Ca²⁺-dependent redox modulation of SERCA 2b by ERp57

Changes in the cytosolic Ca²⁺ concentration control many essential cellular responses (Dolmetsch et al., 1997; Berridge et al., 1998; Li et al., 1998; Hajnoczky et al., 2003). To coordinate these functions Ca²⁺ signals are flexible, yet precisely regulated. At intermediate IP₃ concentrations (i.e., ∼300 nM), cytosolic Ca²⁺ oscillates in Xenopus oocytes (Camacho and Lechleiter, 1993, 1995; John et al., 1998; Roderick et al., 2000; Falcke et al., 2003). Ca²⁺ oscillations are triggered by opening of the inositol 1,4,5-trisphosphate receptor (IP₃R) channel, whereas reuptake of the cation into the ER lumen is due to the activity of the sarco ER calcium ATPases (SERCAs). Consistent with this model, overexpression of SERCA pumps increases IP₃ induced Ca²⁺ oscillations (Camacho and Lechleiter, 1993). SERCA pumps play a critical role not only in clearing cytosolic Ca²⁺, but also in maintaining ER Ca²⁺ concentrations ([Ca²⁺]_ER). At rest, the free [Ca²⁺]_ER is ∼300 μM, whereas cytosolic Ca²⁺ concentration is three to four orders of magnitude lower (∼5–50 nM; Pozzan et al., 1994).

The lumen of the ER is a specialized protein-folding environment. It contains molecular chaperones such as calreticulin (CRT), calnexin (CLNX), and ER protein 57 (also known as ER-60, GRP58; ERp57; High et al., 2000; Ellgaard and Helenius, 2003; Kostova and Wolf, 2003; Schrag et al., 2003). Optimal [Ca²⁺]_ER is necessary for protein folding (Ashby and Tepikin, 2001). ER Ca²⁺ depletion inhibits protein folding and maturation (Hardman and Wetmore, 1996; Chen et al., 1997) and facilitates protein degradation (Ramsden et al., 2000). Ca²⁺ can also regulate the formation of chaperone complexes in the ER (Corbett et al., 1999).

ERp57 is a ubiquitous ER thiol-dependent oxidoreductase that promotes the formation of intra- or intermolecular disulfide bonds during glycoprotein folding (Marcus et al., 1996; High et al., 2000; Ellgaard and Helenius, 2003; Kostova and Wolf, 2003; Schrag et al., 2003). ERp57 has also been identified as a key component in the assembly of class I major histocompatibility complexes (Hughes and Cresswell, 1998; Lindquist et al., 1998; Morrice and Powis, 1998; Antoniou et al., 2002; Dick et al., 2002; Bouvier, 2003). Critical to our paper is the demonstration, firmly established in the literature, of a specific interaction between either CLNX or CRT with ERp57 (Oliver et al., 1997, 1999; Van der Wal et al., 1998; Zapun et al., 1998; High et al., 2000; Frickel et al., 2002).

Our group discovered that CRT, as well as CLNX, inhibited Ca²⁺ oscillations using the Xenopus oocyte system (Camacho and Lechleiter, 1995; John et al., 1998; Roderick et al., 2000). The precise molecular mechanism responsible for this inhibitory effect is not completely known, however, CRT and CLNX may directly regulate SERCA 2b or they may recruit other enzymes such as ERp57 to cause the effect. Two conserved cysteines in the longest ER facing loop 4 (L4) of SERCA 2b are potential targets of ERp57. In this paper, we address the issues of whether ERp57 modulates Ca²⁺ oscillations through an interaction with L4 thiols; whether this interaction requires the enzymatic activity of ERp57; whether the interaction is specific and Ca²⁺ dependent; and whether CRT is required to recruit ERp57 to L4.

