Biochemical Journal

Research article

Structural analysis of the Sil1–Bip complex reveals the mechanism for Sil1 to function as a nucleotide-exchange factor

Ming Yan , Jingzhi Li , Bingdong Sha

Abstract

Sil1 functions as a NEF (nucleotide-exchange factor) for the ER (endoplasmic reticulum) Hsp70 (heat-shock protein of 70 kDa) Bip in eukaryotic cells. Sil1 may catalyse the ADP release from Bip by interacting directly with the ATPase domain of Bip. In the present study we show the complex crystal structure of the yeast Bip and the NEF Sil1 at the resolution of 2.3 Å (1 Å=0.1 nm). In the Sil1–Bip complex structure, the Sil1 molecule acts as a ‘clamp’ which binds lobe IIb of the Bip ATPase domain. The binding of Sil1 causes the rotation of lobe IIb ~ 13.5° away from the ADP-binding pocket. The complex formation also induces lobe Ib to swing in the opposite direction by ~ 3.7°. These conformational changes open up the nucleotide-binding pocket in the Bip ATPase domain and disrupt the hydrogen bonds between Bip and bound ADP, which may catalyse ADP release. Mutation of the Sil1 residues involved in binding the Bip ATPase domain compromise the binding affinity of Sil1 to Bip, and these Sil1 mutants also abolish the ability to stimulate the ATPase activity of Bip.

  • Bip
  • crystal structure
  • heat-shock protein of 70 kDa (Hsp70)
  • nucleotide-exchange factor (NEF)
  • Sil1

INTRODUCTION

In eukaryotic cells, the ER (endoplasmic reticulum) is one of the major organelles where nascent protein folding occurs [14]. The ER contains a large number of proteins, which include secreted proteins, membrane proteins and glycosylated proteins [5]. As a result, the protein concentration within the ER is exceptionally high (~ 100 mg/ml). An extensive protein-folding network is required for proper folding in the ER lumen [6,7]. Molecular chaperones such as Bip, J-proteins, Sil1, Grp94 and Grp170 collectively function to facilitate protein folding, translocation and degradation. Mutations in this protein-folding network may cause severe genetic diseases in animals, such as diabetes, developmental disorders and neurodegenerative diseases [8].

Bip, an ER-resident luminal Hsp70 (heat-shock protein of 70 kDa), plays central roles in the molecular chaperone machinery [5,9]. Bip can bind the unfolded polypeptide immediately after it enters the lumen during translocation, to prevent possible misfolding and promote protein folding [1013]. All Hsp70s contain an N-terminal ATPase domain and a C-terminal peptide-binding domain. The structures of the cytosolic Hsc70 (heat-shock cognate 70 stress protein)/Hsp70 have been extensively studied [1418]. The N-terminal ATPase domain of Hsp70 contains two structurally similar lobes (I and II). A deep cleft is formed between the two lobes, which is the binding site for the nucleotide. Hydrolysis of ATP causes a conformational change that is transmitted to the C-terminal peptide-binding domain, which modulates the affinity of Hsp70 for unfolded substrates [6,7,19].

The intrinsic ATPase activity of Hsp70 is low. Two types of co-chaperones, Hsp40 and NEF (nucleotide-exchange factor), are required to work with Hsp70 to fold polypeptides efficiently. The J-domain of Hsp40 can stimulate the ATP hydrolysis of Hsp70 [20,21]. The NEFs for Hsp70 can bind the ATPase domain of Hsp70 to facilitate ADP release after ATP hydrolysis, which could promote the ATPase activity of Hsp70 [7]. Rebinding of ATP and the concomitant dissociation of the NEF completes the exchange reaction.

Three types of NEFs [BAG domain proteins, HspBP1 (Hsp-binding protein 1) and Hsp110] have been identified in the cytosol of eukaryotes [2224]. Hsp110, an Hsp70 homologue, has been shown to have nucleotide exchange activity [25,26]. The bacterial NEF GrpE has also been identified [27]. A number of crystal structures of the Hsp70 and NEF complexes have been determined [17,18,2224,2831]. Several NEFs, including GrpE, BAG domain proteins and Hsp110, can trigger a conformational change in lobe II of the Hsp70 ATPase domain to open up the cleft between lobes I and II, although the NEFs themselves contain distinct molecular structures. However, structural and biophysical data suggested that HspBP1 may be able to displace lobe I of the Hsp70 ATPase domain to reduce the affinity to the nucleotide [23].

