- Email: [email protected]
Crystal Structure of the bb′ Domains of the Protein Disulfide Isomerase ERp57
Structure 14, 1331–1339, August 2006 ª2006 Elsevier Ltd All rights reserved
DOI 10.1016/j.str.2006.06.019
Crystal Structure of the bb0 Domains of the Protein Disulfide Isomerase ERp57 Guennadi Kozlov,1,4 Pekka Maattanen,1,4 Joseph D. Schrag,2 Stephanie Pollock,1 Miroslaw Cygler,2 Bhushan Nagar,1 David Y. Thomas,1 and Kalle Gehring1,3,* 1 Biochemistry Department McGill University 3655 Promenade Sir William Osler Montre´al, Que´bec H3G 1Y6 Canada 2 Biotechnology Research Institute National Research Council of Canada 6100 Royalmount Avenue Montre´al, Que´bec H4P 2R2 Canada
Summary The synthesis of proteins in the endoplasmic reticulum (ER) is limited by the rate of correct disulfide bond formation. This process is carried out by protein disulfide isomerases, a family of ER proteins which includes general enzymes such as PDI that recognize unfolded proteins and others that are selective for specific proteins or classes. Using small-angle X-ray scattering and X-ray crystallography, we report the structure of a selective isomerase, ERp57, and its interactions with the lectin chaperone calnexin. Using isothermal titration calorimetry and NMR spectroscopy, we show that the b0 domain of ERp57 binds calnexin with micromolar affinity through a conserved patch of basic residues. Disruption of this binding site by mutagenesis abrogates folding of RNase B in an in vitro assay. The relative positions of the ERp57 catalytic sites and calnexin binding site suggest that activation by calnexin is due to substrate recruitment rather than a direct stimulation of ERp57 oxidoreductase activity. Introduction In order to fold in the oxidative environment of the endoplasmic reticulum (ER), proteins need to acquire the proper arrangement of cysteine disulfide bonds. To meet this challenge, the cell produces a large number of different ER protein disulfide isomerases to catalyze the oxidation and isomerization of disulfides (Ellgaard and Ruddock, 2005; Tu and Weissman, 2004). The best-known and most abundant of these is the primary oxidant PDI. For N-glycosylated proteins, a dedicated pathway exists mediated by the protein disulfide isomerase ERp57, the lectin-like molecular chaperones calnexin (CNX) and calreticulin (CRT), and UDP-glucose: glycoprotein glucosyltransferase (UGGT) (Ellgaard and Frickel, 2003; Helenius and Aebi, 2004; High et al., 2000; Parodi, 2000; Schrag et al., 2003). Following syn-
*Correspondence: [email protected] 3 Lab address: http://www.mcgnmr.ca 4 These authors contributed equally to this work.
thesis and N-glycosylation, the glycan is trimmed by the removal of two glucose residues to generate a Glc1Man9GlcNAc2 glycoprotein that binds to CNX or CRT (Ou et al., 1993; Zapun et al., 1997). Together with ERp57 (Frickel et al., 2002; Pollock et al., 2004; Zapun et al., 1998), this allows for correct formation of the disulfide bonds and further maturation of the protein through glycan processing. A quality control system exists in which UGGT recognizes unfolded Man9GlcNAc2 glycoproteins and reglucosylates them for CNX/CRT binding and further ERp57 processing. This pathway, known as the calnexin cycle, is responsible for the folding of proteins such as viral hemagglutinin, rhodopsin, transferrin, and major histocompatibility complex MHC class I molecules (Ou et al., 1993; Rosenbaum et al., 2006; Solda et al., 2006). Recent evidence suggests that ERp57 may also play a structural role in the MHC class I complex with TAP (transporter associated with antigen processing), where it forms a covalent, disulfidelinked complex with tapasin (Garbi et al., 2006; Peaper et al., 2005). ERp57 (also called ER-60 or GRP58) is composed of four thioredoxin-like domains abb0 a0 with CGHC catalytic motifs in the N- and C-terminal a and a0 domains. ERp57 is highly similar to PDI both in its domain organization and sequence. Similarity is highest in the catalytic a and a0 domains (w50% identity) and lowest in the bb0 domains (20%). ERp57 possesses oxidoreductase activity (Bourdi et al., 1995; Hirano et al., 1995; Srivastava et al., 1993) but is most active when coupled to CNX/ CRT (Zapun et al., 1998). The physical association between ERp57 and CNX/CRT has been shown by crosslinking (Oliver et al., 1997, 1999) and by NMR (Frickel et al., 2002; Pollock et al., 2004), but the mechanism of CNX/CRT stimulation of ERp57 activity is unknown. Recently, the X-ray crystal structure of yeast PDI was reported at 2.4 A˚ (Tian et al., 2006). The structure of PDI shows an overall U shape with the catalytic sites of the two active thioredoxin domains facing each other. The interior surface of the U is hydrophobic and likely contributes to PDI’s ability to bind unfolded proteins and catalyze the formation of correct disulfides. An NMR characterization of the a and a0 domains of ERp57 confirmed that they adopt a thioredoxin-like fold (Pirneskoski et al., 2004; Silvennoinen et al., 2001, 2005). The structure of the luminal domain of CNX has also been determined by X-ray crystallography. The protein is characterized by two main structural components, a globular lectin domain and an extended proline-rich structure, termed the P domain (Schrag et al., 2001). The P domain is composed of four modules which form a hook-like arm. Previous studies using NMR spectroscopy and mutagenesis found that the tip of the P domain of CNX or CRT binds to ERp57 (Frickel et al., 2002; Leach et al., 2002; Pollock et al., 2004). Here we report structural studies of ERp57 using a combination of small-angle X-ray scattering (SAXS), X-ray diffraction, and NMR techniques. We show that the shape of ERp57 in solution is highly similar to the crystal structure of yeast PDI but that the 2.0 A˚ X-ray
Structure 1332
Table 1. Data Collection and Refinement Statistics Data Collection Space group Unit cell (A˚) Resolution range (A˚) Observed reflections Unique reflections Completeness (%) Overall (I/sI)
P21 a = 76.3, b = 62.4, c = 99.3 50–2.0 455,000 121,811 99.4 (96.1) 14.0 (2.6)
Refinement and Model Quality Resolution range (A˚) Rfree, number of reflections Rwork, number of reflections Rmsd bond lengths (A˚) Rmsd bond angles ( ) Average B factors (A˚2), number of atoms Protein Main chain atoms Side chain atoms Water Ramachandran plota % of residues in Most favorable regions Additional allowed regions a
32.7–2.0 0.250 (3,162) 0.194 (59,353) 0.025 1.932
43.1 (6,087) 42.1 (2,776) 43.6 (2,858) 45.7 (453)
93.1 6.9
Laskowski et al. (1993).
structure of the bb0 domains reveals significant differences in the CNX binding site, determined by NMR chemical shift mapping. Mutagenesis and in vitro folding assays demonstrate a requirement for the CNX-ERp57 interaction for efficient glycoprotein folding but not oxidoreductase activity. Results Structure of the bb0 Fragment We crystallized the bb0 fragment of ERp57 and determined its three-dimensional structure by X-ray diffrac-
tion using multiwavelength anomalous diffraction (MAD) phasing (Table 1). The crystals belong to the P21 space group and contain three molecules in the asymmetric unit. The structure was refined to 2.0 A˚ resolution (Figure 1A). The chains were traced from P134 (A135 in chain B) to L365. Five N-terminal residues corresponding to the cloning linker and the last 15 residues of the construct including K366–G376 from the linker following the b0 domain were not present in the electron density maps, indicating that these parts were disordered in the crystal. The structure contains two thioredoxin-like domains connected by a short linker. The b domain extends from A135 to F241, and the b0 domain from P245 to Y364 (Figure 2A). The two domains are very similar and can be superimposed using the backbone atoms of 96 residues with a root-mean-square deviation (rmsd) of 2.8 A˚. Each domain has a central five-stranded b sheet (Figure 1B). The b sheet has the order of b strands b1b3-b2-b4-b5, which are all parallel with the exception of strand b4. The four a helices of each domain are of various lengths and enclose the b sheets with two helices on either side. The a2 helix in the b0 domain (A273– D292) is significantly longer than any other helix in the structure. The long loop between strands b4 and b5 of domain b contains a short a-helical segment (S209– L211) that interacts with helix a2 of the b0 domain, forming an important part of the b-b0 interface. The b and b0 domains of ERp57 form an essentially rigid molecule, and as a result, all three chains in the asymmetric unit can be superimposed using their entire backbone with an rmsd of 0.8–1.0 A˚ (Figure 1C). The short three residue linker between these domains is involved in interdomain contacts and has temperature factors similar to proximal regions in both domains. A structural similarity search using the Dali database (Holm and Sander, 1995) gave scores that were generally higher for the b domain, which has higher sequence similarity to other thioredoxin-like domains. The best match was the second thioredoxin-like domain from Figure 1. X-Ray Crystal Structure of the bb0 Fragment of ERp57 (A) Crystal packing of the three molecules in the asymmetric unit cell. (B) Secondary structure of the b and b0 domains. (C) Overlay of the three molecules showing the rigidity of the bb0 fragment.
Crystal Structure of the bb0 Fragment of ERp57 1333
Figure 2. Comparison of Yeast PDI and Human ERp57 Structures (A) Domain structure of ERp57, showing the four thioredoxin-like domains and the C-terminal polylysine tract (triangle). (B) Structure-based sequence alignment. Secondary structure elements are shown in red for ERp57 and in light blue for PDI. Asterisks mark the locations of sequence insertions. Blue triangles mark lysine and arginine residues in ERp57 essential for binding the CNX P domain. Residues in the b0 domain of ERp57 are underlined. (C) Two structural superpositions of the b and b0 domains from PDI and ERp57. Asterisks mark the insertions in the b domain of ERp57 (dark red) and in the b0 domain of PDI (light blue).
calsequestrin (Z = 14.9); both domains could be superimposed using their backbone atoms of 101 residues with an rmsd of 1.9 A˚. The tandem of b and b0 domains superimposed on the calsequestrin structure with an rmsd of 3.7 A˚ over 204 residues, indicating similar relative orientation of these domains in calsequestrin. The similarity of PDI to calsequestrin has been previously noted (Tian et al., 2006). The closest structural homolog for the complete bb0 fragment of ERp57 was yeast PDI, which could be superimposed with an rmsd of 3.5 A˚ (Z = 14.1) over 199 residues. A structure-based sequence alignment of the bb0 fragments of human ERp57 and yeast PDI highlights the similarities and differences between the two proteins (Figure 2B). The biggest differences are two 6–7 residue inserts in the b domain and a long deletion in the b0 domain of ERp57 relative to PDI (Figure 2C). The first insert comprises helix a3. The second insert enlarges the b4-b5 loop in the b domain, with important structural and functional consequences. In ERp57, this loop contains residues that participate in CNX binding (see below). It also creates an extensive contact surface between the b and b0 domains, not seen in yeast PDI. As a result, the angle between the b and b0 domains in ERp57 differs by roughly 15 from the angle in PDI. This loop may also contribute to the rigidity of the b and b0 pair. In contrast to the b domain, which is smaller in yeast PDI, the b0 domain contains an additional helix as part of the loop following sheet b50 . Notwithstanding their structural similarity, the bb0 domains of ERp57 and PDI differ significantly in their surface charges. The largest difference is in helix a20 of the b0 domain, which carries a formal charge of +4 in ERp57 and 24 in PDI (Figure 2B). The positive charge of this region of ERp57 likely plays an important role in its specific binding to the highly negatively charged P domain of CNX.