We used a confocal Ca²⁺ oscillation assay as a tool to investigate the modulation of the SERCA 2b pump activity. In this assay, increases in Ca²⁺ oscillation frequency (shorter period between oscillations and/or decay time [t_1/2] of individual waves) reflect increased Ca²⁺ ATPase activity and vice versa (Camacho and Lechleiter, 1993, 1995, 2000; John et al., 1998; Roderick et al., 2000; Falcke et al., 2003). To test whether ERp57 modulates SERCA 2b activity in Xenopus oocytes, we coexpressed SERCA 2b with ERp57. We found that coexpression reduced the frequency of Ca²⁺ oscillations compared with overexpression of SERCA 2b by itself. A histogram of period and t_1/2 shows that both were significantly prolonged suggesting that Ca²⁺ uptake into the ER is reduced (Fig. 1 A). Western blots demonstrated that expression levels of SERCA 2b were the same regardless of whether ERp57 was coexpressed with the pump (Fig. 1 B). These results are consistent with the hypothesis that ERp57 modulates ER thiol groups in SERCA 2b.

To test whether the thiol groups in L4 are important in modulating Ca²⁺ oscillations enhanced by SERCA 2b, we imaged Ca²⁺ activity in oocytes overexpressing either the oxidizable form of SERCA 2b or two L4 cysteine–deficient mutants (SERCA 2b-C1SC2S and SERCA 2b-C1AC2A) that cannot form a disulfide bond and are a constitutively reduced form. Because these mutants did not express as efficiently as SERCA 2b in oocytes, we lowered expression levels of the pump to match expression levels of the mutants (Fig. 2 A). Under these conditions, we found that both mutants exhibited a higher frequency of Ca²⁺ oscillations and more rapid cytosolic Ca²⁺ uptake (i.e., shorter t_1/2) than SERCA 2b (Fig. 2 B). We also overexpressed a single missense mutation found in a Darier disease pedigree that involves a cysteine to glycine mutation in L4 (SERCA 2b-C875G; Ruiz-Perez et al., 1999; Sakuntabhai et al., 1999; Ahn et al., 2003). When overexpressed in the oocytes, this mutant also exhibited a higher frequency of Ca²⁺ oscillations than the wild-type SERCA 2b (unpublished data). Together, with the previous data, our results suggest that ERp57 promotes disulphide bond formation in L4, thereby reducing the activity of the Ca²⁺ pump. When the disulfide bridge in L4 is disrupted, the Ca²⁺ ATPase appears to exhibit higher pump activity.

Two conserved thioredoxin motifs in ERp57 are thought to be responsible for catalytic activity of the enzyme (Hirano et al., 1995). One motif is positioned near the NH₂-terminus (T1) and the other is near the COOH-terminus (T2). Two single mutants, ERp57-T1 and ERp57-T2, and a double mutant, ERp57-T1T2, were generated by mutagenesis of the relevant cysteines into serines in each motif (Fig. 3 A). To characterize the mutant ERp57 proteins, we generated GST fusion proteins and measured in vitro the catalytic activity using an insulin turbidity assay (Holmgren, 1979; Hirano et al., 1995). Protein disulfide isomerase (PDI), another ER resident oxidoreductase, was used as a positive control for this assay. We cloned the rat isoform and made purified GST-PDI. Both GST-PDI and GST-ERp57 exhibited strong enzymatic activity, although the latter exhibited slower kinetics. In comparison to ERp57, the ERp57-T2 mutant exhibited ∼70% activity, whereas the ERp57-T1 mutant exhibited only ∼30% activity. The double mutation ERp57-T1T2 completely abolished ERp57 enzymatic activity. There was no activity when GST alone or enzyme was absent in the assay (Fig. 3 B). Purified GST and GST fusion proteins are shown (Fig. 3 C). Together, these results suggest that the T1 motif plays a critical role in ERp57 enzymatic activity.