Sil1 has been identified as an ER NEF which functions co-operatively with the ER Hsp70 Bip [3234]. Grp170 (Lhs1 in yeast) can function as another ER NEF, possibly using a mechanism distinct from Sil1 [32,35]. It has been suggested that Sil1 represents a paralogue of HspBP1 [23,34,36]. The sequence identity and sequence similarity between yeast Sil1 and human HspBP1 NEF regions are 13% and 30% respectively. Sil1 has also been revealed as an important component in ERAD (ER-associated degradation) machinery and the unfolded protein response [8,37]. In the mouse model, disruption of Sil1 causes accumulation of misfolded proteins in the ER and nucleus, which leads to neurodegeneration [38]. In humans, mutations in Sil1 have been found in individuals with Marinesco–Sjögren syndrome, an autosomal recessive cerebellar ataxia complicated by cataracts, developmental delay and myopathy [3942]. Yeast Sil1 contains 421 amino acid residues with a signal peptide at the N-terminus. Human Sil1 (461 amino acids, also named BAP) shares 20% sequence identity with yeast Sil1 [34]. It is not clear how Sil1 interacts with the Bip ATPase domain to facilitate nucleotide release.

MATERIALS AND METHODS

Protein expression, purification and crystallization

The gene encoding Saccharomyces cerevisiae Sil120–421, Sil1C (Sil1113–421) and the Bip ATPase domain (amino acids 43–426) were cloned into the vector pET28b for protein expression in Escherichia coli. The purified recombinant Sil1C and Bip were mixed using the molar ratio 2:1 and the Sil1–Bip complex was further purified using a Superdex-200 column (GE Healthcare). The complex was concentrated to 20 mg/ml in 20 mM Hepes buffer (pH 7.0), containing 150 mM NaCl and 1 mM MgCl2 and subjected to crystallization trials. Long needle-shaped crystals (0.5 mm×0.1 mm×0.1mm) were obtained by the hanging-drop vapor-diffusion method using Linbro plates at room temperature (22 °C). The well solution consisted of 1 ml of 100 mM Hepes buffer (pH 7.0) containing 25% (w/v) PEG [poly(ethylene glycol)] 4000 and 0.2 M ammonium sulfate. The Se-Met (L-selenomethionine) Sil1C–Bip complex crystals were grown in similar conditions to the native protein complex.

Structure determinations

The native Sil1–Bip crystals diffracted X-rays to 2.3 Å (1 Å=0.1 nm) in the beamline SER-CAT (Southeast Regional Collaborative Access Team) at APS (Advanced Photo Source). The SAD (systems analysis and design) data using Se-Met Sil1C–Bip crystals were collected in the beamline A1 at CHESS (Cornell High Energy Synchrotron Source). The crystals were flash frozen at 100 K in a N2 gas stream in the cryoprotectant consisting of 100 mM Hepes buffer (pH 7.0) containing 24% (w/v) PEG 4000, 0.2 M ammonium sulfate and 20% (v/v) glycerol. The Se-Met Sil1–Bip crystals diffracted X-ray to 3.1 Å. The data set revealed that the crystals belong to space group of P21212 with unit cell parameters of a=226.353 Å, b=116.586 Å, c=55.844 Å, α=90.00°, β=90.00° and γ=90.00°. The Se-Met Sil1–Bip crystals are very sensitive to X-ray radiation and only one wavelength data set can be collected using one crystal. Therefore the SAD method was utilized to carry out the structure determination using SOLVE [43]. From the possible 22 Se atoms in the primary sequence, 14 Se atoms can be identified. The phases from the SAD phasing were further improved by use of the program RESOLVE [44]. The resultant electron density map was readily traceable. The COOT program was utilized to model the yeast Sil1–Bip into the electron density map [45].