Small-Angle X-Ray Scattering of Full-Length ERp57 In order to determine the overall shape and domain organization of ERp57, we carried out an analysis by small-angle X-ray scattering (SAXS). Small-angle scattering can be used to calculate the radius of gyration (Rg) and the maximum dimensions of the molecule (Dmax), but it is also sensitive to the overall shape of a molecule and can be used to generate molecular shape reconstructions and test structural models (Koch et al., 2003). We collected scattering data on full-length ERp57 with no evidence of aggregation up to 10 mg/ml (Figure 3A). Independent support of our SAXS data comes from two recent papers: a SAXS study of human PDI (Li et al., 2006) and the crystal structure of yeast PDI (Tian et al., 2006). Qualitatively, there is good agreement between the pair distance distribution functions P(r) calculated for ERp57 and human PDI with the presence of a marked, double-maximum centered around 40 A˚ in both curves (Figure 3A, insert) (Li et al., 2006). The theoretical scattering curve calculated using the yeast PDI crystal structure and the experimental curve for fulllength ERp57 are also very similar, with a low residual coefficient c of 4.76 (Figure 3A) (Svergun et al., 1995). The fit between the scattering curves was quite sensitive to changes in the angles between the domains in the atomic model; replacing the bb0 domains of PDI with the domains from ERp57 did not improve the fit. While the yeast PDI structure suggests that the a and a0 domains are relatively mobile (Tian et al., 2006), our SAXS data imply that, at low resolution, the average conformation of ERp57 in solution closely resembles the PDI crystal structure. The program GASBOR (Svergun et al., 2001) was used to generate structural models based on the ERp57 SAXS data. The program uses simulated annealing to match the experimental data to a computed scattering curve from a pseudoatom model. The models, consisting of
Structure 1334
Figure 3. Small-Angle X-Ray Scattering of Full-Length ERp57 (A) Fit of the experimental ERp57 scattering data (red) and the scattering curve calculated from the PDI X-ray structure (green). R(g) = 31.5. The insert shows the experimental P(r) plot. (B) Pseudo-atom reconstruction (GASBOR) of ERp57 superposed on the X-ray structure of yeast PDI (Tian et al., 2006). The domains of PDI are labeled. In the lower view, the a domain is hidden on the backside and not labeled.
one scatterer (pseudoatom) per amino acid, were aligned, averaged, and filtered based on occupancy using DAMAVER (Volkov and Svergun, 2003) to obtain a ‘‘most probable’’ model, which clearly shows four globular domains (Figure 3B). Superposition of the yeast PDI X-ray structure and DAMAVER model shows that the relative positions of the domains observed in the crystal are conserved in ERp57 in solution. Most notably, both structures have an overall U shape and place the two catalytically active domains a and a0 on one side of the molecule and the b and b0 domains on the other (Figure 3B). Identification of Individual Residues in the CNX Binding Site The high-quality spectra of the b0 domain of ERp57 allowed us to use NMR to identify the CNX binding site. The 1H-15N HSQC spectrum was characteristic of a folded, well-behaved protein with good signal dispersion (see Figure S1 in Supplemental Data available with this article online). Complete backbone resonance assignments for the b0 domain were obtained using standard triple-resonance experiments on a doubly (15N, 13 C) labeled sample. Titration of 15N-labeled b0 domain with unlabeled CNX P domain residues P310–K381 caused amide chemical shift changes for residues D266 (0.25 ppm), V267 (0.20 ppm), A273 (0.14 ppm), R280 (0.18 ppm), N281 (0.23 ppm), and R344 (0.17 ppm). These lie in the region following strand b20 , in the long helix a2 and the region preceding helix a40 (Figure 4B). Other residues in this region such as K274, N277, Y278, and W279 may be involved in binding, but their chemical shift changes could not be measured because of overlapping or missing peaks. Analysis of surface charge distribution reveals that the chemical shift changes localize to a large positively charged patch of residues (Figure 4C). This was not surprising, as the tip of the P domain is highly acidic and would be expected to interact electrostatically with a positively charged surface. This area also corresponds
to the highest sequence conservation in the bb0 domains across ERp57 proteins from other species (Figure 4D). Among the most conserved, surface-accessible residues are the basic amino acids: K214, K274, and R282. The corresponding surface and residues in the b0 domain of yeast PDI are negatively charged, consistent with its lack of interaction with CNX (Figure 2B). Mutagenesis Confirms the Binding Site Isothermal titration calorimetry (ITC) was used to measure the binding between different fragments of ERp57 and CNX (Figure 5). Full-length ERp57 binds to the twomodule P domain fragment of CNX with an affinity of 26 mM. The binding is exothermic, which reflects favorable van der Waals, hydrogen bond, and electrostatic interactions. These results are comparable to the binding of ERp57 to CRT, also measured by ITC (Frickel et al., 2002). The affinity of the P domain for the bb0 fragment was slightly higher than for full-length ERp57. This could reflect greater accessibility of the binding site in the bb0 fragment. The smallest ERp57 fragment that bound the P domain was the b0 domain, which bound with 50 times lower affinity. Together, these measurements demonstrate that the b0 domain contains the principal elements for ERp57 binding to CNX, while the b domain provides additional contacts necessary for high affinity. To verify the identification of the CNX binding site, we generated site-directed alanine substitution mutants and tested their binding using NMR and ITC. Two mutants, K274A and R282A, were designed to remove electrostatic interactions in the conserved, basic surface of the b0 domain. We generated a K214A mutant in the b domain to test the role of the b4-b5 loop in binding. As a negative control, a lysine residue (K332) on the opposite side of the b0 domain was mutated to alanine (Figure 5A). The specificity of the P domain-ERp57 interaction was also tested with an M347A mutation in the two-module P domain of CNX. This conserved residue occurs at the tip of the P domain and shifts in HSQC spectra upon bb0
Crystal Structure of the bb0 Fragment of ERp57 1335
half of the effect of deletion of the entire b domain. The K332A mutant showed binding affinity identical to wild-type, confirming that binding is specific and is not related to the overall charge of the bb0 domains.
Figure 4. Identification of the CNX Binding Site on ERp57 (A) Plot of chemical shift changes in the 15N-labeled ERp57 b0 domain upon addition of unlabeled CNX P domain. (B) Mapping of the chemical shifts onto the bb0 structure. Red indicates a large chemical shift change, white indicates no change, and gray indicates no data. The largest changes in the lower b0 domain are labeled. (C) Electrostatic potential of bb0 showing the abundance of basic residues in the P domain binding site. The right-hand surface is in the same orientation as the representation in (B); the left-hand surface is rotated by 180 . (D) Sequence conservation of surface residues on the ERp57 bb0 domains. Highly conserved residues in 50 protein sequences most closely related to human ERp57 are shown in green. The P domain binding site corresponds to the largest area of conserved amino acid residues.
binding (Figure S2), but its direct role in ERp57 binding has never been tested. All the mutant proteins were correctly folded, as verified by 1D NMR spectra and comparison with spectra for the corresponding wild-type fragments (Figures S3–S5). Mutation of either R282 in ERp57 or M347 in CNX completely abrogated binding, as measured by ITC and NMR titrations. Both mutations caused a >1000fold loss in affinity based on the upper limit (w5 mM) for detecting binding by NMR. Less dramatic effects were observed with the K214A mutation, which showed an 8-fold loss of affinity (Figure 5C). The side chain of K214 appears to be the most important recognition element in the b domain, as its deletion accounted for over
ERp57-CNX Interaction Is Essential for In Vitro Folding of RNase B We studied the competence of ERp57 and CNX for catalyzing protein folding in vitro using RNase B as a model substrate and low-pH native gels (Zapun et al., 1998). Folding occurs in an ER-like redox buffer containing the luminal domain of CNX and full-length ERp57. The required substrate Glc1 RNase B carries a single N-glycan of the form Glc1Man8–9GlcNAc2 at Asn34. This Glc1labeled RNase B is specifically bound by the lectin domain of CNX (Zapun et al., 1997), which stimulates the disulfide exchange activity of ERp57 to refold the substrate. As previously reported (Zapun et al., 1998), the folding reaction is dependent on the presence of the terminal glucose residue on the substrate. The commercial source of RNase B contains w90% of Man5–7GlcNAc2 RNase B, which cannot be labeled with 3H-glucose by UGGT. This unlabeled RNase B can be visualized in the gel assay by Coomassie blue staining and remains mostly unfolded after 80 min (Figure 5D, CB). 3H-glucosylated RNase B, when incubated with wild-type ERp57 and CNX, was largely folded after 40 min (Figure 5D, AR). The ability of the ERp57-CNX complex to mediate folding of 3H-labeled RNase B was completely dependent on a functional interaction between ERp57 and CNX. The point mutants K214A, K274A, and R282A of full-length ERp57 were compromised in their ability to fold Glc1-labeled RNase B (Figure 5D). Similarly, omission of either CNX or ERp57 from the reaction led to only a minor amount of spontaneous folding of RNase B, comparable to that observed with unlabeled RNase B (Figure 5D). In contrast, the K332A mutant, which does not affect CNX binding, catalyzed rapid folding. As a control for nonspecific effects, we examined the NMR spectra of the mutant proteins and characterized their thiol reductase activity against a small-molecule substrate, di(o-aminobenzyl)-labeled oxidized glutathione (diabz-GSSG) (Raturi et al., 2005). All of the ERp57 mutants were folded (Figures S4 and S5) and had full reductase activity (our unpublished data). Significantly, the addition of the luminal domain of CNX or the twomodule P domain did not stimulate ERp57 activity toward diabz-GSSG (Figure S6). Discussion The X-ray structure of the bb0 domains provides atomic resolution details of the CNX binding site on ERp57. This site consists of a large positively charged patch of highly conserved residues (K214, K274, and R282) that are essential for the function of ERp57. These residues engage the negatively charged residues of the CNX P domain, which along with the critical methionine residue (M347) comprise the tip of the P domain. While the b0 domain provides the majority of the binding site, the b4-b5 loop of the b domain affords additional contacts (K214) with the P domain, strengthening the interaction.
Structure 1336
Figure 5. Mutagenesis of the CNX-ERp57 Interaction Sites Showing the Requirement for the ERp57-CNX Interaction for Protein Refolding (A) Structure of ERp57 bb0 domains showing space-filling models of the residues mutated. (B) ITC thermogram of the interaction of the CNX two-module P domain binding to the bb0 domains of ERp57. (C) Table of ITC binding affinity constants for wild-type and mutant proteins. (D) In vitro refolding assays of RNase B with wild-type and mutant ERp57 proteins and the luminal domain of CNX. 3H-glucose-labeled RNase B was refolded for the time indicated and run on low-pH nondenaturing PAGE. Autoradiograms (AR) show the relative amounts of natively folded (N) and unfolded (U) 3H-Glc1-RNase B. For the wild-type assay, the Coomassie blue (CB)-stained gel shows the spontaneous folding of unlabeled RNase B. The right panels show autoradiograms at 80 min for four mutant ERp57 proteins and negative controls where CNX or ERp57 was omitted from the folding reaction.
