Identification by Mutational Analysis of Amino Acid Residues Essential in the Chaperone Function of Calreticulin*

Calreticulin is a Ca2+-binding chaperone that resides in the lumen of the endoplasmic reticulum and is involved in the regulation of intracellular Ca2+ homeostasis and in the folding of newly synthesized glycoproteins. In this study, we have used site-specific mutagenesis to map amino acid residues that are critical in calreticulin function. We have focused on two cysteine residues (Cys88 and Cys120), which form a disulfide bridge in the N-terminal domain of calreticulin, on a tryptophan residue located in the carbohydrate binding site (Trp302), and on certain residues located at the tip of the “hairpin-like” P-domain of the protein (Glu238, Glu239, Asp241, Glu243, and Trp244). Calreticulin mutants were expressed in crt-/- fibroblasts, and bradykinin-dependent Ca2+ release was measured as a marker of calreticulin function. Bradykinin-dependent Ca2+ release from the endoplasmic reticulum was rescued by wild-type calreticulin and by the Glu238, Glu239, Asp241, and Glu243 mutants. The Cys88 and Cys120 mutants rescued the calreticulin-deficient phenotype only partially (∼40%), and the Trp244 and Trp302 mutants did not rescue it at all. We identified four amino acid residues (Glu239, Asp241, Glu243, and Trp244) at the hairpin tip of the P-domain that are critical in the formation of a complex between ERp57 and calreticulin. Although the Glu239, Asp241, and Glu243 mutants did not bind ERp57 efficiently, they fully restored bradykinin-dependent Ca2+ release in crt-/- cells. This indicates that binding of ERp57 to calreticulin may not be critical for the chaperone function of calreticulin with respect to the bradykinin receptor.

The endoplasmic reticulum (ER) 4 plays a critical role in the folding of secretory and membrane-associated proteins and glycoproteins (1). The ER contains several unique groups of molecular chaperones, including disulfide isomerase and its homolog ERp57, immunophilins, the Hsp70 and Hsp90 family of proteins, and the lectin-like chaperones calreticulin and calnexin (1). These chaperones and "folding sensors" play a central role in protein quality control in the ER (1,2). ERp57 catalyzes disulfide bond exchange in substrates that are bound to calreticulin or calnexin (3). Calreticulin and calnexin are homologous proteins that both interact with monoglucosylated, trimmed intermediates of the N-linked core glycans on newly synthesized glycoproteins (4 -6).
Biochemical and structural studies have demonstrated that calreticulin comprises three distinct structural domains: the amino-terminal N-domain, which is globular, the central P-domain, which is folded into an "extended arm," and the carboxyl-terminal C-domain (7). NMR (8), structural modeling (9,10), and biochemical studies (11,12) have indicated that, together, the globular N-domain and the "extended arm" P-domain of calreticulin form a functional protein-folding unit (9). The N-domain of calreticulin contains a carbohydrate binding site (13) and one disulfide bond (14). In a recent study, site-specific mutation of histidine residues in calreticulin revealed that His 153 , which is located in the N-domain of the protein, is essential for its chaperone function (12). The "extended arm" structure of the P-domain is involved in the association of calreticulin with ERp57, a thiol-disulfide oxidoreductase that is a close homolog of protein-disulfide isomerase (15)(16)(17).
Calreticulin deficiency is embryonic lethal, and cells derived from calreticulin knockout embryos have impaired Ca 2ϩ homeostasis (11). For example, in calreticulin-deficient cells, folding of the bradykinin receptor is altered, and this impairs its ability to initiate InsP 3 -dependent Ca 2ϩ release (11). The InsP 3 -dependent Ca 2ϩ release is, however, fully restored in calreticulin-deficient cells that express recombinant calreticulin (11,12). Therefore, in this study, we used restoration of bradykinin-dependent Ca 2ϩ release from the ER to compare the function of several site-specific calreticulin mutants with that of wild-type calreticulin. We focused on two cysteine residues that form a disulfide bridge in the N-domain of calreticulin, on a centrally located tryptophan residue in the carbohydrate binding site, and on several residues located at the tip of the "extended arm" of the protein's P-domain. We have identified two tryptophan residues that are critical for the chaperone function of calreticulin: Trp 302 , which is located in the carbohydrate binding pocket and Trp 244 , which is found at the tip of the P-domain's extended arm. In contrast, mutation of the cysteine residues disrupted the chaperone function only partially. We have also identified four amino acid residues at the tip of the "extended arm" of the P-domain that are critical in the formation of an ERp57-calreticulin complex.