The subsequent structure refinement was performed by using the CNS program against the native data set of 2.3 Å resolution [46]. A Ramachandran plot of the final model using the Probity program (http://kinemage.biochem.duke.edu) revealed that 99.2% of the residues were in allowed regions. The outliers in the Ramachandran plot have been carefully corrected.

The crystal structure of the yeast Bip ATPase domain free of nucleotide or in complex with ADP were determined by using the molecular replacement method to the resolutions of 2.2 Å or 2.3 Å respectively (Supplementary Figure S1 and Table S1 at http://www.BiochemJ.org/bj/438/bj4380447add.htm). The co-ordinates from the human Bip ATPase domain (PDB code 3IUC) were utilized as the searching model. Program Phaser was utilized for the model search. The structure refinement was carried out using Refmac5.

Sil1 mutagenesis

The Sil1 mutations were generated by using the QuikChange® kit from Invitrogen.

The affinity measurements between Sil1 and the Bip ATPase domain

Measurement of the binding affinities between Sil1 and Bip ATPase domain was carried out by use of an isothermal titration calorimeter (MicroCal) at room temperature. Sil1 (or its mutants) and Bip were dialysed against the same buffer (10 mM Hepes, pH 7.0, 150 mM NaCl and 1 mM 2-mercaptoethanol). Bip (or its mutant) was added to the calorimetric cell and Sil1 (or its mutants) was injected into the cell using a 250 μl injection syringe. The heat released was obtained by integrating the calorimetric output curves. Pure buffers were injected into the Bip protein as control experiments. The heat releases from the control experiments were subtracted from the experimental data before the data were utilized for Kd fitting. The Kd values and the binding ratios were calculated using the software supplied with the calorimeter.

ATPase activity assay

Wild-type Sil1 (or its mutants) and the Bip ATPase domain were incubated at 30 °C in the assay buffer (20 mM Hepes buffer, pH 7.0, containing 2 mM MgCl2, 25 mM KCl and 2 mM ATP). A total of 1 μM protein was used for each reaction. The ATPase activity of Bip was measured using the Malachite Green assay as described previously [47].

RESULTS AND DISCUSSION

Crystal structure of the Sil1–Bip ATPase domain complex

The S. cerevisiae Sil120–421 and Bip ATPase domain (amino acids 43–426) without the N-terminal targeting signals were expressed in E. coli and purified to homogeneity. Full-length Sil1 appears to form a dimer in solution as shown in the gel filtration profiles (Figure 1A). The direct interactions between the purified Sil1 and Bip ATPase domain can be measured using ITC (isothermal titration calorimetry) (Figure 1B). The ITC data fitting showed tight binding between Sil1 and the Bip ATPase domain with a Kd of 13 nM. The data fitting also indicated that a Sil1 dimer binds a Bip ATPase domain monomer. Full-length yeast Sil120–421 appeared to be unstable in solution. It degraded to a smaller stable fragment within 2 weeks at 4 °C (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/438/bj4380447add.htm). The N-terminal sequencing and MS analysis indicated that the smaller fragment of Sil1 represented Sil1C. Interestingly, full-length Sil1 forms a dimer in solution, whereas Sil1C migrates as a monomer, as shown in the gel-filtration profiles (Figure 1A), which suggests that the N-terminal fragment of Sil1 may be responsible for the dimer formation of Sil1.

Figure 1 Sil1C may retain the ability to interact with and stimulate the ATPase activity of the Bip ATPase domain

(A) The gel-filtration profiles of full-length Sil1 [Sil1(20–421)] and Sil1C [Sil1(113–421)] when loaded on to a Superdex-200 column (GE Healthcare). The filled triangles indicate the eluted Sil120–421 and Sil1C peaks. The eluted peak positions of the molecular mass standards are shown and the molecular masses are marked in kDa. (B) The ITC data showing that Sil120–421 (left-hand panel) and Sil1C (right-hand panel) can interact with the Bip ATPase domain. The data fitting showed that Sil120–421 and Sil1C bound the Bip ATPase domain with Kd values of 13 nM and 341 nM respectively. (C) Experiments using the ATPase activity assay indicated that both Sil120–421 and Sil1C can stimulate the ATPase activity of the Bip ATPase domain.