Mutations previously shown to compromise the ERp57-CRT interaction (V267, Y269, K274, R280, N281, and F299) correlate well with our results (Russell et al., 2004). Among these, K274, R280, and N281 are surface exposed and likely hydrogen bond with the negatively charged P domain. Y269 is partially surface accessible and could form part of a hydrophobic pocket at the base of the b20 -a20 loop for the binding of M347 of CNX. On the other hand, V267 and F299 are both buried in the hydrophobic core and are unable to directly interact with the P domain. These mutations are both located under helix a20 and may affect binding via indirect contacts or through distortion of the b-b0 interdomain interface. Intriguingly, the CNX binding site in the b0 domain corresponds to the position of the catalytic cysteines in the b2-a2 loop of thioredoxins and the primary peptide binding site in the b0 domain of PDI (Pirneskoski et al., 2004). This supports earlier suggestions that the substrate binding domains in protein disulfide isomerases were created through gene duplication of catalytic domains and evolutionary modification of their active sites (loss of the catalytic cysteines) to favor protein binding (Kanai et al., 1998). Our SAXS results demonstrate that the solution structure of ERp57 is highly similar to the crystal structure of PDI from yeast. This places the CNX binding site on the
opposite face of ERp57 relative to the catalytic thiols (Figure 6). The long hook-like P domain is ideally shaped to curve around ERp57 and position the lectin binding site of CNX next to the catalytic a and a0 domains. The SAXS model also suggests that the positively charged C terminus of ERp57 is too far from the CNX binding site on the b0 domain of ERp57 to participate in direct binding of the P domain. This basic stretch, previously identified as interacting with CNX (Pollock et al., 2004), may interact with the phospholipid membrane, thus affecting binding of CNX and the efficiency of ERp57mediated protein folding. In PDI, the C-terminal sequence is acidic and forms a highly solvent-exposed a helix. How does the CNX-ERp57 complex work to fold proteins? The long distance (>40 A˚) between the CNX binding site and the catalytic cysteines argues against a CNX-induced conformational change or direct stimulation of the oxidoreductase activity of the a and a0 domains. Instead, the structure of ERp57 suggests that CNX serves to bind and position substrates next to the catalytic cysteines. The loss of either glycoprotein binding to CNX or CNX binding to ERp57 abrogates ERp57-catalyzed protein folding. Because the reactions done here are performed at low substrate concentrations, decreasing the KM of the folding reaction has the same effect as a stimulation of the reaction Vmax. More
Crystal Structure of the bb0 Fragment of ERp57 1337
from the Guinier plot by using PRIMUS, GNOM, and CRYSOL. The pair distance distribution function (P[r]) was calculated by using the program GNOM. Scattering curves and the goodness of fit (c) based on atomic models were obtained by using CRYSOL. Crystallization Crystals were obtained by the hanging drop equilibration of 14 mg/ml protein in 50 mM Tris-HCl, 0.15 M NaCl, 1 mM DTT (pH 7.5) against 30% (w/v) PEG3350, 0.1 M (NH4)2SO4, 0.1 M HEPES buffer (pH 7.5) for 1–3 days at 20 C.
Figure 6. Model of Full-Length ERp57 The a and a0 domains from the yeast PDI structure (light gray) were positioned on the bb0 fragment of ERp57 (dark gray) to show the relative positions of the thiol reactive cysteine residues (yellow) and the three basic residues, K214, K274, and R282, of the CNX binding site (blue).
kinetic studies are needed to test this hypothesis and elucidate the functional aspects of the CNX-ERp57 interaction. Experimental Procedures Protein Expression, Preparation, and Purification Full-length human ERp57 lacking the signal sequence (A24–L505) and the bb0 (P134–G376) and b0 (F241–G376) fragments were cloned as GST fusions between the BamH1 and Not1 sites of pGEX-6P-1 (Amersham Pharmacia) and expressed in Escherichia coli BL21(DE3). To prepare labeled protein for NMR experiments, cells were grown in M9 minimal medium with 15N-NH4Cl and 13C-U-glucose. Selenomethionine-labeled protein was produced in the E. coli methionine auxotroph DL41(DE3) grown on LeMaster medium (Hendrickson et al., 1990). The GST fusion proteins were purified by affinity chromatography on glutathione-Sepharose resin and tags were removed by cleavage with Prescission protease, leaving N-terminal Gly-Pro-Leu-Gly-Ser and C-terminal Ala-Ala-Ala-Ser extensions. The cleaved protein was additionally purified using size exclusion chromatography and concentrated using Centricon 3000 concentrators. The P domain two-module fragment of canine CNX (residues 310–381) was expressed and purified as described (Pollock et al., 2004). The luminal domain of canine CNX (Ou et al., 1995) and recombinant rat UDP-glucose:glycoprotein glucosyltransferase (UGGT) (Tessier et al., 2000) were expressed in a baculovirus/sf9 insect cell system as described previously (Zapun et al., 1998). Estimates of protein concentrations were based on theoretical UV absorption coefficients. Preparation of CNX and ERp57 Point Mutants Mutants were prepared by Quickchange site-directed mutagenesis (Stratagene) with mismatched primers and verified by DNA sequencing. SAXS Small-angle X-ray scattering data were collected at beamline 12.3.1 at the Advanced Light Source (Lawrence Berkeley National Laboratory, Berkeley, CA) with protein concentrations of 2, 5, and 10 mg/ml. Data were scaled by using the program PRIMUS (Konarev et al., 2003) and analyzed by the programs GNOM (Svergun, 1992) and CRYSOL (Svergun et al., 1995). Radii of gyration (Rg) were computed
Structure Solution and Refinement Three-wavelength MAD data were collected at beamline X8C at the National Synchrotron Light Source, Brookhaven National Laboratory (Table 1) and processed with HKL2000 (Otwinowski and Minor, 1997) and d*TREK (Pflugrath, 1999). The structure was determined using SHELXC/D/E (Sheldrick and Schneider, 1997) and ARP/ wARP (Perrakis et al., 1997). Data to 2.1 A˚ resolution at the peak wavelength (0.9795 A˚) were used to locate Se atoms with a figure of merit = 0.66. Density modification with the program ARP/wARP (Perrakis et al., 1997) allowed automated model building of w75% of the residues. The partial model obtained from ARP/wARP was extended manually with O (Jones et al., 1991) and improved by refinement using REFMAC (Murshudov et al., 1999) with the translation libration screw (TLS) option (Winn et al., 2003). Figures were prepared with PyMOL (Delano, 2002) and GRASP (Nicholls et al., 1991). The protein contact potential was calculated by PyMOL using default vacuum electrostatics ranging and color coded from 280 kT (red) to 80 kT (blue), where k is the Boltzmann constant and T is temperature. Coordinates have been deposited in the RCSB Protein Data Bank under ID code 2H8L. NMR Spectroscopy NMR resonance assignments of the b0 fragment of ERp57 in 50 mM Na2HPO4, 100 mM NaCl, 1 mM DTT (pH 6.8) were carried out using HNCACB, CBCA(CO)NH, HNCA, and 15N-noesy-HMQC experiments at 30 C at 500, 600, and 800 MHz. Amide signal assignments have been deposited in the Biological Magnetic Resonance Bank under ID code BMRB-7162. Isothermal Titration Calorimetry Measurements Experiments were carried out on a MicroCal VP-ITC titration calorimeter (MicroCal, Northampton, MA) in 50 mM Tris-HCl, 100 mM NaCl, 1 mM NaN3 (pH 6.8) at 15 C. Data were processed using ORIGIN and the binding isotherm was fit to a single-site binding model to determine the binding constant. RNase B Folding The RNase B folding assay was carried out as previously described (Zapun et al., 1998) with noted changes. The Glc1 form of RNase B was prepared from Man8GlcNAc2 or Man9GlcNAc2 RNase B (Sigma) using UGGT and 3H-UDP-glucose. Briefly, 500 mg of RNase B was unfolded and reduced for 20 min at room temperature in 0.1 M Tris-HCl (pH 8.0) containing 6 M guanidine HCl and 20 mM DTT. Unfolded RNase B was desalted in 0.1% formic acid on a NAP-5 column (Amersham Biosciences) and lyophilized. RNase B (z120 mM) was then glucosylated with 5–15 mM, 10–30 mCi/mmol [3H]UDP-glucose using a crude concentrate of recombinant rat UGGT expressed in sf9 insect cells (Tessier et al., 2000) in 25 mM HEPES, 5 mM CaCl2 containing 200 mM 1-deoxynojirimycin for 2 hr at 37 C. For each folding assay, the labeling reaction was adjusted to 150 mM NaCl, 2.5 mM CaCl2, 0.5 mM oxidized glutathione, 2 mM reduced glutathione containing 24 mM purified calnexin luminal domain (Ou et al., 1995) and 8 mM ERp57 or its mutants and incubated at room temperature. Time points were taken by blocking free thiols with 0.25 volumes of 0.5 M iodoacetamide in 1.5 M Tris-HCl (pH 8.7) and storing on ice before analyzing by low-pH nondenaturing PAGE (Creighton, 1979). Gels were Coomassie stained, destained, and treated with Amplify (Amersham Biosciences) for 30 min before drying and exposing to Kodak Biomax MS film for 24–96 hr. Supplemental Data Supplemental data include six figures showing the identification of the ERp57 binding site on the CNX P domain, 2D NMR spectrum
Structure 1338
of the b0 domain of CNX, 1D NMR spectra of the point mutants in the CNX P domain, ERp57 bb0 domains, full-length ERp57 showing that they are correctly folded, and results of GSSG reduction by ERp57. Supplemental data are available at http://www.structure.org/cgi/ content/full/14/8/1331/DC1.
Kanai, S., Toh, H., Hayano, T., and Kikuchi, M. (1998). Molecular evolution of the domain structures of protein disulfide isomerases. J. Mol. Evol. 47, 200–210.
Acknowledgments
Konarev, P.V., Volkov, V.V., Sokolova, A.V., Koch, M.H.J., and Svergun, D.I. (2003). PRIMUS: a Windows PC-based system for smallangle scattering data analysis. J. Appl. Crystallogr. 36, 1277–1282.