EXPERIMENTAL PROCEDURES
Materials-Trypsin, malate dehydrogenase (MDH), bradykinin, and Dulbecco's modified Eagle's medium were from Sigma. Fetal bovine serum was from Invitrogen. SDS-PAGE reagents and molecular weight makers were from Bio-Rad. Effectene Transfection reagent and Ni 2ϩ -nitrilotriacetic acid-agarose beads were from Qiagen. CM5 sensor chips, the amine coupling kit, and the BIA evaluation analysis program were from BIAcore Inc. All chemicals were of the highest grade available.
Expression and Purification of Recombinant Proteins, Aggregation Assay, and Intrinsic Fluorescence Measurement-Proteins were expressed in Top10FЈ E. coli cells. His-tagged proteins were purified by one-step Ni 2ϩ -nitrilotriacetic acid-agarose affinity chromatography in native condition (12). Over 90% of the protein was purified to homogeneity by one-step Ni 2ϩ -nitrilotriacetic acid-agarose column chromatography. Protein concentration was determined by a Beckman System 6300 amino acid analyzer or by using Bio-Rad protein assay reagent using bovine serum albumin as a standard (12). The protein aggregation assay was carried out in a mixture containing varying amounts of wildtype and mutant calreticulin at room temperature (12,21). Samples were incubated at 42°C in 50 mM sodium phosphate, pH 7.5, and monitored for light scattering (12,21). The denatured MDH (0.25 M) was suspended in a buffer containing 10 mM Tris-HCl, pH 7.0, 150 mM NaCl, 5 mM CaCl 2 (21), followed by the addition of wild-type or mutant calreticulin (0.25 M). IgY was isolated from chicken egg yolk according to the protocol of the EGGstract IgY purification system and used for an aggregation assay (12). Light scattering was measured using a spectrofluorometer system C43/2000 (PTI) equipped with a temperaturecontrolled cell holder; the excitation and emission wavelengths were set to 320 and 360 nm, respectively (12). Intrinsic fluorescence measurements were performed at 25°C in a spectrofluorometer system C43/ 2000 as described (12). The excitation wavelength was set to 286 nm, and the range of emission wavelength was set to 295-450 mm. The effect of Zn 2ϩ on the intrinsic fluorescence of protein was evaluated at a wavelength of 334 nm.
Proteolytic Digestions and CD Analysis-10 g of purified, recombinant wild-type and mutant calreticulin expressed in E. coli were incubated with trypsin at 1:100 (trypsin/protein; w/w), with a final trypsin concentration of 10 mg/ml (22). The proteins were separated in SDS-PAGE (10% acrylamide) and stained with Coomassie Blue (12). CD analysis was performed at 25°C using a Jasco J720 spectropolarimeter (22).