Sil1C can bind the Bip ATPase domain with a lower affinity (Kd≈341 nM) compared with the full-length Sil1, as indicated by ITC studies (Figure 1B). The stoichiometry for the Sil1C monomer is to interact with the Bip ATPase domain monomer at a 1:1 ratio. The ATPase activity of Bip can be measured using the Malachite Green assay [47]. The data from the ATPase activity assay suggested that Sil1C might be equivalent to full-length Sil1 in stimulating the Bip ATPase activity (Figure 1C). It is likely that Sil1C retains the co-chaperone activity of the full-length Sil1. In the present paper, we introduce the term ‘Sil1C’ to refer to Sil1113–421.

The crystal structure of the protein complex of Sil1C and the Bip ATPase domain was determined to 2.3 Å resolution using the SAD method (Table 1). The final model of the complex structure contains residues 125–406 of Sil1 and residues 47–426 of Bip. In the Sil1C–Bip ATPase domain structure, one Sil1 molecule interacts with one Bip molecule, which is consistent with the ITC data.

View this table:
Table 1 Data collection, phasing and refinement statistics for the Sil1–Bip complex structure

The highest resolution shell is shown in parentheses.

The Sil1C molecule forms a monomer in the crystal structure. The Sil1C monomer has an elongated molecular shape which consists of 16 α-helices (A1–A16) and no β-strands (Figures 2A and 2B). The central helices A3–A14 form four Armadillo-like repeats (ARM1–ARM4). The ARM repeats are constructed from three α-helices and pack into a right-handed superhelix to produce a gently curved elongated molecule. In the Sil1–Bip complex structure, the Sil1C molecule acts as a ‘clamp’ that wraps around the middle region of lobe IIb of the Bip ATPase domain for ~ 180° (Figure 2A). The Sil1C molecule utilizes its concaved inner surface to embrace lobe IIb of Bip. The N-terminus of Sil1C interacts with the inner side of lobe IIb that is close to lobe I, whereas the C-terminus of Sil1C is positioned on the opposite side of lobe IIb (Figure 2A). Recent biochemical and genetic studies suggested that Sil1 interacted with lobe IIb of the ATPase domain [32], which is consistent with our structural data.

Figure 2 The Sil1–Bip ATPase domain complex structure

(A) The Sil1–Bip ATPase domain complex structure. The Sil1C structure is shown in ribbons and the Bip ATPase domain is shown in surface drawing. The N-terminus and C-terminus of Sil1C are labelled, as are lobes Ia, Ib, IIa and IIb of Bip. (B) The Sil1C structure shown in ribbons. The helices A1–A16 are labelled, as are the N-terminus and C-terminus and the ARM repeats of Sil1C. The flexible regions are shown by broken lines. The orientation of Sil1C in this Figure is generated by rotating Sil1C in (A) by ~ 90° along the horizontal axis to show the concaved interface of Sil1. (C) Structural comparison between the Sil1–Bip complex and the HspBP1–Hsp70 complex. The Sil1C structure (blue) is superimposed with the HspBp1 structure (magenta). The Sil1-bound Bip1 lobe IIb fragment is in green ribbons and the HspBP1-bound Hsp70 lobe IIb fragment is in red ribbons. (D) The surface areas on lobe IIb of Bip and Hsp70 that are involved in binding Sil1 and HspBP1. Lobe IIb of Bip is positioned as the same orientation as that from Hsp70. The surface of Bip that is involved in binding Sil1 is shown in red and the surface of Hsp70 that interacts with HspBP1 is in blue. The orientation of Bip lobe IIb in this Figure is similar as that in the right-hand panel of (A).