We thank Tara Sprules for help in running NMR experiments, Tudor Moldoveanu for advice in initial crystallization trials, Olivia Groover and Ekaterina Pomerantseva for technical assistance, and Arun Raturi and Bulent Mutus for the gift of diabz-GSSG. Grateful acknowledgment is given to Greg Hura, who collected the SAXS data on SYBYL beamline 12.3.1 at the Advanced Light Source, which is supported by the U.S. Department of Energy under contract DE-AC02-05CH11231. This work was supported by a grant from the Canadian Institutes of Heath Research to K.G. and D.Y.T. Received: May 4, 2006 Revised: June 10, 2006 Accepted: June 13, 2006 Published: August 15, 2006 References Bourdi, M., Demady, D., Martin, J.L., Jabbour, S.K., Martin, B.M., George, J.W., and Pohl, L.R. (1995). cDNA cloning and baculovirus expression of the human liver endoplasmic reticulum P58: characterization as a protein disulfide isomerase isoform, but not as a protease or a carnitine acyltransferase. Arch. Biochem. Biophys. 323, 397–403. Creighton, T.E. (1979). Intermediates in the refolding of reduced ribonuclease A. J. Mol. Biol. 129, 411–431. Delano, W.L. (2002). The PyMOL Molecular Graphics System (San Carlos, CA: DeLano Scientific). Ellgaard, L., and Frickel, E.M. (2003). Calnexin, calreticulin, and ERp57: teammates in glycoprotein folding. Cell Biochem. Biophys. 39, 223–247. Ellgaard, L., and Ruddock, L.W. (2005). The human protein disulphide isomerase family: substrate interactions and functional properties. EMBO Rep. 6, 28–32. Frickel, E.M., Riek, R., Jelesarov, I., Helenius, A., Wuthrich, K., and Ellgaard, L. (2002). TROSY-NMR reveals interaction between ERp57 and the tip of the calreticulin P-domain. Proc. Natl. Acad. Sci. USA 99, 1954–1959. Garbi, N., Tanaka, S., Momburg, F., and Hammerling, G.J. (2006). Impaired assembly of the major histocompatibility complex class I peptide-loading complex in mice deficient in the oxidoreductase ERp57. Nat. Immunol. 7, 93–102. Helenius, A., and Aebi, M. (2004). Roles of N-linked glycans in the endoplasmic reticulum. Annu. Rev. Biochem. 73, 1019–1049. Hendrickson, W.A., Horton, J.R., and LeMaster, D.M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of threedimensional structure. EMBO J. 9, 1665–1672.
Koch, M.H., Vachette, P., and Svergun, D.I. (2003). Small-angle scattering: a view on the properties, structures and structural changes of biological macromolecules in solution. Q. Rev. Biophys. 36, 147–227.
Laskowski, R.A., MacArthur, M.W., Moss, D.S., and Thornton, J.M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. Leach, M.R., Cohen-Doyle, M.F., Thomas, D.Y., and Williams, D.B. (2002). Localization of the lectin, ERp57 binding, and polypeptide binding sites of calnexin and calreticulin. J. Biol. Chem. 277, 29686–29697. Li, S.J., Hong, X.G., Shi, Y.Y., Li, H., and Wang, C.C. (2006). Annular arrangement and collaborative actions of four domains of proteindisulfide isomerase: a small angle X-ray scattering study in solution. J. Biol. Chem. 281, 6581–6588. Murshudov, G.N., Vagin, A.A., Lebedev, A., Wilson, K.S., and Dodson, E.J. (1999). Efficient anisotropic refinement of macromolecular structures using FFT. Acta Crystallogr. D Biol. Crystallogr. 55, 247– 255. Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. Oliver, J.D., van der Wal, F.J., Bulleid, N.J., and High, S. (1997). Interaction of the thiol-dependent reductase ERp57 with nascent glycoproteins. Science 275, 86–88. Oliver, J.D., Roderick, H.L., Llewellyn, D.H., and High, S. (1999). ERp57 functions as a subunit of specific complexes formed with the ER lectins calreticulin and calnexin. Mol. Biol. Cell 10, 2573– 2582. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Ou, W.J., Cameron, P.H., Thomas, D.Y., and Bergeron, J.J. (1993). Association of folding intermediates of glycoproteins with calnexin during protein maturation. Nature 364, 771–776. Ou, W.J., Bergeron, J.J., Li, Y., Kang, C.Y., and Thomas, D.Y. (1995). Conformational changes induced in the endoplasmic reticulum luminal domain of calnexin by Mg-ATP and Ca2+. J. Biol. Chem. 270, 18051–18059. Parodi, A.J. (2000). Protein glucosylation and its role in protein folding. Annu. Rev. Biochem. 69, 69–93. Peaper, D.R., Wearsch, P.A., and Cresswell, P. (2005). Tapasin and ERp57 form a stable disulfide-linked dimer within the MHC class I peptide-loading complex. EMBO J. 24, 3613–3623. Perrakis, A., Sixma, T.K., Wilson, K.S., and Lamzin, V.S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple-refined dummy atomic models. Acta Crystallogr. D Biol. Crystallogr. 53, 448–455. Pflugrath, J.W. (1999). The finer things in X-ray diffraction data collection. Acta Crystallogr. D Biol. Crystallogr. 55, 1718–1725.
High, S., Lecomte, F.J., Russell, S.J., Abell, B.M., and Oliver, J.D. (2000). Glycoprotein folding in the endoplasmic reticulum: a tale of three chaperones? FEBS Lett. 476, 38–41.
Pirneskoski, A., Klappa, P., Lobell, M., Williamson, R.A., Byrne, L., Alanen, H.I., Salo, K.E., Kivirikko, K.I., Freedman, R.B., and Ruddock, L.W. (2004). Molecular characterization of the principal substrate binding site of the ubiquitous folding catalyst protein disulfide isomerase. J. Biol. Chem. 279, 10374–10381.
Hirano, N., Shibasaki, F., Sakai, R., Tanaka, T., Nishida, J., Yazaki, Y., Takenawa, T., and Hirai, H. (1995). Molecular cloning of the human glucose-regulated protein ERp57/GRP58, a thiol-dependent reductase. Identification of its secretory form and inducible expression by the oncogenic transformation. Eur. J. Biochem. 234, 336–342.
Pollock, S., Kozlov, G., Pelletier, M.F., Trempe, J.F., Jansen, G., Sitnikov, D., Bergeron, J.J., Gehring, K., Ekiel, I., and Thomas, D.Y. (2004). Specific interaction of ERp57 and calnexin determined by NMR spectroscopy and an ER two-hybrid system. EMBO J. 23, 1020–1029.
Holm, L., and Sander, C. (1995). Dali: a network tool for protein structure comparison. Trends Biochem. Sci. 20, 478–480.
Raturi, A., Vacratsis, P.O., Seslija, D., Lee, L., and Mutus, B. (2005). A direct, continuous, sensitive assay for protein disulphide-isomerase based on fluorescence self-quenching. Biochem. J. 391, 351–357.
Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119.
Rosenbaum, E.E., Hardie, R.C., and Colley, N.J. (2006). Calnexin is essential for rhodopsin maturation, Ca2+ regulation, and photoreceptor cell survival. Neuron 49, 229–241.
Crystal Structure of the bb0 Fragment of ERp57 1339
Russell, S.J., Ruddock, L.W., Salo, K.E., Oliver, J.D., Roebuck, Q.P., Llewellyn, D.H., Roderick, H.L., Koivunen, P., Myllyharju, J., and High, S. (2004). The primary substrate binding site in the b0 domain of ERp57 is adapted for endoplasmic reticulum lectin association. J. Biol. Chem. 279, 18861–18869. Schrag, J.D., Bergeron, J.J., Li, Y., Borisova, S., Hahn, M., Thomas, D.Y., and Cygler, M. (2001). The structure of calnexin, an ER chaperone involved in quality control of protein folding. Mol. Cell 8, 633– 644. Schrag, J.D., Procopio, D.O., Cygler, M., Thomas, D.Y., and Bergeron, J.J. (2003). Lectin control of protein folding and sorting in the secretory pathway. Trends Biochem. Sci. 28, 49–57. Sheldrick, G., and Schneider, T. (1997). SHELXL: high-resolution refinement. Methods Enzymol. 277, 319–343. Silvennoinen, L., Karvonen, P., Koivunen, P., Myllyharju, J., Kivirikko, K., and Kilpelainen, I. (2001). Assignment of 1H, 13C and 15N resonances of the a0 domain of ERp57. J. Biomol. NMR 20, 385–386. Silvennoinen, L., Koivunen, P., Myllyharju, J., Kilpelainen, I., and Permi, P. (2005). NMR assignment of the N-terminal domain a of the glycoprotein chaperone ERp57. J. Biomol. NMR 33, 136. Solda, T., Garbi, N., Hammerling, G.J., and Molinari, M. (2006). Consequences of ERp57 deletion on oxidative folding of obligate and facultative clients of the calnexin cycle. J. Biol. Chem. 281, 6219– 6226. Srivastava, S.P., Fuchs, J.A., and Holtzman, J.L. (1993). The reported cDNA sequence for phospholipase C a encodes protein disulfide isomerase, isozyme Q-2 and not phospholipase-C. Biochem. Biophys. Res. Commun. 193, 971–978. Svergun, D.I. (1992). Determination of the regularization parameter in indirect-transform methods using perceptual criteria. J. Appl. Crystallogr. 25, 495–503. Svergun, D., Barberato, C., and Koch, M.H.J. (1995). CRYSOL— a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Crystallogr. 28, 768– 773. Svergun, D.I., Petoukhov, M.V., and Koch, M.H. (2001). Determination of domain structure of proteins from X-ray solution scattering. Biophys. J. 80, 2946–2953. Tessier, D.C., Dignard, D., Zapun, A., Radominska-Pandya, A., Parodi, A.J., Bergeron, J.J., and Thomas, D.Y. (2000). Cloning and characterization of mammalian UDP-glucose glycoprotein: glucosyltransferase and the development of a specific substrate for this enzyme. Glycobiology 10, 403–412. Tian, G., Xiang, S., Noiva, R., Lennarz, W.J., and Schindelin, H. (2006). The crystal structure of yeast protein disulfide isomerase suggests cooperativity between its active sites. Cell 124, 61–73. Tu, B.P., and Weissman, J.S. (2004). Oxidative protein folding in eukaryotes: mechanisms and consequences. J. Cell Biol. 164, 341–346. Volkov, V.V., and Svergun, D.I. (2003). Uniqueness of ab initio shape determination in small-angle scattering. J. Appl. Crystallogr. 36, 860–864. Winn, M.D., Murshudov, G.N., and Papiz, M.Z. (2003). Macromolecular TLS refinement in REFMAC at moderate resolutions. Methods Enzymol. 374, 300–321. Zapun, A., Petrescu, S.M., Rudd, P.M., Dwek, R.A., Thomas, D.Y., and Bergeron, J.J. (1997). Conformation-independent binding of monoglucosylated ribonuclease B to calnexin. Cell 88, 29–38. Zapun, A., Darby, N.J., Tessier, D.C., Michalak, M., Bergeron, J.J., and Thomas, D.Y. (1998). Enhanced catalysis of ribonuclease B folding by the interaction of calnexin or calreticulin with ERp57. J. Biol. Chem. 273, 6009–6012.
Accession Numbers Coordinates of the structure of the bb0 domains of ERp57 have been deposited in the RCSB Protein Data Bank under ID code 2H8L. Amide signal assignments have been deposited in the Biological Magnetic Resonance Bank under ID code BMRB-7162.