Surface Plasmon Resonance Analysis-The sensor chip was activated, coupled, and blocked using the amine coupling kit from BIAcore. Specifically, two lanes on a CM5 chip were activated using a 1:1 mixture of N-hydroxysuccinimide/N-ethyl-NЈ-(3-dimethylaminopropyl)-carbodiimide hydrochloride for 7 min at a flow rate of 5 ml/min. Purified calreticulin or calreticulin mutants were then immobilized in 10 mM FIGURE 1. Expression of calreticulin mutants in calreticulin-deficient mouse embryonic fibroblasts. Western blot analysis and immunostaining of calreticulin in mouse embryonic fibroblasts was carried out as described under "Experimental Procedures" (12). Upper panel, proteins from calreticulin-deficient cells expressing different calreticulin mutants were lysed, separated by SDS-PAGE, transferred to nitrocellulose membrane, and probed with rabbit anti-hemagglutinin (anti-HA tag) antibody. Lower panel, calreticulin-deficient cells expressing different mutants were probed with anti-calreticulin antibody. Scale bar, 10 m. sodium acetate buffer, pH 4, at a flow rate of 5 ml/min, and this was followed by blocking with 1 M ethanolamine, pH 8.5, for 7 min at a flow rate of 5 ml/min. A control channel on the sensor chip was activated and blocked, using amino-coupling reagent, without immobilization of protein. The protein bound to the control channel was subtracted from the protein bound to the other channels to determine specific binding. The flow buffer contained 20 mM Tris, pH 7, 135 mM KCl, 2 mM CaCl 2 , 0.05% Tween 20, 200 M phenylmethylsulfonyl fluoride, 100 M benzamidine, and a mixture of protease inhibitors (23). Binding was carried out at 20°C at a flow rate of 30 ml/min. The binding data were analyzed by nonlinear least squares curve fitting. Association and dissociation constants, k a and k d , were generated from the association and dissociation curves from the BIAcore experiments. ERp57 was expressed in E. coli and purified as previously described (24).
Bradykinin-induced Ca 2ϩ Release in Cells Expressing Calreticulin Mutants-Calreticulin-deficient cells have inhibited bradykinindependent Ca 2ϩ release, which is restored by expression of full-length recombinant calreticulin ( Fig. 2A) (11). We took advantage of this, using the recovery of bradykinin-dependent Ca 2ϩ release from the ER, in crt Ϫ/Ϫ cells, as a model system in which to compare the function of calreticulin and the various mutants we had generated. Fig. 2A shows that, as expected, bradykinin caused a rapid and transient increase in the cytoplasmic Ca 2ϩ concentration in wild-type cells but not in crt Ϫ/Ϫ cells (11). Also, as expected, the expression of recombinant calreticulin in the crt Ϫ/Ϫ cells fully restored the bradykinin-dependent Ca 2ϩ release ( Fig. 2A).
Next, we measured bradykinin-dependent Ca 2ϩ release in calreticulin-deficient cells (crt Ϫ/Ϫ ) expressing the various calreticulin mutants. In crt Ϫ/Ϫ cells expressing recombinant calreticulin with mutations at the tip of extended arm of the P-domain, bradykinin-dependent Ca 2ϩ release was partially recovered ( Fig. 2A). Specifically, expression of the mutants Glu 238 , Glu 239 , Asp 241 , and Glu 243 restored bradykinindependent Ca 2ϩ release to 55, 75, 73, and 78% of the control, respectively ( Fig. 2A). In contrast, expression of the Trp 302 calreticulin mutant did not restore bradykinin-dependent Ca 2ϩ release, and expression of Trp 244 mutant restored only Ͼ10% of the activity ( Fig. 2A). However, Trp 244 and Trp 302 calreticulin mutants had no effect on Ca 2ϩ capacity of the ER stores (Fig. 2B). The expression of SERCA and InsP 3 receptors (Fig. 2C) was not altered in crt Ϫ/Ϫ cells expressing Trp 244 and Trp 302 mutants. In digitionin-permeabilized cells, InsP 3 -induced Ca 2ϩ release from ER was indistinguishable in wild-type cells and in crt Ϫ/Ϫ cells expressing Trp 244 and Trp 302 calreticulin mutants (data not shown), indicating that InsP 3 receptors were functional in these cells. Furthermore, flow cytometry analysis revealed that the cell surface localization of the bradykinin receptor was also not affected by in cells the expression of Trp 244 and Trp 302 calreticulin mutants (Fig. 2D). We concluded that Trp 244 and Trp 302 residues play an essential role in the function of calreticulin as far as folding of the bradykinin receptor is concerned.
Last, we investigated the functional consequence of mutating the cysteine residues involved in formation of the disulfide bridge in the N-domain of calreticulin (Cys 88 and Cys 120 ). Fig. 2A shows that bradykinin-dependent Ca 2ϩ release was increased by only 38 and 40%, compared with the control, in cells expressing the Cys 88 and Cys 120 mutants, respectively. These data demonstrate that the cysteine residues are also critical in determining the ability of calreticulin to function properly as a chaperone.
