Supplementary data. Supplementary Material and Methods. Cloning. The coding sequences of eIF2B subunits α, β and δ were amplified by PCR as individual ...
Supplementary data Supplementary Material and Methods Cloning The coding sequences of eIF2B subunits α, β and δ were amplified by PCR as individual exons from Chaetomium thermophilum genomic DNA and introduced into the expression vector according to the InFusion protocol (Clontech). The coding sequence for full‐length eIF2Bα was introduced into a modified pET15b vector cleaved with the restriction enzymes NcoI and XhoI. The coding sequences for full‐length cteIF2Bβ and ‐δ were introduced into the pGEX‐6P1 vector cleaved with BamHI and XhoI. The constructs encoding cteIF2BβΔ123‐148 and cteIF2Bδ130‐C were generated by deletion of the corresponding coding regions from the original plasmids (wild‐type eIF2Bβ and eIF2Bδ in pGEX‐6P1). The deletion constructs as well as site‐directed mutagenesis to generate point mutants of cteIF2Bα, ‐β or ‐δ were performed according to the QickChange protocol (Stratagene). Protein expression and purification Expression: All constructs were expressed in E. coli BL21(DE3) Rosetta II cells (Novagene). Cells containing the respective plasmids were grown in 2xYT medium (supplemented with antibiotics) at 37 °C with shaking at 220 rpm until they reached an OD600 of 0.8. Subsequently the cell cultures were transferred to 16 °C and the expression was induced by the addition of isopropyl‐β‐D‐ thiogalactopyranosid (IPTG) to a final concentration of 0.5 mM. The cells were harvested after 16 hours at 16 °C. Purification of eIF2Bα: Cells containing N‐terminally His‐tagged eIF2Bα (wild‐type and mutants) were resuspended in L‐100‐His buffer (20 mM Hepes (pH 7.5), 100 mM KCl, 10% glycerol, 15 mM imidazole, 2 mM β‐mercaptoethanol) containing a mixture of protease inhibitors. Cell lysis was performed in a microfluidizer (Microfluidics, USA), and cell debris was removed by centrifugation for 30 min at 30,000 xg. The supernatant was loaded onto a HisTrap column (GE Healthcare) equilibrated in L‐100‐His buffer. The column was then washed with 2 column volumes of L‐100‐His buffer and bound protein was eluted with a linear gradient into elution buffer (L‐100‐His buffer with 350 mM imidazole). Fractions containing the target protein were desalted in L‐100‐His buffer on a HiPrep Desalting column (GE Healthcare), before adding TEV protease in a 1:100 weight ratio of protease to target protein to remove the His‐tag. The cleaved protein was separated from the His‐tag and uncleaved protein by a second HisTrap run. The flow‐through was finally concentrated to a volume of 5 ml and applied to a Superdex S‐200 gelfiltration column (GE Healthcare) equilibrated in G‐100 buffer (10 mM Hepes (pH 7.5), 100 mM KCl, 10% glycerol
1
and 2 mM DTT). The protein was concentrated to ~20 mg/ml, flash‐frozen in liquid nitrogen and stored at ‐80 °C. Purification of eIF2Bβ and eIF2Bδ: All constructs (wild‐type and mutants) were expressed as N‐terminally GST‐tagged fusion proteins. Cell lysis was performed as described for eIF2Bα with the difference that L‐ 500 buffer (20 mM Hepes (pH 7.5), 500 mM KCl, 10% glycerol, 4 mM β‐mercaptoethanol) was used to resuspend the cells. After removal of cell debris by centrifugation, the supernatant was applied to a GSTrap column (GE Healthcare) equilibrated in L‐500 buffer. After washing the column with 4 column volumes of L‐500 buffer, bound protein was eluted with 30 mM reduced glutathione in L‐500 buffer. Fractions containing the target protein were pooled and desalted on a HiPrep Desalting column equilibrated in desalting buffer (10 mM Hepes (pH 7.5), 200 mM KCl, 10% glycerol, 4 mM β‐ mercaptoethanol). The pooled protein was incubated over night at 4 °C with Prescission protease in a 1:100 weight ratio of protease to target protein to remove the GST‐tag. In order to remove the GST‐ tagged protease, cleaved GST and uncleaved fusion protein the sample was applied to a second GSTrap column equilibrated in desalting buffer. The flow‐through was pooled and concentrated to 5 ml before loading onto a Superdex S‐200 gelfiltration column equilibrated in G‐100 buffer. eIF2Bβ, eIF2Bβ∆123‐148, and eIF2BβA286E which eluted as pure protein in a single peak, were concentrated to 15‐20 mg/ml and stored at ‐80 °C. eIF2Bδ usually suffered strong degradation resulting in the cleavage of some of the full‐ length protein into two fragments migrating at 37 kDa and 20 kDa according to SDS‐PAGE. As a consequence, eIF2Bδ eluted in two major peaks from the S‐200 column