To determine the relationship between ERp57 catalytic activity with Ca²⁺ pump activity, we coexpressed wild-type or mutant ERp57 with SERCA 2b. As expected, ERp57 had the highest effect in reducing the frequency of Ca²⁺ oscillations (i.e., reduced pump activity). The ERp57-T2 mutant that had 70% enzymatic activity in the insulin assay exhibited intermediate levels of pump inhibition. Neither ERp57-T1 (30% enzyme activity) nor ERp57-T1T2 (devoid of enzyme activity) affected the Ca²⁺ wave period and t_1/2 that was enhanced by SERCA 2b overexpression (Fig. 4, A and B). Western blots demonstrated that ERp57 and its mutants as well as SERCA 2b expression levels were similar throughout expression groups (Fig. 4 C). Independent confirmation that all ERp57 mutants were correctly processed in the ER was provided by in vitro translations supplemented with canine microsomes (Fig. 4 D). Together, these results strongly suggest that ERp57 oxidoreductase activity is responsible for the modulation of Ca²⁺ uptake by SERCA 2b.

To investigate the interaction between ERp57 and its potential target L4, we performed an in vitro GST pull-down experiment under various Ca²⁺ concentrations: the highest being 300 μM to mimic full Ca²⁺ store content and the lowest being 10 μM to mimic Ca²⁺ store depletion. Interestingly, ERp57 binds to L4 in a Ca²⁺-dependent manner: the interaction is strongest at high Ca²⁺ (300 μM) and is significantly reduced at low Ca²⁺ (10 μM; Fig. 5 A). To test the specificity of this interaction we performed a similar GST pull-down experiment with GST-PDI. Neither GST-PDI nor GST alone interacted with the L4 (Fig. 5 B). This suggests that the interaction between ERp57 and L4 is specific. Moreover, this interaction occurs preferentially at high Ca²⁺ concentrations that are indicative of full Ca²⁺ stores (Pozzan et al., 1994).

Translational scanning studies of SERCAs indicate that they have intrinsic sequences that determine topologic insertion into the ER membrane (Bayle et al., 1995, 1997). To ascertain that the L4 construct was correctly inserted into the ER, a GFP-L4 fusion construct was coexpressed in Xenopus oocytes with a marker for ER localization, a DsRed-IP₃R fusion protein. As a control, cytosolic GFP was also coexpressed with DsRed-IP₃R. As expected, GFP-L4 colocalized with the IP₃R, whereas GFP had a more diffused cytosolic expression pattern (Fig. 5 C). These data indicate that the L4 construct was correctly localized in the ER.

Because the association of ERp57 with the L4 substrate is stronger at higher Ca²⁺ concentrations, we tested whether the intrinsic activity of ERp57 is Ca²⁺ dependent. We measured in vitro the catalytic activity of purified GST-ERp57 at the same range of [Ca²⁺] used in the GST pull-down assay (300, 150, 50, and 10 μM). The activity of ERp57 was only mildly but not significantly dependent on Ca²⁺ at the concentrations measured (Fig. 6 A). A similar assay was also performed for GST-PDI and in this case we found no Ca²⁺ dependence of its activity (Fig. 6 B).

CRT and CLNX inhibit Ca²⁺ oscillations via an interaction with the COOH-terminal sequence of SERCA 2b where a glycosylation consensus motif exists (John et al., 1998; Roderick et al., 2000). Furthermore, CRT does not interact with SERCA 2a, which lacks the 11th transmembrane segment and a COOH terminus in the ER lumen (John et al., 1998). ERp57 binds to CRT's P domain (Zapun et al., 1998; Oliver et al., 1999; Frickel et al., 2002; Leach et al., 2002). Based on these observations, we hypothesized that CRT binds the COOH-terminal sequence of the pump and recruits ERp57 to modulate L4 thiol groups and consequently pump activity. Because SERCA 2a also has the L4, we tested whether ERp57 modulates SERCA 2a directly or whether the effects are specific to SERCA 2b via CRT. We coexpressed ERp57 with SERCA 2a and found that ERp57 does not modulate Ca²⁺ oscillations in oocytes overexpressing this pump isoform (Fig. 7 A). We also coexpressed ERp57 with two mutants that eliminated the consensus glycosylation motif in SERCA 2b (SERCA 2b-N1036A and SERCA 2b-S1038A). We find that ERp57 does not modulate Ca²⁺ oscillations mediated by these two pump mutants (Fig. 7 B). Histograms of period and t_1/2 are shown in Fig. 7 C, whereas Western blots for these experiments are shown in Fig. 7 D. Together, these experiments demonstrate that CRT is required to recruit ERp57 to the L4.