Effect of Calreticulin Mutants on Thermal Aggregation of MDH-In order to further examine the role of the residues Cys 88 , Cys 120 , Trp 244 , and Trp 302 in the chaperone activity of calreticulin, we exploited an in vitro assay used previously (12,21). In this assay, we measured the ability of calreticulin to prevent thermal aggregation of MDH, a nonglycosylated substrate (12,21). MDH is susceptible to heat-induced aggregation at 42°C, as measured by light scattering, and this can be partially reversed in the presence of molecular chaperones (12, 21, 26 -28). For example, full-length recombinant calreticulin effectively prevents MDH thermal-induced aggregation in vitro (12,21). Using this assay, we compared the ability of mutant calreticulins to prevent thermal aggregation of MDH. To do this, we expressed the recombinant proteins in E. coli and purified them (Fig. 3A).
As previously observed, heat-induced aggregation of MDH was significantly reduced in the presence of wild-type calreticulin (21) (Fig.  3A). The Cys 88 , Cys 120 , Glu 238 , Glu 239 , Asp 241 , and Glu 243 calreticulin mutants also all prevented heat-induced aggregation of MDH and to an extent similar to that observed for the wild-type protein (data not shown). In contrast, the Trp 302 and Trp 244 calreticulin mutants were not effective in preventing heat-induced aggregation of MDH (Fig. 3A), suggesting that these residues are critical in the ability of calreticulin to prevent MDH aggregation. We also tested the effectiveness of calreticulin mutants in preventing aggregation of a glycosylated substrate, IgY (21). As expected, full-length calreticulin effectively prevented aggregation of chemically denatured IgY (Fig. 3B). In keeping with the results of the MDH-refolding experiments (Fig. 3A), Trp 302 and Trp 244 calreticulin mutants did not efficiently prevent aggregation of IgY (Fig. 3B).
Structural Analysis of Calreticulin Mutants-Recently, we have shown that mutation of the residue His 153 in calreticulin caused significant conformational changes in the protein (12). Since the mutations used in this study may also have affected the conformation of calreticulin, we compared the conformations of the mutants with wild-type calreticulin, using intrinsic fluorescence and limited proteolysis analysis.
The intrinsic fluorescence emission of a protein is affected by the movement of charged groups and by changes in hydrophobicity, and it provides information on changes in tertiary structure. Fluorescence emission spectra were measured for wild-type calreticulin and for calreticulin mutants at varying concentrations of Zn 2ϩ . Fig. 4A shows that calreticulin has an emission maximum at 334 nm and that there was a progressive increase in the intensity of fluorescence in the presence of increasing concentrations of Zn 2ϩ (12). This is indicative of Zn 2ϩ -dependent conformational changes in the protein.
Fluorescence emission spectra for the calreticulin mutants Cys 88 , Cys 120 , Glu 238 , Glu 239 , Asp 241 , Glu 243 , and Trp 302 were indistinguishable from the spectrum observed for the wild-type protein (data not shown),

. Effects of Trp 302 and Trp 244 calreticulin mutants on thermal aggregation of MDH (A) and IgY (B).
MDH or IgY was incubated in the presence or absence of wildtype calreticulin or calreticulin mutants as indicated, followed by monitoring aggregation at 360 nm (21). Box, SDS-PAGE of recombinant wild-type (wt) and Trp 244 and Trp 302 mutants of calreticulin used for aggregation and conformational change studies is shown.
indicating that mutation of these residues in calreticulin did not significantly affect the conformation of the protein. In contrast, the mutant Trp 244 showed larger changes in intrinsic fluorescence at lower Zn 2ϩ concentrations than did wild-type calreticulin (Fig. 4B), the maximal change being observed by 0.25 mM Zn 2ϩ . The effect of Zn 2ϩ on the precise conformation of calreticulin was enhanced in the absence of the Trp 244 residue (Fig. 4C). Ca 2ϩ -dependent changes in intrinsic fluorescence were also measured for all mutants. There was no significant difference in the Ca 2ϩ -dependent intrinsic fluorescence observed in wild-type calreticulin, in the mutant Trp 244 (Fig. 4, D and E), and in the mutants Cys 88 , Cys 120 , Glu 238 , Glu 239 , Asp 241 , Glu 243 , Trp 302 (data not shown).