Sil1 has been suggested to be a homologue of HspBP1, which catalyses nucleotide exchange for Hsp70 [23,34,36]. Both the Sil1 structure and HspBP1 structure contained four central ARM repeats flanked by two N-terminal and C-terminal helices [23]. Sil1 and HspBP1 share limited sequence identity (13%) between these two NEFs (see Supplementary Figure S3 at http://www.BiochemJ.org/bj/438/bj4380447add.htm) for Hsp70 and they fold into similar protein structures. The superimposition of Sil1C and HspBP1 indicated that the RMSD (root mean square deviation) for the co-ordinates of the main-chain atoms of Sil1C and HspBP1 is 4.0 Å (Figure 2C and Supplementary Movie S1 at http://www.BiochemJ.org/bj/438/bj4380447add.htm). In the HspBP1–Hsp70 ATPase domain complex structure, lobe I of the Hsp70 ATPase domain has been removed for crystallization purposes [23]. In the Sil1–Bip complex, both lobe I and II of the Bip ATPase domain are present in the crystal structure. Both Sil1C and HspBP1 utilize the concave surface to interact with Hsp70 at the lobe IIb subdomain. However, the mechanism of how Sil1 binds Bip is significantly different from how HspBP1 interacts with Hsp70. When Sil1C is superimposed on HspBP1 for the two complex structures, the Sil1-bound Bip rotates ~ 36° away from the HspBP1-bound Hsp70. Moreover, the Bip molecule translates ~ 8.7Å from Hsp70 (Figure 2C). The molecular surface of Bip lobe II that is involved in binding Sil1 is significantly different from that of Hsp70 lobe II for HspBP1 binding (Figure 2D).

The interactions between Sil1 and the Bip ATPase domain

In the complex structure, Sil1C binds the lobe IIb region of the Bip ATPase domain primarily via extensive charge–charge interactions and numerous hydrogen bond formations. The contact areas between Sil1 and the Bip ATPase domain bury 2963.02 Å2 of accessible surface area, which accounts of 10.6% of the Sil1C surface area and 8.5% of the Bip ATPase domain surface area. A major interaction site and a minor interaction site are present in the complex structure for Sil1C to bind the Bip ATPase domain (Figure 3A). The N-terminal helices A2, A5, A8 and A11 of Sil1C interact with the inner surface and the top of lobe IIb of the Bip ATPase domain and form the major interaction site, whereas the C-terminal helices A14 and A16 of Sil1C bind the opposite side of lobe IIb that constitutes the minor interaction site. SilC interacts with Bip by utilizing surfaces that are significantly different from that of HspBP1 for Hsp70 binding (Figure 3A).

Figure 3 The interactions between Sil1C and the Bip ATPase domain

(A) The upper panel shows the interaction surfaces of Sil1C for the Bip ATPase domain. The major interaction site is in red and the minor interaction site is in green. The residues constituting the interaction sites are labelled in single letter code. The orientation of the Sil1C molecule in this panel is similar to that in Figure 2(B). The lower panel shows the interaction surface of HspBP1 for Hsp70 binding. The HspBP1 molecule is positioned in the similar orientation as that for SilC in the upper panel. The residues within HspBP1 that are involved in binding Hsp70 and conserved between Sil1 and HspBP1 are shown in blue and the residues of HspBP1 that are responsible for binding Hsp70 but not conserved between Sil1 and HspBP1 are in red (see also Supplementary Figure S2 at http://www.BiochemJ.org/bj/438/bj4380447add.htm). (B) The interactions between Glu160 and His163 from helix A2 of Sil1C with Bip in the major interaction site. Sil1C is in cyan and Bip is in green. The water molecules involved are shown as dark blue spheres.

At the major interaction site, the side chain of His163 from A2 of Sil1C forms a strong hydrogen bond (2.83 Å in length) with the side chain of Glu311 of Bip. This is consistent with the observation that mutation of Glu311 of Bip abolished the Sil1C and Bip interactions [48]. The side chain of His163 from Sil1C also interacts with the carbonyl oxygen of Asp330 of Bip through an ordered water molecule. Glu160 from helix A2 of Sil1C generates a salt bridge with Arg310 of Bip. Arg205 from helix A5 of Sil1C forms strong charge–charge interactions with Asp330 of Bip (Figure 3B). In short, the Sil1–Bip interface is rather polar in character, containing numerous salt bridges (Sil1–Bip, His163–Glu311, Glu160–Arg310, Arg205–Asp330 and Lys285–Glu328) and hydrogen bonds (Sil1–Bip, Thr201–Asp330, Ser202–Lys314 and Ser249–Asp330) at the major interface.