DOI 10.1016/j.str.2006.06.019
Crystal Structure of the bb0 Domains of the Protein Disulfide Isomerase ERp57 Guennadi Kozlov,1,4 Pekka Maattanen,1,4 Joseph D. Schrag,2 Stephanie Pollock,1 Miroslaw Cygler,2 Bhushan Nagar,1 David Y. Thomas,1 and Kalle Gehring1,3,* 1 Biochemistry Department McGill University 3655 Promenade Sir William Osler Montre´al, Que´bec H3G 1Y6 Canada 2 Biotechnology Research Institute National Research Council of Canada 6100 Royalmount Avenue Montre´al, Que´bec H4P 2R2 Canada
Summary The synthesis of proteins in the endoplasmic reticulum (ER) is limited by the rate of correct disulfide bond formation. This process is carried out by protein disulfide isomerases, a family of ER proteins which includes general enzymes such as PDI that recognize unfolded proteins and others that are selective for specific proteins or classes. Using small-angle X-ray scattering and X-ray crystallography, we report the structure of a selective isomerase, ERp57, and its interactions with the lectin chaperone calnexin. Using isothermal titration calorimetry and NMR spectroscopy, we show that the b0 domain of ERp57 binds calnexin with micromolar affinity through a conserved patch of basic residues. Disruption of this binding site by mutagenesis abrogates folding of RNase B in an in vitro assay. The relative positions of the ERp57 catalytic sites and calnexin binding site suggest that activation by calnexin is due to substrate recruitment rather than a direct stimulation of ERp57 oxidoreductase activity. Introduction In order to fold in the oxidative environment of the endoplasmic reticulum (ER), proteins need to acquire the proper arrangement of cysteine disulfide bonds. To meet this challenge, the cell produces a large number of different ER protein disulfide isomerases to catalyze the oxidation and isomerization of disulfides (Ellgaard and Ruddock, 2005; Tu and Weissman, 2004). The best-known and most abundant of these is the primary oxidant PDI. For N-glycosylated proteins, a dedicated pathway exists mediated by the protein disulfide isomerase ERp57, the lectin-like molecular chaperones calnexin (CNX) and calreticulin (CRT), and UDP-glucose: glycoprotein glucosyltransferase (UGGT) (Ellgaard and Frickel, 2003; Helenius and Aebi, 2004; High et al., 2000; Parodi, 2000; Schrag et al., 2003). Following syn-
*Correspondence: [email protected] 3 Lab address: http://www.mcgnmr.ca 4 These authors contributed equally to this work.
thesis and N-glycosylation, the glycan is trimmed by the removal of two glucose residues to generate a Glc1Man9GlcNAc2 glycoprotein that binds to CNX or CRT (Ou et al., 1993; Zapun et al., 1997). Together with ERp57 (Frickel et al., 2002; Pollock et al., 2004; Zapun et al., 1998), this allows for correct formation of the disulfide bonds and further maturation of the protein through glycan processing. A quality control system exists in which UGGT recognizes unfolded Man9GlcNAc2 glycoproteins and reglucosylates them for CNX/CRT binding and further ERp57 processing. This pathway, known as the calnexin cycle, is responsible for the folding of proteins such as viral hemagglutinin, rhodopsin, transferrin, and major histocompatibility complex MHC class I molecules (Ou et al., 1993; Rosenbaum et al., 2006; Solda et al., 2006). Recent evidence suggests that ERp57 may also play a structural role in the MHC class I complex with TAP (transporter associated with antigen processing), where it forms a covalent, disulfidelinked complex with tapasin (Garbi et al., 2006; Peaper et al., 2005). ERp57 (also called ER-60 or GRP58) is composed of four thioredoxin-like domains abb0 a0 with CGHC catalytic motifs in the N- and C-terminal a and a0 domains. ERp57 is highly similar to PDI both in its domain organization and sequence. Similarity is highest in the catalytic a and a0 domains (w50% identity) and lowest in the bb0 domains (20%). ERp57 possesses oxidoreductase activity (Bourdi et al., 1995; Hirano et al., 1995; Srivastava et al., 1993) but is most active when coupled to CNX/ CRT (Zapun et al., 1998). The physical association between ERp57 and CNX/CRT has been shown by crosslinking (Oliver et al., 1997, 1999) and by NMR (Frickel et al., 2002; Pollock et al., 2004), but the mechanism of CNX/CRT stimulation of ERp57 activity is unknown. Recently, the X-ray crystal structure of yeast PDI was reported at 2.4 A˚ (Tian et al., 2006). The structure of PDI shows an overall U shape with the catalytic sites of the two active thioredoxin domains facing each other. The interior surface of the U is hydrophobic and likely contributes to PDI’s ability to bind unfolded proteins and catalyze the formation of correct disulfides. An NMR characterization of the a and a0 domains of ERp57 confirmed that they adopt a thioredoxin-like fold (Pirneskoski et al., 2004; Silvennoinen et al., 2001, 2005). The structure of the luminal domain of CNX has also been determined by X-ray crystallography. The protein is characterized by two main structural components, a globular lectin domain and an extended proline-rich structure, termed the P domain (Schrag et al., 2001). The P domain is composed of four modules which form a hook-like arm. Previous studies using NMR spectroscopy and mutagenesis found that the tip of the P domain of CNX or CRT binds to ERp57 (Frickel et al., 2002; Leach et al., 2002; Pollock et al., 2004). Here we report structural studies of ERp57 using a combination of small-angle X-ray scattering (SAXS), X-ray diffraction, and NMR techniques. We show that the shape of ERp57 in solution is highly similar to the crystal structure of yeast PDI but that the 2.0 A˚ X-ray
Structure 1332
Table 1. Data Collection and Refinement Statistics Data Collection Space group Unit cell (A˚) Resolution range (A˚) Observed reflections Unique reflections Completeness (%) Overall (I/sI)
P21 a = 76.3, b = 62.4, c = 99.3 50–2.0 455,000 121,811 99.4 (96.1) 14.0 (2.6)
Refinement and Model Quality Resolution range (A˚) Rfree, number of reflections Rwork, number of reflections Rmsd bond lengths (A˚) Rmsd bond angles ( ) Average B factors (A˚2), number of atoms Protein Main chain atoms Side chain atoms Water Ramachandran plota % of residues in Most favorable regions Additional allowed regions a
32.7–2.0 0.250 (3,162) 0.194 (59,353) 0.025 1.932
43.1 (6,087) 42.1 (2,776) 43.6 (2,858) 45.7 (453)
93.1 6.9
Laskowski et al. (1993).
structure of the bb0 domains reveals significant differences in the CNX binding site, determined by NMR chemical shift mapping. Mutagenesis and in vitro folding assays demonstrate a requirement for the CNX-ERp57 interaction for efficient glycoprotein folding but not oxidoreductase activity. Results Structure of the bb0 Fragment We crystallized the bb0 fragment of ERp57 and determined its three-dimensional structure by X-ray diffrac-
tion using multiwavelength anomalous diffraction (MAD) phasing (Table 1). The crystals belong to the P21 space group and contain three molecules in the asymmetric unit. The structure was refined to 2.0 A˚ resolution (Figure 1A). The chains were traced from P134 (A135 in chain B) to L365. Five N-terminal residues corresponding to the cloning linker and the last 15 residues of the construct including K366–G376 from the linker following the b0 domain were not present in the electron density maps, indicating that these parts were disordered in the crystal. The structure contains two thioredoxin-like domains connected by a short linker. The b domain extends from A135 to F241, and the b0 domain from P245 to Y364 (Figure 2A). The two domains are very similar and can be superimposed using the backbone atoms of 96 residues with a root-mean-square deviation (rmsd) of 2.8 A˚. Each domain has a central five-stranded b sheet (Figure 1B). The b sheet has the order of b strands b1b3-b2-b4-b5, which are all parallel with the exception of strand b4. The four a helices of each domain are of various lengths and enclose the b sheets with two helices on either side. The a2 helix in the b0 domain (A273– D292) is significantly longer than any other helix in the structure. The long loop between strands b4 and b5 of domain b contains a short a-helical segment (S209– L211) that interacts with helix a2 of the b0 domain, forming an important part of the b-b0 interface. The b and b0 domains of ERp57 form an essentially rigid molecule, and as a result, all three chains in the asymmetric unit can be superimposed using their entire backbone with an rmsd of 0.8–1.0 A˚ (Figure 1C). The short three residue linker between these domains is involved in interdomain contacts and has temperature factors similar to proximal regions in both domains. A structural similarity search using the Dali database (Holm and Sander, 1995) gave scores that were generally higher for the b domain, which has higher sequence similarity to other thioredoxin-like domains. The best match was the second thioredoxin-like domain from Figure 1. X-Ray Crystal Structure of the bb0 Fragment of ERp57 (A) Crystal packing of the three molecules in the asymmetric unit cell. (B) Secondary structure of the b and b0 domains. (C) Overlay of the three molecules showing the rigidity of the bb0 fragment.
Crystal Structure of the bb0 Fragment of ERp57 1333
Figure 2. Comparison of Yeast PDI and Human ERp57 Structures (A) Domain structure of ERp57, showing the four thioredoxin-like domains and the C-terminal polylysine tract (triangle). (B) Structure-based sequence alignment. Secondary structure elements are shown in red for ERp57 and in light blue for PDI. Asterisks mark the locations of sequence insertions. Blue triangles mark lysine and arginine residues in ERp57 essential for binding the CNX P domain. Residues in the b0 domain of ERp57 are underlined. (C) Two structural superpositions of the b and b0 domains from PDI and ERp57. Asterisks mark the insertions in the b domain of ERp57 (dark red) and in the b0 domain of PDI (light blue).
calsequestrin (Z = 14.9); both domains could be superimposed using their backbone atoms of 101 residues with an rmsd of 1.9 A˚. The tandem of b and b0 domains superimposed on the calsequestrin structure with an rmsd of 3.7 A˚ over 204 residues, indicating similar relative orientation of these domains in calsequestrin. The similarity of PDI to calsequestrin has been previously noted (Tian et al., 2006). The closest structural homolog for the complete bb0 fragment of ERp57 was yeast PDI, which could be superimposed with an rmsd of 3.5 A˚ (Z = 14.1) over 199 residues. A structure-based sequence alignment of the bb0 fragments of human ERp57 and yeast PDI highlights the similarities and differences between the two proteins (Figure 2B). The biggest differences are two 6–7 residue inserts in the b domain and a long deletion in the b0 domain of ERp57 relative to PDI (Figure 2C). The first insert comprises helix a3. The second insert enlarges the b4-b5 loop in the b domain, with important structural and functional consequences. In ERp57, this loop contains residues that participate in CNX binding (see below). It also creates an extensive contact surface between the b and b0 domains, not seen in yeast PDI. As a result, the angle between the b and b0 domains in ERp57 differs by roughly 15 from the angle in PDI. This loop may also contribute to the rigidity of the b and b0 pair. In contrast to the b domain, which is smaller in yeast PDI, the b0 domain contains an additional helix as part of the loop following sheet b50 . Notwithstanding their structural similarity, the bb0 domains of ERp57 and PDI differ significantly in their surface charges. The largest difference is in helix a20 of the b0 domain, which carries a formal charge of +4 in ERp57 and 24 in PDI (Figure 2B). The positive charge of this region of ERp57 likely plays an important role in its specific binding to the highly negatively charged P domain of CNX.