To determine whether calreticulin mutations affected conformational changes in the protein, we carried out CD analysis of the purified proteins. CD spectra for Glu 238 , Glu 239 , Asp 241 , and Glu 243 were indistinguishable from the spectra obtained for wild-type calreticulin (data not shown). However, the CD spectra of wild-type calreticulin and the Cys 88 , Cys 120 , Trp 244 , and Trp 302 mutants were very similar in shape but not identical (Fig. 4F). Trp 244 and Trp 302 mutants exhibited a slight increase in ␤-sheet-␤-turn, whereas Cys 80 and Cys 120 showed a slight decrease in ␤-sheet-␤-turn (Fig. 4F). These minor changes in conformation of Cys 88 , Cys 120 , Trp 244 , and Trp 302 mutations may have contributed, in part, to inhibition of their chaperone function.
Ca 2ϩ -dependent resistance to trypsin is an excellent measure of calreticulin folding (12,22). Therefore, we also investigated possible conformational differences between calreticulin and the calreticulin mutants using limited proteolysis. Specifically, we examined the effect of mutating the residues Cys 88 , Cys 120 , Glu 238 , Glu 239 , Asp 241 , Glu 243 , Trp 244 , and Trp 302 on calreticulin's resistance to trypsin. In agreement with our earlier observations (12,22), in the presence of 2 mM Ca 2ϩ , wild-type calreticulin is relatively resistant to trypsin digestion (trypsin/ calreticulin ϭ 1:100) (Fig. 5A). Mutation of Glu 239 at the tip of the extended arm of the P-domain of calreticulin (Fig. 5B) and of Trp 302 in the carbohydrate binding region (Fig. 5C) had no effect on the sensitivity of the protein to trypsin digestion. Both mutants were relatively resistant to digestion, similar to the wild-type protein. The mutants Cys 88 , Cys 120 , Glu 238 , and Asp 241 were also all resistant to trypsin digestion (data not shown). In contrast, the Ca 2ϩ -dependent trypsin resistance of calreticulin was lost when the residues Glu 243 and Trp 244 were mutated (Fig. 5, D and E). Calreticulin binds Ca 2ϩ at high affinity/low capacity and low affinity/high capacity sites (20), and it is possible that changes in Ca 2ϩ binding to the protein could alter its sensitivity to trypsin. There-fore, we measured Ca 2ϩ binding to calreticulin and the calreticulin mutants. Ca 2ϩ binding to the high affinity and to the high capacity Ca 2ϩ binding sites in calreticulin was not affected in any of the calreticulin mutants used in the present studies (Table 1). We conclude that the increased susceptibility to Ca 2ϩ -dependent proteolysis seen in the mutants Glu 243 and Trp 244 indicates that mutation of these amino acid residues leads to significant alteration in calreticulin's structure.
Association of ERp57 with Calreticulin-Calreticulin associates with ERp57, a thiol-disulfide oxidoreductase and a close homolog of proteindisulfide isomerase (15,29,30). ERp57 binds to the tip of the hairpin-like P-domain of calreticulin (16,17) and promotes disulfide bond formation in glycoprotein substrates (3,31). Several of the calreticulin residues mutated in this study (Glu 238 , Glu 239 , Asp 241 , Glu 243 , and Trp 244 ) are within the region believed to contain the ERp57 binding site (16,17,32). We made a number of attempts to show ERp57-calreticulin interactions using GST pull-down experiments. Unfortunately, these experiments were difficult to control and reproduce because a significant amount of both calreticulin and ERp57 binds nonspecifically to the nickel affinity beads. Therefore, we used BiAcore sensor chips to test whether any of the mutated residues are involved in promoting ERp57calreticulin association. Recombinant calreticulin and the calreticulin mutants were immobilized to a BIAcore sensor chip, and the binding of recombinant ERp57 in the flow buffer was measured. Figure 6 shows that ERp57 bound to calreticulin with high affinity. Mutation of the residue Glu 238 in calreticulin, located at the tip of the P-domain, had no

Amino Acids Essential in Calreticulin Chaperone Function
significant effect on ERp57 binding to the immobilized protein (Fig. 6). In contrast, the mutants Glu 239 and Glu 243 showed significantly reduced binding of ERp57, and the mutants Asp 241 and Trp 244 did not bind ERp57 at all (Fig. 6). Interestingly, mutation of the Trp 302 residue, which is localized in the N-terminal carbohydrate binding site away from the P-domain of calreticulin, significantly enhanced ERp57 binding to the protein (Fig. 6).