The conformational changes of Bip caused by Sil1 binding

Sil1 functions as a NEF for the ER-resident Hsp70 Bip. To investigate the mechanisms as to how Sil1 catalyses ADP release from the Bip ATPase domain, we have determined the crystal structure of the Bip ATPase domain with and without the bound ADP to the resolution of 2.2 Å and 2.3 Å respectively (Supplementary Figure S1 and Table S1). Bip interacts with ADP at the nucleotide-binding pocket, which is located between lobe I and II. The binding of ADP causes lobe IIb of Bip to swing open for ~ 7.1°, however, little conformational changes can also be observed for lobe I and IIa after ADP binding (Supplementary Figure S1). These data are consistent with previous structural observations of the Hsp70 ATPase domain [49].

When the Bip ATPase domain structure in the Sil1–Bip complex is compared with the ADP-bound Bip ATPase domain structure, significant conformational changes can be seen to occur in the Bip ATPase domain structure (Figure 4A). In the Sil1–Bip complex structure, lobe IIb of Bip is swung away from the ADP-binding pocket for ~ 13.5° compared with the ADP-bound Bip ATPase domain structure.

Figure 4 Conformational changes in the Bip ATPase domain generated by Sil1 binding

(A) The Bip ATPase domain in the Sil1–Bip complex is superimposed with the ADP-bound Bip ATPase domain. The Bip ATPase domain from the Sil1–Bip complex is in green and the ADP-bound Bip ATPse domain is in red. The bound ADP is shown as a stick drawing. (B) The Sil1–Bip ATPase domain interactions at lobe Ib of Bip. The Bip ATPase domain in the Sil1–Bip complex is superimposed with the ADP-bound Bip ATPase domain structure. The ADP-bound Bip ATPase domain is shown in red ribbons and the Sil1-bound Bip ATPase domain is in green. Sil1C is shown in cyan ribbons. Lobe I and II of Bip are labelled, as are residues Asn103 and Arg310 from Bip and residues His167 and Glu160 from Sil1C. The charge–charge interaction between Arg310 and Glu160 in the Sil1–Bip complex is indicated (3.7 Å). (C) Sil1C binding generates dislocations of several residues involved in binding the adenine and the ribose ring of the ADP. The ADP-bound Bip ATPase domain is in cyan and the Bip ATPase domain from the Sil1–Bip complex is in green. Bound ADP is shown as a stick drawing. (C) was generated by rotating the portion around the dotted area in (A) by ~ 60° along the horizontal axis.

Interestingly, Sil1C binding to Bip also causes lobe Ib to rotate ~ 3.7° away from the ADP-binding pocket. Sil1C makes contacts with Bip at the lobe Ib region. The side chain of His167 from A3 of Sil1C forms a hydrogen bond with that of Asn103 of Bip (Figure 4B). In the Sil1-free Bip ATPase domain structure, the carbonyl oxygen of Asn103 from lobe Ib of Bip forms a hydrogen bond with the side chain of Arg310 from lobe IIb of Bip and this interaction may help to associate lobe Ib and IIb together. The Sil1C binding breaks this hydrogen bond and pushes lobe Ib and IIb away from each other (Figure 4B). His167 of Sil1C forms a hydrogen bond with Asn103 of Bip, and Glu160 from Sil1C establishes a charge–charge interaction with Arg310 from lobe IIb of Bip. The contacts between Sil1C and lobe Ib of Bip may play important roles to swing lobe Ib away from the nucleotide-binding pocket by ~ 3.7°.

The complex formation of Sil1 and the Bip ATPase domain causes the movements of lobe Ib and lobe IIb to the opposite directions, which significantly opens up the ADP-binding pocket located between lobes I and II in the Bip ATPase domain. This conformational change results in the dislocations of ~ 3 Å for the residues located on lobe IIb that are involved in binding ADP (Figure 4C). In the Bip–ADP complex structure, the adenine group is stabilized by forming strong hydrogen bonds with Ser320 from lobe IIb of Bip. The hydroxy groups from the ribose ring constitute hydrogen bonds with Glu313 and Lys316 from lobe IIb of Bip (Figure 4c). Sil1 binding to the Bip ATPase domain generates the dislocations of ~ 3 Å for the residues involved in binding ADP. These conformational changes may disrupt those above-mentioned hydrogen bonds between ADP and Glu313, Lys316 and Ser320 from Bip, which may play important roles in catalysing ADP release from Bip.