containing either the full‐length protein or the 37‐kDa fragment. According to MS analysis, the 37‐kDa fragment contained residues 128‐ 466 of eIF2Bδ. Full‐length eIF2Bδ and the 37‐kDa fragment were pooled separately, concentrated to 10‐ 15 mg/ml and stored at ‐80 °C. Deletion constructs and point mutations of eIF2Bβ and ‐δ were purified according to the same protocol. Protein crystallization and structure determination eIF2Bβ. Initial crystallization trials for full‐length eIF2Bβ were performed by sitting‐drop vapor diffusion with commercially available standard screens. No crystals were obtained at any of the tested protein concentrations (between 6 and 15 mg/ml) or temperatures (4 and 20 °C). In order to improve the crystallizability of the protein, we decided to remove residues 123 to 148 by deletion of the corresponding coding sequence in the expression vector. According to multiple sequence alignments of C. thermophilum eIF2Bβ with its homologs from other species, these residues are not conserved and seem to correspond to an extended loop region between two conserved α‐helices that is idiosyncratic to 2
cteIF2Bβ (Figure S1). Thus, we reasoned that its removal would have no negative impact on the overall structure of the protein, which is supported by the fact that it still forms a complex with eIF2B subunits α and δ. In initial crystallization trials with the construct eIF2BβΔ123‐148, crystals were obtained after 5 days at 20 °C in a condition containing 100 mM HEPES (pH 7.5) and 1.0 M tri‐sodium citrate. After optimization, diffraction quality crystals were obtained with 9.5 mg/ml protein at 20 °C in 100 mM HEPES (pH 6.8) and 1.33 M tri‐sodium citrate. X‐ray diffraction data were collected at BL 14.1 (HZB, BESSY, Berlin) (1). The phase problem was solved by molecular replacement using the program PHASER (2) and the atomic coordinates of the N‐terminal domain of the Bacillus subtilis 5‐methylthioribose 1‐ phosphate isomerase (PDB: 2YVK) and the C‐terminal domain of Homo sapiens eIF2Bα (PDB: 3ECS) as independent search models. The structure was refined in trigonal space group R3 at a resolution of 2.54 Å using the program PHENIX (3). Missing regions of the peptide chain were built manually in Coot (4). The final model contains two molecules per asymmetric unit (see Table 1 for details of data collection and refinement). eIF2Bδ. Initial crystallization trials for eIF2Bδ were performed by sitting‐drop vapor diffusion with commercially available standard screens. With the truncated version of eIF2Bδ (36 kDa fragment) initial crystals grew with 10 and 12 mg/ml protein at 20 °C in a condition containing 100 mM MES (pH 6.0) and 1.0 M (NH4)2SO4. After optimization, diffraction quality crystals were obtained in 100 mM MES (pH 6.2) and 0.8 M (NH4)2SO4. X‐ray diffraction data were collected at BL 14.1 (HZB, BESSY, Berlin) (1). The phase problem was solved by molecular replacement using PHASER (2). As search models the atomic coordinates of α‐helices 2‐4 (residues 27‐93) of the N‐terminal domain of Homo sapiens eIF2Bα (PDB: 3ECS), and residues 116‐295 of its C‐terminal domain were used. The structure was refined in primitive orthorhombic space group P212121 at a resolution of 2.55 Å using the program PHENIX (3). The final model contains two molecules per asymmetric unit (see Table 1 for details of data collection and refinement). 3
Supporting Figures
Figure S1. Multiple sequence alignment of eIF2Bβ orthologs. Highly conserved residues are highlighted in dark blue, conserved residues in light blue and variable residues in gray; the numbering above the alignment corresponds to the C. thermophilum ortholog (cteIF2Bβ). Residues that were removed in the crystallized cteIF2BβΔ123‐148 construct are highlighted in red. Helices α5 and α6 with the kink region as 4
well as the C‐terminal helix α12 are indicated as magenta cylinders. Surface exposed Gcd‐ and Gcn‐ mutations are indicated with S. cerevisiae numbering below the alignment in orange and pink, respectively; exposed VWM/CACH mutations are indicated with H. sapiens numbering in green; the corresponding C. thermophilum numbers are given in brackets. Mutations of cteIF2Bβ used in this study are indicated by diamonds and the introduced amino‐acid above the alignment. Species abbreviations: ct – Chaetomium thermophilum; sc – Saccharomyces cerevisiae; os ‐ Oryza sativa; sp – Schizzosaccharomyces pombe; dd – Dictyostelium dyscoideum; tr ‐ Takifugu rubripes; gg – Gallus gallus; oc – Oryctolagus cuniculus; hs – Homo sapiens; bt – Bos taurus; mm – Mus musculus; rn – Rattus norvegicus.