DISCUSSION
In this study, we used calreticulin-deficient cells to carry out a functional analysis of specific amino acid residues in calreticulin, a lectin-like chaperone in the ER. In calreticulin-deficient cells, folding of the bradykinin receptor is altered, which impairs its ability to initiate InsP 3 -dependent Ca 2ϩ release (11). This is an excellent model to investigate the function of calreticulin and to monitor the effect of specific amino acid mutations. Previously, we have used this approach to identify a histidine residue essential for the chaperone function of calreticulin (12). In the present study, we focused on amino acid residues located in the carbohydrate binding pocket of calreticulin, on the disulfide bridge in the N-domain, and on several residues at the tip of the extended arm of the P-domain. We report identification of two tryptophan residues that appear to be critical for chaperone function of calreticulin: Trp 302 located in the carbohydrate binding pocket and Trp 244 found at the tip of the extended arm of the P-domain. Furthermore, we identified four other amino acid residues at the tip of the extended arm of the P-domain of calreticulin that are important in the binding interaction between ERp57 and calreticulin. We also found that mutation of the cysteine residues that form the disulfide bridge also disrupted the chaperone function of calreticulin by ϳ60%.
Calreticulin and calnexin bind their glycoprotein substrates primarily via monoglucosylated glycans (Glc 1 Man 7 -9 GlcNAc 2 ). The three-dimensional structure of the luminal domain of calnexin shows a binding site for the terminal glucose moiety of the carbohydrate within the globular N-domain of the protein (10). Molecular modeling of calreticulin, based on the structure of calnexin, indicates that these proteins may have a similar carbohydrate binding pocket (9) (Fig. 7). Structural and modeling analyses indicate that the glucose binding pocket in both proteins contains a centrally located tryptophan residue and several charged amino acid residues (9,10). Recent in vitro mutational analysis of calreticulin indicates that the residues Tyr 109 , Met 131 , Asp 135 , and Asp 317 , localized in the putative carbohydrate binding site, are involved in binding of the Glc 1 Man 9 GlcNAc 2 sugar moiety (13). Asp 135 and Tyr 109 are probably the most important contributors toward polar interactions between the sugar and calreticulin (13). Here we show that the centrally located Trp 302 residue in calreticulin is essential for the chaperone function of the protein in vivo, probably because of a critical role in carbohydrate binding. Mutation of Trp 302 may produce important structural changes in the sugar binding site and/or in the N-terminal domain, which lead to the loss of the chaperone function of calreticulin. This is supported by our observation that the Trp 302 mutant failed to prevent thermal aggregation of MDH. Also, the residue Trp 302 may form a hydrogen bond with the Man 3 of the sugar moiety (13). Interestingly, although calreticulin forms functional complexes with ERp57 via its P-domain (16,17,32), the Trp 302 calreticulin mutant exhibited greatly increased binding of ERp57. This indicates that a single amino acid mutation in the globular N-domain of calreticulin affects the structure and function of the distant P-domain.