Small-Angle X-Ray Scattering of Full-Length ERp57 In order to determine the overall shape and domain organization of ERp57, we carried out an analysis by small-angle X-ray scattering (SAXS). Small-angle scattering can be used to calculate the radius of gyration (Rg) and the maximum dimensions of the molecule (Dmax), but it is also sensitive to the overall shape of a molecule and can be used to generate molecular shape reconstructions and test structural models (Koch et al., 2003). We collected scattering data on full-length ERp57 with no evidence of aggregation up to 10 mg/ml (Figure 3A). Independent support of our SAXS data comes from two recent papers: a SAXS study of human PDI (Li et al., 2006) and the crystal structure of yeast PDI (Tian et al., 2006). Qualitatively, there is good agreement between the pair distance distribution functions P(r) calculated for ERp57 and human PDI with the presence of a marked, double-maximum centered around 40 A˚ in both curves (Figure 3A, insert) (Li et al., 2006). The theoretical scattering curve calculated using the yeast PDI crystal structure and the experimental curve for fulllength ERp57 are also very similar, with a low residual coefficient c of 4.76 (Figure 3A) (Svergun et al., 1995). The fit between the scattering curves was quite sensitive to changes in the angles between the domains in the atomic model; replacing the bb0 domains of PDI with the domains from ERp57 did not improve the fit. While the yeast PDI structure suggests that the a and a0 domains are relatively mobile (Tian et al., 2006), our SAXS data imply that, at low resolution, the average conformation of ERp57 in solution closely resembles the PDI crystal structure. The program GASBOR (Svergun et al., 2001) was used to generate structural models based on the ERp57 SAXS data. The program uses simulated annealing to match the experimental data to a computed scattering curve from a pseudoatom model. The models, consisting of
Structure 1334
Figure 3. Small-Angle X-Ray Scattering of Full-Length ERp57 (A) Fit of the experimental ERp57 scattering data (red) and the scattering curve calculated from the PDI X-ray structure (green). R(g) = 31.5. The insert shows the experimental P(r) plot. (B) Pseudo-atom reconstruction (GASBOR) of ERp57 superposed on the X-ray structure of yeast PDI (Tian et al., 2006). The domains of PDI are labeled. In the lower view, the a domain is hidden on the backside and not labeled.
one scatterer (pseudoatom) per amino acid, were aligned, averaged, and filtered based on occupancy using DAMAVER (Volkov and Svergun, 2003) to obtain a ‘‘most probable’’ model, which clearly shows four globular domains (Figure 3B). Superposition of the yeast PDI X-ray structure and DAMAVER model shows that the relative positions of the domains observed in the crystal are conserved in ERp57 in solution. Most notably, both structures have an overall U shape and place the two catalytically active domains a and a0 on one side of the molecule and the b and b0 domains on the other (Figure 3B). Identification of Individual Residues in the CNX Binding Site The high-quality spectra of the b0 domain of ERp57 allowed us to use NMR to identify the CNX binding site. The 1H-15N HSQC spectrum was characteristic of a folded, well-behaved protein with good signal dispersion (see Figure S1 in Supplemental Data available with this article online). Complete backbone resonance assignments for the b0 domain were obtained using standard triple-resonance experiments on a doubly (15N, 13 C) labeled sample. Titration of 15N-labeled b0 domain with unlabeled CNX P domain residues P310–K381 caused amide chemical shift changes for residues D266 (0.25 ppm), V267 (0.20 ppm), A273 (0.14 ppm), R280 (0.18 ppm), N281 (0.23 ppm), and R344 (0.17 ppm). These lie in the region following strand b20 , in the long helix a2 and the region preceding helix a40 (Figure 4B). Other residues in this region such as K274, N277, Y278, and W279 may be involved in binding, but their chemical shift changes could not be measured because of overlapping or missing peaks. Analysis of surface charge distribution reveals that the chemical shift changes localize to a large positively charged patch of residues (Figure 4C). This was not surprising, as the tip of the P domain is highly acidic and would be expected to interact electrostatically with a positively charged surface. This area also corresponds
to the highest sequence conservation in the bb0 domains across ERp57 proteins from other species (Figure 4D). Among the most conserved, surface-accessible residues are the basic amino acids: K214, K274, and R282. The corresponding surface and residues in the b0 domain of yeast PDI are negatively charged, consistent with its lack of interaction with CNX (Figure 2B). Mutagenesis Confirms the Binding Site Isothermal titration calorimetry (ITC) was used to measure the binding between different fragments of ERp57 and CNX (Figure 5). Full-length ERp57 binds to the twomodule P domain fragment of CNX with an affinity of 26 mM. The binding is exothermic, which reflects favorable van der Waals, hydrogen bond, and electrostatic interactions. These results are comparable to the binding of ERp57 to CRT, also measured by ITC (Frickel et al., 2002). The affinity of the P domain for the bb0 fragment was slightly higher than for full-length ERp57. This could reflect greater accessibility of the binding site in the bb0 fragment. The smallest ERp57 fragment that bound the P domain was the b0 domain, which bound with 50 times lower affinity. Together, these measurements demonstrate that the b0 domain contains the principal elements for ERp57 binding to CNX, while the b domain provides additional contacts necessary for high affinity. To verify the identification of the CNX binding site, we generated site-directed alanine substitution mutants and tested their binding using NMR and ITC. Two mutants, K274A and R282A, were designed to remove electrostatic interactions in the conserved, basic surface of the b0 domain. We generated a K214A mutant in the b domain to test the role of the b4-b5 loop in binding. As a negative control, a lysine residue (K332) on the opposite side of the b0 domain was mutated to alanine (Figure 5A). The specificity of the P domain-ERp57 interaction was also tested with an M347A mutation in the two-module P domain of CNX. This conserved residue occurs at the tip of the P domain and shifts in HSQC spectra upon bb0
Crystal Structure of the bb0 Fragment of ERp57 1335
half of the effect of deletion of the entire b domain. The K332A mutant showed binding affinity identical to wild-type, confirming that binding is specific and is not related to the overall charge of the bb0 domains.
Figure 4. Identification of the CNX Binding Site on ERp57 (A) Plot of chemical shift changes in the 15N-labeled ERp57 b0 domain upon addition of unlabeled CNX P domain. (B) Mapping of the chemical shifts onto the bb0 structure. Red indicates a large chemical shift change, white indicates no change, and gray indicates no data. The largest changes in the lower b0 domain are labeled. (C) Electrostatic potential of bb0 showing the abundance of basic residues in the P domain binding site. The right-hand surface is in the same orientation as the representation in (B); the left-hand surface is rotated by 180 . (D) Sequence conservation of surface residues on the ERp57 bb0 domains. Highly conserved residues in 50 protein sequences most closely related to human ERp57 are shown in green. The P domain binding site corresponds to the largest area of conserved amino acid residues.
binding (Figure S2), but its direct role in ERp57 binding has never been tested. All the mutant proteins were correctly folded, as verified by 1D NMR spectra and comparison with spectra for the corresponding wild-type fragments (Figures S3–S5). Mutation of either R282 in ERp57 or M347 in CNX completely abrogated binding, as measured by ITC and NMR titrations. Both mutations caused a >1000fold loss in affinity based on the upper limit (w5 mM) for detecting binding by NMR. Less dramatic effects were observed with the K214A mutation, which showed an 8-fold loss of affinity (Figure 5C). The side chain of K214 appears to be the most important recognition element in the b domain, as its deletion accounted for over
ERp57-CNX Interaction Is Essential for In Vitro Folding of RNase B We studied the competence of ERp57 and CNX for catalyzing protein folding in vitro using RNase B as a model substrate and low-pH native gels (Zapun et al., 1998). Folding occurs in an ER-like redox buffer containing the luminal domain of CNX and full-length ERp57. The required substrate Glc1 RNase B carries a single N-glycan of the form Glc1Man8–9GlcNAc2 at Asn34. This Glc1labeled RNase B is specifically bound by the lectin domain of CNX (Zapun et al., 1997), which stimulates the disulfide exchange activity of ERp57 to refold the substrate. As previously reported (Zapun et al., 1998), the folding reaction is dependent on the presence of the terminal glucose residue on the substrate. The commercial source of RNase B contains w90% of Man5–7GlcNAc2 RNase B, which cannot be labeled with 3H-glucose by UGGT. This unlabeled RNase B can be visualized in the gel assay by Coomassie blue staining and remains mostly unfolded after 80 min (Figure 5D, CB). 3H-glucosylated RNase B, when incubated with wild-type ERp57 and CNX, was largely folded after 40 min (Figure 5D, AR). The ability of the ERp57-CNX complex to mediate folding of 3H-labeled RNase B was completely dependent on a functional interaction between ERp57 and CNX. The point mutants K214A, K274A, and R282A of full-length ERp57 were compromised in their ability to fold Glc1-labeled RNase B (Figure 5D). Similarly, omission of either CNX or ERp57 from the reaction led to only a minor amount of spontaneous folding of RNase B, comparable to that observed with unlabeled RNase B (Figure 5D). In contrast, the K332A mutant, which does not affect CNX binding, catalyzed rapid folding. As a control for nonspecific effects, we examined the NMR spectra of the mutant proteins and characterized their thiol reductase activity against a small-molecule substrate, di(o-aminobenzyl)-labeled oxidized glutathione (diabz-GSSG) (Raturi et al., 2005). All of the ERp57 mutants were folded (Figures S4 and S5) and had full reductase activity (our unpublished data). Significantly, the addition of the luminal domain of CNX or the twomodule P domain did not stimulate ERp57 activity toward diabz-GSSG (Figure S6). Discussion The X-ray structure of the bb0 domains provides atomic resolution details of the CNX binding site on ERp57. This site consists of a large positively charged patch of highly conserved residues (K214, K274, and R282) that are essential for the function of ERp57. These residues engage the negatively charged residues of the CNX P domain, which along with the critical methionine residue (M347) comprise the tip of the P domain. While the b0 domain provides the majority of the binding site, the b4-b5 loop of the b domain affords additional contacts (K214) with the P domain, strengthening the interaction.
Structure 1336
Figure 5. Mutagenesis of the CNX-ERp57 Interaction Sites Showing the Requirement for the ERp57-CNX Interaction for Protein Refolding (A) Structure of ERp57 bb0 domains showing space-filling models of the residues mutated. (B) ITC thermogram of the interaction of the CNX two-module P domain binding to the bb0 domains of ERp57. (C) Table of ITC binding affinity constants for wild-type and mutant proteins. (D) In vitro refolding assays of RNase B with wild-type and mutant ERp57 proteins and the luminal domain of CNX. 3H-glucose-labeled RNase B was refolded for the time indicated and run on low-pH nondenaturing PAGE. Autoradiograms (AR) show the relative amounts of natively folded (N) and unfolded (U) 3H-Glc1-RNase B. For the wild-type assay, the Coomassie blue (CB)-stained gel shows the spontaneous folding of unlabeled RNase B. The right panels show autoradiograms at 80 min for four mutant ERp57 proteins and negative controls where CNX or ERp57 was omitted from the folding reaction.