The working mechanisms for Sil1 to function as a NEF for Bip

Sil1 has been identified as a NEF that promotes nucleotide release from Bip. On the basis of our crystal structure of the Sil1–Bip complex, we propose that Sil1 binds lobe IIb of the Bip ATPase domain as a ‘clamp’. Sil1 also makes contacts with lobe Ib of Bip, which may provide a pivot point for the clamp to open up the nucleotide-binding pocket of Bip (Figure 5A). The Sil1–Bip complex formation generates conformational changes to swing lobe Ib and IIb away from the nucleotide-binding pocket for ~ 3.7° and 13.5° respectively. These conformational changes abolish the hydrogen bonds between ADP and the residues from lobe IIb of Bip, which will subsequently release ADP from Bip. Sil1 binding causes little conformational changes in lobe Ia and IIa.

Figure 5 Sil1 functions as a nucleotide-exchange factor

(A) The proposed working model for Sil1 to function as a novel NEF for Bip. (B) The sequence alignment among Sil1 family members. The Sil1 members include those from S. cerevisiae (Sc), Homo sapiens (Hs), Mus musculus (Mm), Xenopus laevis (Xl), Drosophila melanogaster (Dm) and Candida albicans (Ca). The conserved residues are shaded.

Three types of cytosolic NEFs (BAG-1, HspBP1 and Hsp110) have been identified in the eukaryotic cells. In the crystal structure of BAG-1 complexed with the Hsp70 ATPase domain, the BAG-1 structure contains three helices and interacts with Hsp70 from the top of lobe IIb [22,24,28,30]. Hsp110 proteins are homologous with Hsp70 and consist of a nucleotide-binding domain, a β-sandwich domain, and a three helix bundle domain. Hsp110 utilizes the nucleotide-binding domain and the three helix bundle domain to interact with Hsp70 [17,18,50]. In comparison, the Sil1C structure contains four ARM motifs. BAG-1 and Hsp110 binding keeps Hsp70 in an ‘open’ conformation where lobe IIb is swung away by ~ 14° and ~ 27° respectively. Sil1C and Bip complex formation pushes IIb away ~ 13.5° from the nucleotide-binding pocket to release ADP. Sil1C and HspBP1 share similar protein folds and interact with Hsp70 at the lobe IIb region. It is likely that Sil1 and HspBP1 may utilize an evolutionarily conserved mechanism to catalyse nucleotide release from Hsp70 by interacting with the lobe IIb regions of Hsp70. Sil1C and HspBP1 use distinct surface areas to bind Hsp70 (Figures 2C and 2D). Moreover, it was hypothesized that HspBP1 binding may cause major conformational changes in lobe Ib of Hsp70, whereas Sil1C binding generates a modest rotation of 3.7° for lobe Ib of Bip [23]. It has been reported that lobe I of the ATPase domain of Hsp70 is progressively degraded by protease in the presence of HspBP1, but not in the presence of BAG [23]. Our results from the present study showed that lobe I of the Bip ATPase domain complexed with Sil1C is quite stable in the presence of protease (results not shown).

It is interesting to note that the full-length Sil1 dimer can interact with one Bip molecule with high affinity (Kd≈13 nM), whereas the Sil1C monomer binds one Bip molecule with relatively lower affinity (Kd≈341 nM). It is possible that, within the full-length Sil1 dimer, one Sil1 monomer binding to Bip may cause allosteric conformational changes to prevent the other Sil1 monomer binding to Bip. The Sil1 N-terminal domain (residues 21–112) may also contribute to the interaction with the Bip ATPase domain, which explains the fact that full-length Sil1 exhibits a tight binding affinity to Bip.