5
Figure S2. Multiple sequence alignment of eIF2Bδ orthologs. Highly conserved residues are highlighted in dark blue, conserved residues in light blue and variable residues in gray; the numbering above the alignment corresponds to the C. thermophilum ortholog (cteIF2Bδ). The poorly conserved N‐terminal tail which is absent from the crystal structures is not included in the alignment; the black arrow marks the first residue defined in the isolated crystal structure, the red arrow marks the first residue defined in cteIF2Bδ of the eIF2B(βδ)2 complex structure. Helices α5 and α6 with the kink region as well as the C‐ terminal helix α12 are indicated as magenta cylinders. For helix α12 the secondary structure prediction, as calculated by the Psipred server (http://bioinf.cs.ucl.ac.uk/psipred/), is shown behind the alignment. Surface exposed Gcd‐ and Gcn‐ mutations are indicated with S. cerevisiae numbering below the alignment in orange and pink, respectively; exposed VWM/CACH mutations are indicated with H. 6
sapiens numbering in green; the corresponding C. thermophilum numbers are given in brackets. Mutations of cteIF2Bδ used in this study are indicated by diamonds and the introduced amino‐acid above the alignment. Species abbreviations: ct – Chaetomium thermophilum; sc – Saccharomyces cerevisiae; sp – Schizzosaccharomyces pombe; dd – Dictyostelium dyscoideum; oc – Oryctolagus cuniculus; hs – Homo sapiens; bt – Bos taurus; mm – Mus musculus; rn – Rattus norvegicus.
7
Figure S3. Multiple sequence alignment of eIF2Bα orthologs. Highly conserved residues are highlighted in dark blue, conserved residues in light blue and variable residues in gray; the numbering above the alignment corresponds to the H. sapiens ortholog (hseIF2Bα). Helices α5 and α6 with the kink region, as well as the C‐terminal helix α12 are indicated as magenta cylinders. Surface exposed Gcd‐ and Gcn‐ mutations are indicated with S. cerevisiae numbering below the alignment in orange and pink; the corresponding H. sapiens numbers are given in brackets. Exposed VWM/CACH mutations are indicated with H. sapiens numbering in green. Mutations of cteIF2Bα used in this study are indicated by diamonds 8
and the introduced amino‐acid above the alignment. Species abbreviations: hs – Homo sapiens; sc – Saccharomyces cerevisiae; ct – Chaetomium thermophilum; sp – Schizzosaccharomyces pombe; ce – Caenorhabditis elegans; dd – Dictyostelium dyscoideum; dm – Drosophila melanogster; pa ‐ Pongo abelii; mf – Macaca fascicularis; bt – Bos taurus; rn – Rattus norvegicus; mm – Mus musculus.