The P-domain of calreticulin forms an unusual extended arm-like structure, where the C and N terminus of the domain are in close proximity (8). The structure is stabilized by three short antiparallel ␤-sheets. Three small hydrophobic clusters containing two tryptophan rings, including Trp 244 , provide additional stability (8). In this study, mutation of Trp 244 resulted in one of the most severe effects on calreticulin function. The Trp 244 mutant of calreticulin was unable to restore bradykinindependent Ca 2ϩ release in calreticulin-deficient cells, it did not prevent thermally induced aggregation of MDH, and it did not bind ERp57. Biophysical studies of this mutant of calreticulin suggested that conformational changes might be responsible for its loss of chaperone activity and lost ability to form a complex with ERp57. Fig. 8 shows that Trp 244 plays a critical role in determining the structure of the tip of the extended arm in the P-domain of calreticulin (8). The indole rings of  residues Trp 244 and Trp 236 provide essential stability for the P-domain and the region involved in binding of ERp57 (Fig. 8A). The mutant Trp 244 has a disturbed structure in this region, probably creating an unstable cavity (Fig. 8, B and D). This, in turn, must have a significant effect on ability of the substrate (carbohydrate) to bind to the globular N-domain of calreticulin. NMR and biochemical analyses of the P-domain of calreticulin indi-cate that the tip of the extended arm binds ERp57 (16,17,32). NMR studies revealed that ERp57 may bind to a region of calreticulin encompassing residues 251-255 (16). We investigated several amino acid residues at the tip of the P-domain to determine their role in ERp57-calreticulin interactions (for the location of specific residues, see Fig. 7). Mutation of Glu 238 did not effect ERp57 binding to calreticulin, whereas the mutants Asp 241 and Trp 244 did not bind any measurable ERp57, and the mutants Glu 239 and Glu 243 showed significantly reduced binding of ERp57. We conclude that the negatively charged residues Glu 239 , Asp 241 , and Glu 243 and the residue Trp 244 are essential for formation of the ERp57calreticulin complex. The residues Glu 239 , Asp 241 , and Glu 243 are located on the concave side of the tip of the P-domain, and they probably provide electrostatic forces critical in formation of the complex with ERp57. It is unlikely that Trp 244 is directly involved in ERp57-calreticulin interactions, but rather it seems to play a in role maintaining the structural stability of the tip of the P-domain. It was surprising to find that although the mutants Glu 239 , Asp 241 , and Glu 243 did not bind ERp57 efficiently, they were able to fully restore bradykinin-dependent calcium release in crt Ϫ/Ϫ cells, suggesting that ERp57 binding to calreticulin may not be critical for its chaperone function as far as the bradykinin receptor is concerned. Glutathione S-transferase pull-down analysis with the P-domain of calnexin indicates that ERp57, as well as binding to calreticulin, also binds to the P-domain of calnexin (32). Interestingly, NMR analysis showed that certain single amino acid mutations on the tip of the P-domain of calnexin had no significant effect on ERp57 binding; however, double mutants did (30). In calnexin, the residues Asp 342 , Asp 344 , Asp 346 , Asp 348 , Glu 350 , and Glu 352 are all involved, the residues Asp 344 and Glu 352 being the most important (30). This is in agreement with our data that implicate Glu 243 (in calnexin Asp 352 ) and Glu 239 (in calnexin Asp 344 ) as critical residues for ERp57 binding to calre-  ticulin. Thus, it appears that clusters of negatively charged residues at the tip of the P-domain, in both calreticulin and calnexin, play an important role in promoting and stabilizing association with ERp57.
In conclusion, a number of cell biological, molecular, and biophysical studies have been used to identify functionally important regions and amino acid residues in calreticulin (12) (this study). Certain amino acid residues located directly in (Trp 302 ) or near (His 153 ) (12) the carbohydrate pocket are critical for the chaperone function of the protein. Further, modification of a residue(s) in the carbohydrate binding region may have a profound effect on the structural and functional properties (mobility and stability) of the P-domain, as documented by increased ERp57 binding by the calreticulin mutant Trp 302 . Conversely, specific amino acid residues located in the P-domain (Trp 244 ) may influence the function of the lectin-like N-domain of calreticulin. Finally, our data here indicate that the interaction between ERp57 and calreticulin, and presumably with calnexin, may not be a prerequisite for folding of all substrates.