Structure-based mutagenesis of Sil1

To investigate the significance of the interactions between Sil1 and Bip, we have performed structure-based mutagenesis for Sil1. The residues within Sil1C that are critically involved in binding the Bip ATPase domain were mutated. Seven mutants (E160A, H163A, R205A, R246A, K285A, D337A and E390A) were generated. The data clearly showed that mutations of Sil1C at the binding interface severely compromised the affinities between Sil1C and Bip as shown by ITC studies. The mutations of Sil1C also abolished the ability of Sil1C to stimulate the ATPase activity of Bip (results not shown). The mutagenesis data indicated that His163 and Glu390 from Sil1C play critical roles in binding the Bip ATPase domain (Table 2). The mutations of H163A and E390A reduce the affinities between the Sil1C mutants and Bip ATPase domain to undetectable levels by ITC measures. The side chain of His163 of Sil1C forms a strong hydrogen bond with Glu311 of Bip. The side chain of His163 also interacts with the carbonyl oxygen of Asp330 through an ordered water molecule. Glu390 of Sil1C is positioned at the C-terminal helix A16 and forms a hydrogen bond with Ser324 of Bip. This interaction may help to stabilize the otherwise flexible C-terminus of Sil1 and facilitate the complex formation between Sil1 and Bip. His163, Lys285 and Glu390 are highly conserved residues among the Sil1 family members (Figure 5B).

View this table:
Table 2 Mutagenesis studies of Sil1

Full-length Sil1 can interact with the Bip ATPase domain with high affinity (Kd≈13 nM). We have also examined the binding affinities between the full-length Sil1 mutations (E160A and H163A) and the Bip ATPase domain. ITC studies indicate that full-length Sil1 E160A and H163A exhibit much lower affinities to the Bip ATPase domain (Kd≈310 nM for E160A and affinity not detectable for H163A) compared with the wild-type Sil1 (Table 2).

A number of mutations in human Sil1 (BAP) have been found that are linked with Marinesco–Sjögren syndrome [3942]. Most of these mutations cause frame shifts or insertions of stop codons that may disrupt the major part of the human Sil1 structure. However, one of the mis-sense mutations causes a stop codon mutation at Q417X in the C-terminus of human Sil1, which corresponds to T360X in yeast Sil1 (Supplementary Figure S3). The T360X mutation in yeast Sil1 will cause the deletion of C-terminal helices A15 and A16. It is likely that such a truncated Sil1 protein would expose considerable hydrophobic surfaces and hence might be aggregation-prone and unstable. Therefore the Q417X mutation in human Sil1 may also disrupt the association between human Sil1 and Bip by removing the C-terminal helices A15 and A16 and may cause Marinesco–Sjögren syndrome.

To conclude, our structural studies suggested that fungal ER lumenal Sil1 and human cytosolic HspBP1 have a similar structural scaffold, and a series of ARM repeats similar to many other α-solenoid proteins in widely different functional contexts. However, the key residues for the interactions with Hsp70 as well as the respective binding sites on Hsp70 are markedly distinct, leading to different mechanisms for catalysing nucleotide exchange. Mutliple NEFs of Hsp70 have been identified including GrpE, BAG domain proteins, Hsp110, HspBP1 and Sil1. With the exception of HspBP1, all of these Hsp70 NEFs employ virtually the same nucleotide-exchange mechanism, which appears in-built into the structure of all Hsp70 ATPase domains, a flexible mechanical hinge at the linker to subdomain IIb.

AUTHOR CONTRIBUTION

Ming Yan contributed to protein expression, protein crystallization, structure determination and mutagenesis studies. Jingzhi Li contributed to data collection and structure determination. Bingdong Sha contributed to experimental design, structure determination and paper preparation.

FUNDING

This work was supported by the National Institutes of Health [grant number R01 GM65959] and the Army Research Office [grant number 51894LS].

Acknowledgments

We are grateful to the staff scientists at APS beamline SER-CAT and CHESS for their help in data collection.

Footnotes

  • The structural co-ordinates reported for the Sil1–Bip complex, Bip ATPase domain and Bip ATPase domain–ADP complex will appear in the Protein Data Bank under accession codes 3QML, 3QFU and 3QFP respectively.

Abbreviations: ER, endoplasmic reticulum; Hsp, heat-shock protein; HspBP1, Hsp-binding protein 1; ITC, isothermal titration calorimetry; NEF, nucleotide-exchange factor; PEG, poly(ethylene glycol); RMSD, root mean square deviation; SAD, systems analysis and design; Se-Met, L-selenomethionine; Sil1C, Sil1113—421

References

View Abstract