9
Figure S4. SEC‐MALS and pull‐down analysis of interactions between eIF2Bα, eIF2Bβ and/or eIF2Bδ. A) SEC run for eIF2Bδ130‐C lacking the poorly conserved, presumably flexible residues 1‐129. The protein elutes in a single peak at 15.1 ml (compared to 13.1 ml for full‐length eIF2Bδ); the molar mass distribution as calculated by MALS, indicated by a black line inside the peak, yields a MW of 43.6 kDa in good agreement with the expected mass of 37 kDa for the monomer. The peak eluting around 13.9 ml corresponds to the GST dimer (~52 kDa) which was not completely removed during purification. SDS‐ PAGE analysis of the peak fraction is shown as inset. B) SEC experiment with eIF2Bα and eIF2Bβ. The two subunits elute as observed for the individual proteins (Figure 3A/B), with eIF2Bα eluting as a dimer at 13.1 ml and eIF2Bβ eluting as a monomer at 14.8 ml. No complex formation is observed. C) As for eIF2Bβ, no complex formation is observed between eIF2Bα and eIF2Bδ. Both proteins elute at virtually identical elution volumes, giving rise to one single elution peak at 13.1 ml as expected from their respective individual runs (Figure 3A/C). D) SEC‐MALS run for eIF2Bα with eIF2BβΔ123‐148 and eIF2Bδ. The hexameric complex is formed despite the deletion in eIF2BβΔ123‐148. The molar mass distribution as calculated by MALS yields a MW of 251 kDa in good agreement with the mass of 256 kDa observed for the wild‐type complex (Figure 3F). E) GST pull‐downs between eIF2Bα, ‐β, and ‐δ. Lanes 1 to 4 show the isolated proteins as reference (in lane 1 the two bands directly above the 55 kDa marker are GST‐ containing degradation products of GST‐eIF2Bδ). Lanes 5 to 7 show the elution fractions (after the removal of unbound proteins by extensive washing) for the pull‐down experiments of GST‐tagged eIF2Bδ with full‐length eIF2Bβ (Lane 5), with the deletion construct eIF2BβΔ123‐148 (6), and with eIF2Bα alone (7). Both eIF2Bβ constructs bind stably to eIF2Bδ. By contrast, as for the shorter eIF2Bδ130‐C construct (Figure 5A, lane 10), full‐length eIF2Bδ does not form a stable complex with eIF2Bα. Lanes 8 and 9 show the elution fractions for the pull‐downs of GST‐tagged eIF2Bδ with eIF2Bα and full‐length eIF2Bβ (Lane 8) or with eIF2Bα and the deletion construct eIF2BβΔ123‐148 (9). In both cases, a stable complex containing all three subunits is formed.
10
Figure S5. Reconstruction of the model of the eIF2BRSC. A) Within the crystal structure of the eIF2B(βδ)2 complex, the dimer‐dimer interface is formed by three main interfaces (areas I‐III). Area I (orange surface) in eIF2Bβ associates with area III (cyan surface) in eIF2Bδ of the second dimer, while area II 11
(light brown) of eIF2Bβ interacts with the same interface in eIF2Bβ of the second dimer (indicated by arrows). Residues altered by Gcn‐ mutations are colored in pink. B) Association of the two canonical eIF2Bβδ dimers as indicated in A results in the formation of the eIF2B(βδ)2 tetramer as observed in the crystal structure (left), which provides a 2‐fold rotation symmetrical surface area for the likewise 2‐fold rotation symmetrical surfaces of eIF2Bα2 (middle). The homologous (but non‐identical) binding surfaces (areas I‐III) on the three subunits are colored as described in A. The right panel shows the top view onto the eIF2B(βδ)2 complex. C) Association of the eIF2Bα2 dimer to eIF2B(βδ)2 results in the eIF2Bα2(βδ)2 hexamer (eIF2Bα2 is rotated relative to its orientation in B by 180°). In this complex, each area I of eIF2Bα2 is bound by area III of one of the two eIF2Bβ subunits, while areas II and III of eIF2Bα2 associate with areas II and I of the two eIF2Bδ subunits, respectively. The right panel shows the top view onto the eIF2Bα2(βδ)2 complex. D) Cartoon presentation of a tkRBPI tetramer, composed of two canonical homodimers. The orientation is the same as that of the eIF2B(βδ)2 complex in Figure 4A.
12
Figure S6. Ligand Binding by cteIF2B regulatory subunits. A‐E) Thermal shift analysis of the effect of the indicated ligands on the thermal stability of cteIF2Bα (A & B), cteIF2BαS270Y (C), cteIF2Bβ (D), and cteIF2Bδ (E). The normalized relative fluorescence is plotted against the temperature (in Kelvin). F) ITC experiments for the binding of various ligands by wild‐type cteIF2Bα. cteIF2Bα binds tightly to GMP (black squares) and AMP (red triangles) and moderately to ribose 5‐phosphate (R5P; purple circle) and CMP (blue diamonds). Although phosphate ions (PO42‐; green circles) and glucose 6‐phosphate (G6P) show a weak binding signal, thermodynamic parameters could not be determined. No binding signals were observed for the other ligands including ribulose 1,5‐bisphosphate (RuBP), dihydroxyacetone phosphate (DHAP), NADP+, GDP and GTP.