The globular domain of CRT comprises the N and C domains and interacts with substrate glycoproteins, whereas the tip of the P domain binds to ERp57 and promotes disulfide bond formation (Ellgaard et al., 2001, 2002; Schrag et al., 2001; Frickel et al., 2002; Leach et al., 2002). Previous observations from our group demonstrated that CRT-NP (i.e., CRT lacking the C domain) modulated SERCA 2b just like wild-type CRT, indicating that in this case the C domain might not be necessary for the interaction (Camacho and Lechleiter, 1995; John et al., 1998). Thus, CRT most likely targets the COOH-terminal sequence of SERCA 2b via its N domain, whereas the P domain recruits ERp57 to interact with the L4. To test this hypothesis, we overexpressed four dominant negative constructs (L4, CRT-NC, CRT-N, or CRT-P) to determine whether they would interrupt the interaction between endogenous SERCA 2b with the ERp57–CRT complex. We found that oocytes overexpressing these four constructs exhibited shorter t_1/2 values (faster SERCA 2b activity) than control oocytes. In contrast, oocytes overexpressing CRT or CRT-NP exhibited longer t_1/2 values, consistent with previous findings (Camacho and Lechleiter, 1995; John et al., 1998; Fig. 8).

To determine the specificity of these dominant negative constructs for SERCA 2b, we also coexpressed them with either SERCA 2a or SERCA 2b. Oocytes coexpressing L4, CRT-NC, CRT-N, or CRT-P with SERCA 2b exhibited significantly shorter t_1/2 values than SERCA 2b overexpressing oocytes, consistent with our observations presented in the previous paragraph. Note that control oocytes coexpressing CRT with SERCA 2b exhibited the longest t_1/2 values (lowest frequency of Ca²⁺ oscillations; Camacho and Lechleiter, 1995; Fig. 9, A and B). In agreement with the results reported in Fig. 7, we found that Ca²⁺ oscillations enhanced by SERCA 2a were not affected by coexpression with these constructs (Fig. 9 C). Together, these data support our model that CRT recruits ERp57 to the lumenal L4 in SERCA 2b, modulating pump activity to maintain ER Ca²⁺ homeostasis (Fig. 10).

It is well established that CRT or CLNX form a macromolecular complex with ERp57 to assist in glycoprotein folding. In particular, CRT interacts with targeting glycoproteins via its globular domain, whereas the tip of the P domain recruits ERp57 to the glycoprotein substrate to promote intra/inter disulfide bond formation (High et al., 2000; Ellgaard et al., 2001, 2002; Schrag et al., 2001, 2003; Frickel et al., 2002; Leach et al., 2002; Ellgaard and Helenius, 2003; Kostova and Wolf, 2003). From our work using the Ca²⁺ oscillation assay in Xenopus oocytes, we have found that both the N and P domains of CRT are necessary for inhibition of Ca²⁺ oscillations enhanced by SERCA 2b, whereas the C domain is not required (Camacho and Lechleiter, 1995; John et al., 1998; Roderick et al., 2000). In this paper, we are led to the conclusion that CRT uses the N domain to target the COOH-terminal sequence of SERCA 2b, and the P domain to recruit ERp57 to interact with the L4 of SERCA 2b. Several of our findings support this conceptual model. ERp57 does not affect Ca²⁺ oscillations enhanced by SERCA 2a, nor SERCA 2b mutants defective in the COOH-terminal glycosylation motif. Overexpression of L4 or the NC, N, or P domains of CRT exhibit dominant negative effects that presumably act by disrupting the interaction between the endogenous CRT–ERp57 complex and SERCA 2b. In particular, overexpression of the first construct, L4, should bind endogenous ERp57 preventing it from regulating the pump. The second construct, CRT-NC, contains the entire globular domain that binds the glycosylated NFS motif of SERCA 2b, but lacks the ability to recruit ERp57 (Schrag et al., 2001; Leach et al., 2002). Its overexpression should compete with the interaction of endogenous CRT with SERCA 2b, thereby preventing ERp57 from regulating pump activity. The third construct, CRT-N, contains two residues that are responsible for binding glucose (Schrag et al., 2001). Its overexpression has the same effect as CRT-NC, indicating that this construct is sufficient to target the SERCA 2b COOH-terminal sequence. The fourth construct, CRT-P, contains the repeat known as 1112, which is longer than the fragment used by Ellgaard and colleagues (Ellgaard et al., 2002), who found that a shorter fragment (repeat 112) was sufficient to interact with ERp57. Without the globular NC domain, the P domain itself is unable to target the glycan (Leach et al., 2002). Consequently, overexpression of CRT-P should bind endogenous ERp57 and prevent regulation of SERCA 2b.