Mutations previously shown to compromise the ERp57-CRT interaction (V267, Y269, K274, R280, N281, and F299) correlate well with our results (Russell et al., 2004). Among these, K274, R280, and N281 are surface exposed and likely hydrogen bond with the negatively charged P domain. Y269 is partially surface accessible and could form part of a hydrophobic pocket at the base of the b20 -a20 loop for the binding of M347 of CNX. On the other hand, V267 and F299 are both buried in the hydrophobic core and are unable to directly interact with the P domain. These mutations are both located under helix a20 and may affect binding via indirect contacts or through distortion of the b-b0 interdomain interface. Intriguingly, the CNX binding site in the b0 domain corresponds to the position of the catalytic cysteines in the b2-a2 loop of thioredoxins and the primary peptide binding site in the b0 domain of PDI (Pirneskoski et al., 2004). This supports earlier suggestions that the substrate binding domains in protein disulfide isomerases were created through gene duplication of catalytic domains and evolutionary modification of their active sites (loss of the catalytic cysteines) to favor protein binding (Kanai et al., 1998). Our SAXS results demonstrate that the solution structure of ERp57 is highly similar to the crystal structure of PDI from yeast. This places the CNX binding site on the
opposite face of ERp57 relative to the catalytic thiols (Figure 6). The long hook-like P domain is ideally shaped to curve around ERp57 and position the lectin binding site of CNX next to the catalytic a and a0 domains. The SAXS model also suggests that the positively charged C terminus of ERp57 is too far from the CNX binding site on the b0 domain of ERp57 to participate in direct binding of the P domain. This basic stretch, previously identified as interacting with CNX (Pollock et al., 2004), may interact with the phospholipid membrane, thus affecting binding of CNX and the efficiency of ERp57mediated protein folding. In PDI, the C-terminal sequence is acidic and forms a highly solvent-exposed a helix. How does the CNX-ERp57 complex work to fold proteins? The long distance (>40 A˚) between the CNX binding site and the catalytic cysteines argues against a CNX-induced conformational change or direct stimulation of the oxidoreductase activity of the a and a0 domains. Instead, the structure of ERp57 suggests that CNX serves to bind and position substrates next to the catalytic cysteines. The loss of either glycoprotein binding to CNX or CNX binding to ERp57 abrogates ERp57-catalyzed protein folding. Because the reactions done here are performed at low substrate concentrations, decreasing the KM of the folding reaction has the same effect as a stimulation of the reaction Vmax. More
Crystal Structure of the bb0 Fragment of ERp57 1337
from the Guinier plot by using PRIMUS, GNOM, and CRYSOL. The pair distance distribution function (P[r]) was calculated by using the program GNOM. Scattering curves and the goodness of fit (c) based on atomic models were obtained by using CRYSOL. Crystallization Crystals were obtained by the hanging drop equilibration of 14 mg/ml protein in 50 mM Tris-HCl, 0.15 M NaCl, 1 mM DTT (pH 7.5) against 30% (w/v) PEG3350, 0.1 M (NH4)2SO4, 0.1 M HEPES buffer (pH 7.5) for 1–3 days at 20 C.
Figure 6. Model of Full-Length ERp57 The a and a0 domains from the yeast PDI structure (light gray) were positioned on the bb0 fragment of ERp57 (dark gray) to show the relative positions of the thiol reactive cysteine residues (yellow) and the three basic residues, K214, K274, and R282, of the CNX binding site (blue).
kinetic studies are needed to test this hypothesis and elucidate the functional aspects of the CNX-ERp57 interaction. Experimental Procedures Protein Expression, Preparation, and Purification Full-length human ERp57 lacking the signal sequence (A24–L505) and the bb0 (P134–G376) and b0 (F241–G376) fragments were cloned as GST fusions between the BamH1 and Not1 sites of pGEX-6P-1 (Amersham Pharmacia) and expressed in Escherichia coli BL21(DE3). To prepare labeled protein for NMR experiments, cells were grown in M9 minimal medium with 15N-NH4Cl and 13C-U-glucose. Selenomethionine-labeled protein was produced in the E. coli methionine auxotroph DL41(DE3) grown on LeMaster medium (Hendrickson et al., 1990). The GST fusion proteins were purified by affinity chromatography on glutathione-Sepharose resin and tags were removed by cleavage with Prescission protease, leaving N-terminal Gly-Pro-Leu-Gly-Ser and C-terminal Ala-Ala-Ala-Ser extensions. The cleaved protein was additionally purified using size exclusion chromatography and concentrated using Centricon 3000 concentrators. The P domain two-module fragment of canine CNX (residues 310–381) was expressed and purified as described (Pollock et al., 2004). The luminal domain of canine CNX (Ou et al., 1995) and recombinant rat UDP-glucose:glycoprotein glucosyltransferase (UGGT) (Tessier et al., 2000) were expressed in a baculovirus/sf9 insect cell system as described previously (Zapun et al., 1998). Estimates of protein concentrations were based on theoretical UV absorption coefficients. Preparation of CNX and ERp57 Point Mutants Mutants were prepared by Quickchange site-directed mutagenesis (Stratagene) with mismatched primers and verified by DNA sequencing. SAXS Small-angle X-ray scattering data were collected at beamline 12.3.1 at the Advanced Light Source (Lawrence Berkeley National Laboratory, Berkeley, CA) with protein concentrations of 2, 5, and 10 mg/ml. Data were scaled by using the program PRIMUS (Konarev et al., 2003) and analyzed by the programs GNOM (Svergun, 1992) and CRYSOL (Svergun et al., 1995). Radii of gyration (Rg) were computed
Structure Solution and Refinement Three-wavelength MAD data were collected at beamline X8C at the National Synchrotron Light Source, Brookhaven National Laboratory (Table 1) and processed with HKL2000 (Otwinowski and Minor, 1997) and d*TREK (Pflugrath, 1999). The structure was determined using SHELXC/D/E (Sheldrick and Schneider, 1997) and ARP/ wARP (Perrakis et al., 1997). Data to 2.1 A˚ resolution at the peak wavelength (0.9795 A˚) were used to locate Se atoms with a figure of merit = 0.66. Density modification with the program ARP/wARP (Perrakis et al., 1997) allowed automated model building of w75% of the residues. The partial model obtained from ARP/wARP was extended manually with O (Jones et al., 1991) and improved by refinement using REFMAC (Murshudov et al., 1999) with the translation libration screw (TLS) option (Winn et al., 2003). Figures were prepared with PyMOL (Delano, 2002) and GRASP (Nicholls et al., 1991). The protein contact potential was calculated by PyMOL using default vacuum electrostatics ranging and color coded from 280 kT (red) to 80 kT (blue), where k is the Boltzmann constant and T is temperature. Coordinates have been deposited in the RCSB Protein Data Bank under ID code 2H8L. NMR Spectroscopy NMR resonance assignments of the b0 fragment of ERp57 in 50 mM Na2HPO4, 100 mM NaCl, 1 mM DTT (pH 6.8) were carried out using HNCACB, CBCA(CO)NH, HNCA, and 15N-noesy-HMQC experiments at 30 C at 500, 600, and 800 MHz. Amide signal assignments have been deposited in the Biological Magnetic Resonance Bank under ID code BMRB-7162. Isothermal Titration Calorimetry Measurements Experiments were carried out on a MicroCal VP-ITC titration calorimeter (MicroCal, Northampton, MA) in 50 mM Tris-HCl, 100 mM NaCl, 1 mM NaN3 (pH 6.8) at 15 C. Data were processed using ORIGIN and the binding isotherm was fit to a single-site binding model to determine the binding constant. RNase B Folding The RNase B folding assay was carried out as previously described (Zapun et al., 1998) with noted changes. The Glc1 form of RNase B was prepared from Man8GlcNAc2 or Man9GlcNAc2 RNase B (Sigma) using UGGT and 3H-UDP-glucose. Briefly, 500 mg of RNase B was unfolded and reduced for 20 min at room temperature in 0.1 M Tris-HCl (pH 8.0) containing 6 M guanidine HCl and 20 mM DTT. Unfolded RNase B was desalted in 0.1% formic acid on a NAP-5 column (Amersham Biosciences) and lyophilized. RNase B (z120 mM) was then glucosylated with 5–15 mM, 10–30 mCi/mmol [3H]UDP-glucose using a crude concentrate of recombinant rat UGGT expressed in sf9 insect cells (Tessier et al., 2000) in 25 mM HEPES, 5 mM CaCl2 containing 200 mM 1-deoxynojirimycin for 2 hr at 37 C. For each folding assay, the labeling reaction was adjusted to 150 mM NaCl, 2.5 mM CaCl2, 0.5 mM oxidized glutathione, 2 mM reduced glutathione containing 24 mM purified calnexin luminal domain (Ou et al., 1995) and 8 mM ERp57 or its mutants and incubated at room temperature. Time points were taken by blocking free thiols with 0.25 volumes of 0.5 M iodoacetamide in 1.5 M Tris-HCl (pH 8.7) and storing on ice before analyzing by low-pH nondenaturing PAGE (Creighton, 1979). Gels were Coomassie stained, destained, and treated with Amplify (Amersham Biosciences) for 30 min before drying and exposing to Kodak Biomax MS film for 24–96 hr. Supplemental Data Supplemental data include six figures showing the identification of the ERp57 binding site on the CNX P domain, 2D NMR spectrum
Structure 1338
of the b0 domain of CNX, 1D NMR spectra of the point mutants in the CNX P domain, ERp57 bb0 domains, full-length ERp57 showing that they are correctly folded, and results of GSSG reduction by ERp57. Supplemental data are available at http://www.structure.org/cgi/ content/full/14/8/1331/DC1.
Kanai, S., Toh, H., Hayano, T., and Kikuchi, M. (1998). Molecular evolution of the domain structures of protein disulfide isomerases. J. Mol. Evol. 47, 200–210.
Acknowledgments
Konarev, P.V., Volkov, V.V., Sokolova, A.V., Koch, M.H.J., and Svergun, D.I. (2003). PRIMUS: a Windows PC-based system for smallangle scattering data analysis. J. Appl. Crystallogr. 36, 1277–1282.