13
Figure S7. A) Global conformational changes in the tkRBPI hexamer upon the transition from the open apo state (top; PDB 3A11) to the closed substrate‐bound state (bottom; PDB 3VM6). B) Comparison between the inter‐domain cavities in tkRBPI (PDB 3VM6), hseIF2Bα (PDB 3ECS), cteIF2Bβ, and cteIF2Bδ. The cavity in tkRBPI is shown in its closed state with the substrate ribose 1,5‐bisphosphate (R15BP) bound by the catalytic site residues (shown as balls and sticks in pink). In the structure of hseIF2Bα, a sulfate ion occupies the position of the 5’ phosphate of R15BP in tkRBPI. For eIF2Bα, ‐β, and ‐δ Gcd‐ (orange sticks) and VWM/CACH mutations (green) are indicated with the numbering for the S. cerevisiae and human orthologs, respectively. The right panel shows the degree of evolutionary conservation of residues in and around the cavity.
14
Figure S8. Surface conservation and locations of Gcn‐, Gcd‐, and VWM/CACH mutations in the eIF2Bα2 homodimer. A) Cartoon presentation of the canonical hseIF2Bα2 dimer (PDB 3ECS) with the two monomers in light and dark gray. The upper panel shows the view on the backside of the dimer (facing the two α12 helices); the lower panel shows the opposite frontal face with the cavity between N‐ and CTD. B) Conservation of surface residues on eIF2Bα. The highest degree of sequence conservation is found on the backside of the dimer around helix α12 which forms the proposed interface for hexamerization (Figure 6A/C). With exception of the highly conserved inter‐domain cavity, the sequence conservation is significantly lower on the front side which is not involved in interactions with eIF2Bβδ dimers or eIF2(α‐P) (see also Figure S7B). C) Positions of surface exposed Gcd‐ (orange) and VWM/CACH mutations (green) in eIF2Bα2. Gcd‐ mutations are numbered according to the yeast eIF2Bα ortholog with the corresponding residues of the human ortholog given in brackets. Y86 itself does not represent the position of a Gcd‐ mutation but is preceded by a flexible loop not resolved in the crystal structure that contains the Gcd‐ mutation H82Y (L81). D) Positions of Gcn‐ mutations in eIF2Bα2. Numbers in pink correspond to the yeast ortholog; black numbers in brackets give the corresponding residues in human eIF2Bα. Orientations in B‐D are the same as in A. 15
‐
‐
Figure S9. Surface conservation and locations of Gcn , Gcd , and VWM/CACH mutations in the eIF2Bβδ heterodimer. A) Cartoon presentation of the canonical cteIF2Bβδ heterodimer, with eIF2Bβ in yellow and eIF2Bδ in blue. The upper panel shows the view on the backside of the dimer (facing α12); the lower panel shows the opposite frontal face with the inter‐domain cavities. B) Conservation of surface residues in eIF2Bβδ. Like in eIF2Bα2 (see Figure S8B) the highest degree of sequence conservation is found on the backside of the dimer around helix α12 which forms the interface for hexamerization (Figure 6A/C). In addition to the interdomain cavity, eIF2Bδ also shows a high degree of sequence conservation in the frontal face of the NTD (lower panel) which is not found in eIF2Bα or ‐β and may be involved in interactions with the eIF2Bγε catalytic subcomplex (dotted circle; see also lower panel in C and Figure 6C). C) Positions of surface exposed Gcd‐ (orange; yeast numbering) and VWM/CACH mutations (green; human numbering) in eIF2Bβδ; the corresponding residues in the structures of the C. thermophilum orthologs are given in brackets. D) Positions of Gcn‐ mutations in eIF2Bβδ. Numbers in pink correspond to the yeast orthologs; black numbers in brackets give the corresponding residues in C. thermophilum eIF2Bβ and ‐δ. The orientations in B‐D are the same as in A. 16
Supplementary Literature 1. Mueller, U., Darowski, N., Fuchs, M.R., Forster, R., Hellmig, M., Paithankar, K.S., Puhringer, S., Steffien, M., Zocher, G. and Weiss, M.S. (2012) Facilities for macromolecular crystallography at the Helmholtz‐Zentrum Berlin. J Synchrotron Radiat, 19, 442‐449. 2. McCoy, A.J., Grosse‐Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C. and Read, R.J. (2007) Phaser crystallographic software. J Appl Crystallogr, 40, 658‐674. 3. Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse‐Kunstleve, R.W. et al. (2010) PHENIX: a comprehensive Python‐based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr, 66, 213‐221. 4. Emsley, P., Lohkamp, B., Scott, W.G. and Cowtan, K. (2010) Features and development of Coot. Acta Crystallogr D Biol Crystallogr, 66, 486‐501.
17