the binding of 10 μM eIF4A to 50 nM FITC-labeled U₄₀ RNA in the absence of nucleotide (black) and 2.5 μM eIF4A to 50 nM FITC-labeled U₄₀ RNA in the presence of 1 mM ATP (red). (C, D) Isothermal Titration Calorimetry (ITC) graphs of eIF4G HEAT-2 binding to eIF4A in the absence (C) and presence (D) of eIF4G HEAT-1. Note that in the experiment shown in panel D, the concentration of eIF4G HEAT-1 (50 μM) is not saturating: it is only ~4 times higher than the K_D of the eIF4A/HEAT-1 interaction (12 μM, Figure 2C). Therefore, a fraction of eIF4A is not bound to HEAT-1 and the calculated apparent K_D for the interaction between eIF4A and eIF4G HEAT-2 in the presence of HEAT-1 (marked with a star on panel D) should be considered a lower limit of the actual K_D.

Figure 7. Models for the mechanisms of unwinding of mRNA and scanning

(A) Model for the dynamics of the eIF4A/eIF4G interactions eIF4G HEAT-1 (yellow) stimulates ATP binding and the helicase activity of eIF4A by simultaneous binding to both eIF4A domains in the closed ATP-bound conformation (left). eIF4G HEAT-2 (orange) favors the nucleotide-free state by simultaneous binding to both eIF4A domains in an open conformation (right). The interdomain linker of eIF4G (light orange) also binds eIF4A and stabilizes the complex. The arrows indicate directions of rearrangements during the ATP hydrolysis/nucleotide exchange cycle. ATP (not shown) binds at the interface between the two eIF4A domains. HEAT-2 is shown semi-transparent to emphasize that it is not required for the ATP-binding/hydrolysis cycle. (B) Hypothetical model for the organization of the cap-binding complex and the scanning complex. State 1. Model of the cap-binding complex bound at the 5′-cap. The orientation and coloring of eIF4A, eIF4G and eIF4H is as in Figure 4D. The linker between the eIF4G HEAT-1 and HEAT-2 domains is not shown. The mRNA is drawn as a dashed red line. The 5′-cap/eIF4E/4G complex structure shown is from yeast (Gross et al., 2003). State 2. Model for the unwinding of the 5′-proximal region of mRNA. The position and 5′-3′ polarity of the mRNA on eIF4A is modeled based on the structure of eIF4A3 (Andersen et al., 2006)). The direction of translocation/unwinding along the mRNA (5′ to 3′) is indicated by an arrow. State 3. Model for the scanning complex. The small ribosomal subunit (grey semi-transparent surface, with the rRNA backbone shown as ribbon) and the mRNA (red solid ribbon) are from 1JGP.pdb (Yusupova et al., 2001). The direction of scanning of the initiation complex along the mRNA (5′ to 3′) is indicated by an arrow. According to this model, eIF4A is on the front (3′-side) of the scanning initiation complex, in agreement with its helicase function.