We thank Tara Sprules for help in running NMR experiments, Tudor Moldoveanu for advice in initial crystallization trials, Olivia Groover and Ekaterina Pomerantseva for technical assistance, and Arun Raturi and Bulent Mutus for the gift of diabz-GSSG. Grateful acknowledgment is given to Greg Hura, who collected the SAXS data on SYBYL beamline 12.3.1 at the Advanced Light Source, which is supported by the U.S. Department of Energy under contract DE-AC02-05CH11231. This work was supported by a grant from the Canadian Institutes of Heath Research to K.G. and D.Y.T. Received: May 4, 2006 Revised: June 10, 2006 Accepted: June 13, 2006 Published: August 15, 2006 References Bourdi, M., Demady, D., Martin, J.L., Jabbour, S.K., Martin, B.M., George, J.W., and Pohl, L.R. (1995). cDNA cloning and baculovirus expression of the human liver endoplasmic reticulum P58: characterization as a protein disulfide isomerase isoform, but not as a protease or a carnitine acyltransferase. Arch. Biochem. Biophys. 323, 397–403. Creighton, T.E. (1979). Intermediates in the refolding of reduced ribonuclease A. J. Mol. Biol. 129, 411–431. Delano, W.L. (2002). The PyMOL Molecular Graphics System (San Carlos, CA: DeLano Scientific). Ellgaard, L., and Frickel, E.M. (2003). Calnexin, calreticulin, and ERp57: teammates in glycoprotein folding. Cell Biochem. Biophys. 39, 223–247. Ellgaard, L., and Ruddock, L.W. (2005). The human protein disulphide isomerase family: substrate interactions and functional properties. EMBO Rep. 6, 28–32. Frickel, E.M., Riek, R., Jelesarov, I., Helenius, A., Wuthrich, K., and Ellgaard, L. (2002). TROSY-NMR reveals interaction between ERp57 and the tip of the calreticulin P-domain. Proc. Natl. Acad. Sci. USA 99, 1954–1959. Garbi, N., Tanaka, S., Momburg, F., and Hammerling, G.J. (2006). Impaired assembly of the major histocompatibility complex class I peptide-loading complex in mice deficient in the oxidoreductase ERp57. Nat. Immunol. 7, 93–102. Helenius, A., and Aebi, M. (2004). Roles of N-linked glycans in the endoplasmic reticulum. Annu. Rev. Biochem. 73, 1019–1049. Hendrickson, W.A., Horton, J.R., and LeMaster, D.M. (1990). Selenomethionyl proteins produced for analysis by multiwavelength anomalous diffraction (MAD): a vehicle for direct determination of threedimensional structure. EMBO J. 9, 1665–1672.
Koch, M.H., Vachette, P., and Svergun, D.I. (2003). Small-angle scattering: a view on the properties, structures and structural changes of biological macromolecules in solution. Q. Rev. Biophys. 36, 147–227.
Laskowski, R.A., MacArthur, M.W., Moss, D.S., and Thornton, J.M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. Leach, M.R., Cohen-Doyle, M.F., Thomas, D.Y., and Williams, D.B. (2002). Localization of the lectin, ERp57 binding, and polypeptide binding sites of calnexin and calreticulin. J. Biol. Chem. 277, 29686–29697. Li, S.J., Hong, X.G., Shi, Y.Y., Li, H., and Wang, C.C. (2006). Annular arrangement and collaborative actions of four domains of proteindisulfide isomerase: a small angle X-ray scattering study in solution. J. Biol. Chem. 281, 6581–6588. Murshudov, G.N., Vagin, A.A., Lebedev, A., Wilson, K.S., and Dodson, E.J. (1999). Efficient anisotropic refinement of macromolecular structures using FFT. Acta Crystallogr. D Biol. Crystallogr. 55, 247– 255. Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. Oliver, J.D., van der Wal, F.J., Bulleid, N.J., and High, S. (1997). Interaction of the thiol-dependent reductase ERp57 with nascent glycoproteins. Science 275, 86–88. Oliver, J.D., Roderick, H.L., Llewellyn, D.H., and High, S. (1999). ERp57 functions as a subunit of specific complexes formed with the ER lectins calreticulin and calnexin. Mol. Biol. Cell 10, 2573– 2582. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Ou, W.J., Cameron, P.H., Thomas, D.Y., and Bergeron, J.J. (1993). Association of folding intermediates of glycoproteins with calnexin during protein maturation. Nature 364, 771–776. Ou, W.J., Bergeron, J.J., Li, Y., Kang, C.Y., and Thomas, D.Y. (1995). Conformational changes induced in the endoplasmic reticulum luminal domain of calnexin by Mg-ATP and Ca2+. J. Biol. Chem. 270, 18051–18059. Parodi, A.J. (2000). Protein glucosylation and its role in protein folding. Annu. Rev. Biochem. 69, 69–93. Peaper, D.R., Wearsch, P.A., and Cresswell, P. (2005). Tapasin and ERp57 form a stable disulfide-linked dimer within the MHC class I peptide-loading complex. EMBO J. 24, 3613–3623. Perrakis, A., Sixma, T.K., Wilson, K.S., and Lamzin, V.S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple-refined dummy atomic models. Acta Crystallogr. D Biol. Crystallogr. 53, 448–455. Pflugrath, J.W. (1999). The finer things in X-ray diffraction data collection. Acta Crystallogr. D Biol. Crystallogr. 55, 1718–1725.
High, S., Lecomte, F.J., Russell, S.J., Abell, B.M., and Oliver, J.D. (2000). Glycoprotein folding in the endoplasmic reticulum: a tale of three chaperones? FEBS Lett. 476, 38–41.
Pirneskoski, A., Klappa, P., Lobell, M., Williamson, R.A., Byrne, L., Alanen, H.I., Salo, K.E., Kivirikko, K.I., Freedman, R.B., and Ruddock, L.W. (2004). Molecular characterization of the principal substrate binding site of the ubiquitous folding catalyst protein disulfide isomerase. J. Biol. Chem. 279, 10374–10381.
Hirano, N., Shibasaki, F., Sakai, R., Tanaka, T., Nishida, J., Yazaki, Y., Takenawa, T., and Hirai, H. (1995). Molecular cloning of the human glucose-regulated protein ERp57/GRP58, a thiol-dependent reductase. Identification of its secretory form and inducible expression by the oncogenic transformation. Eur. J. Biochem. 234, 336–342.
Pollock, S., Kozlov, G., Pelletier, M.F., Trempe, J.F., Jansen, G., Sitnikov, D., Bergeron, J.J., Gehring, K., Ekiel, I., and Thomas, D.Y. (2004). Specific interaction of ERp57 and calnexin determined by NMR spectroscopy and an ER two-hybrid system. EMBO J. 23, 1020–1029.
Holm, L., and Sander, C. (1995). Dali: a network tool for protein structure comparison. Trends Biochem. Sci. 20, 478–480.
Raturi, A., Vacratsis, P.O., Seslija, D., Lee, L., and Mutus, B. (2005). A direct, continuous, sensitive assay for protein disulphide-isomerase based on fluorescence self-quenching. Biochem. J. 391, 351–357.
Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119.
Rosenbaum, E.E., Hardie, R.C., and Colley, N.J. (2006). Calnexin is essential for rhodopsin maturation, Ca2+ regulation, and photoreceptor cell survival. Neuron 49, 229–241.
Crystal Structure of the bb0 Fragment of ERp57 1339
Russell, S.J., Ruddock, L.W., Salo, K.E., Oliver, J.D., Roebuck, Q.P., Llewellyn, D.H., Roderick, H.L., Koivunen, P., Myllyharju, J., and High, S. (2004). The primary substrate binding site in the b0 domain of ERp57 is adapted for endoplasmic reticulum lectin association. J. Biol. Chem. 279, 18861–18869. Schrag, J.D., Bergeron, J.J., Li, Y., Borisova, S., Hahn, M., Thomas, D.Y., and Cygler, M. (2001). The structure of calnexin, an ER chaperone involved in quality control of protein folding. Mol. Cell 8, 633– 644. Schrag, J.D., Procopio, D.O., Cygler, M., Thomas, D.Y., and Bergeron, J.J. (2003). Lectin control of protein folding and sorting in the secretory pathway. Trends Biochem. Sci. 28, 49–57. Sheldrick, G., and Schneider, T. (1997). SHELXL: high-resolution refinement. Methods Enzymol. 277, 319–343. Silvennoinen, L., Karvonen, P., Koivunen, P., Myllyharju, J., Kivirikko, K., and Kilpelainen, I. (2001). Assignment of 1H, 13C and 15N resonances of the a0 domain of ERp57. J. Biomol. NMR 20, 385–386. Silvennoinen, L., Koivunen, P., Myllyharju, J., Kilpelainen, I., and Permi, P. (2005). NMR assignment of the N-terminal domain a of the glycoprotein chaperone ERp57. J. Biomol. NMR 33, 136. Solda, T., Garbi, N., Hammerling, G.J., and Molinari, M. (2006). Consequences of ERp57 deletion on oxidative folding of obligate and facultative clients of the calnexin cycle. J. Biol. Chem. 281, 6219– 6226. Srivastava, S.P., Fuchs, J.A., and Holtzman, J.L. (1993). The reported cDNA sequence for phospholipase C a encodes protein disulfide isomerase, isozyme Q-2 and not phospholipase-C. Biochem. Biophys. Res. Commun. 193, 971–978. Svergun, D.I. (1992). Determination of the regularization parameter in indirect-transform methods using perceptual criteria. J. Appl. Crystallogr. 25, 495–503. Svergun, D., Barberato, C., and Koch, M.H.J. (1995). CRYSOL— a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Crystallogr. 28, 768– 773. Svergun, D.I., Petoukhov, M.V., and Koch, M.H. (2001). Determination of domain structure of proteins from X-ray solution scattering. Biophys. J. 80, 2946–2953. Tessier, D.C., Dignard, D., Zapun, A., Radominska-Pandya, A., Parodi, A.J., Bergeron, J.J., and Thomas, D.Y. (2000). Cloning and characterization of mammalian UDP-glucose glycoprotein: glucosyltransferase and the development of a specific substrate for this enzyme. Glycobiology 10, 403–412. Tian, G., Xiang, S., Noiva, R., Lennarz, W.J., and Schindelin, H. (2006). The crystal structure of yeast protein disulfide isomerase suggests cooperativity between its active sites. Cell 124, 61–73. Tu, B.P., and Weissman, J.S. (2004). Oxidative protein folding in eukaryotes: mechanisms and consequences. J. Cell Biol. 164, 341–346. Volkov, V.V., and Svergun, D.I. (2003). Uniqueness of ab initio shape determination in small-angle scattering. J. Appl. Crystallogr. 36, 860–864. Winn, M.D., Murshudov, G.N., and Papiz, M.Z. (2003). Macromolecular TLS refinement in REFMAC at moderate resolutions. Methods Enzymol. 374, 300–321. Zapun, A., Petrescu, S.M., Rudd, P.M., Dwek, R.A., Thomas, D.Y., and Bergeron, J.J. (1997). Conformation-independent binding of monoglucosylated ribonuclease B to calnexin. Cell 88, 29–38. Zapun, A., Darby, N.J., Tessier, D.C., Michalak, M., Bergeron, J.J., and Thomas, D.Y. (1998). Enhanced catalysis of ribonuclease B folding by the interaction of calnexin or calreticulin with ERp57. J. Biol. Chem. 273, 6009–6012.
Accession Numbers Coordinates of the structure of the bb0 domains of ERp57 have been deposited in the RCSB Protein Data Bank under ID code 2H8L. Amide signal assignments have been deposited in the Biological Magnetic Resonance Bank under ID code BMRB-7162.