ERp57 and its homologue PDI have different substrate specificities (Freedman et al., 2002; Clissold and Bicknell, 2003). ERp57 specifically associates with N-linked glycoproteins, whereas PDI does not. This specificity is mainly mediated by an interaction with CRT or CLNX (Oliver et al., 1997, 1999; Van der Wal et al., 1998; Zapun et al., 1998; High et al., 2000; Frickel et al., 2002). Using in vitro GST pull-down experiments, we also demonstrated that ERp57 specifically interacts with the L4 of SERCA 2b, whereas PDI did not. It is possible that there is a direct interaction between ERp57 and L4, but it is equally likely that an indirect protein interaction is occurring in a pull-down assay. For example, microsomes contain CRT and glycoproteins (e.g., SERCA 2b), which could then link ERp57 to L4. In fact, the bulk of our results favor the requirement of CRT to recruit ERp57 to the L4.

The mechanisms that regulate SERCA activity from the ER have been until now somewhat unclear. Decreased lumenal Ca²⁺ induces a strong stimulation of SERCA activity in isolated pancreatic acinar cells (Mogami et al., 1998). We demonstrate that the association of ERp57 with the L4 is Ca²⁺ dependent in the physiological range of ER lumenal Ca²⁺ (10, 50, 150, and 300 μM). The Ca²⁺ dependency of the association between ERp57 and L4 is likely to be mediated by CRT, because Ca²⁺ does not regulate the intrinsic enzyme activity of the ERp57. It has been reported that Ca²⁺ is required for CRT binding to oligosaccharides, although these experiments were conducted beyond physiological [Ca²⁺] (10 mM Ca²⁺; Vassilakos et al., 1998). Ca²⁺ has also been shown to modulate the interaction between CRT and ERp57 (Corbett et al., 1999). Together, our data suggest the following role of ER Ca²⁺ in the feedback mechanism that regulates SERCA 2b activity (Fig. 10). When ER Ca²⁺ stores are full (∼300 μM Ca²⁺), ERp57 binds to L4 promoting disulfide bond formation in the loop, which inhibits pump activity. When Ca²⁺ stores become depleted (∼10 μM Ca²⁺), ERp57 dissociates from L4, resulting in the reduced form of SERCA 2b that is more active. The mechanism by which thiol groups in the L4 are reduced after ERp57 dissociates from the pump is unknown. The ER lumen maintains a neutral to slightly oxidized environment (GSH–GSSG ∼1–3; Hwang et al., 1992). Our results suggest that ERp57 must be bound to L4 in order to maintain it in an oxidized state. It is possible that additional ER proteins play a role here. Regardless of the precise underlying mechanism, our data clearly indicate that the reduced form of L4 supports higher pump activity to rapidly refill the ER.

The CLNX–CRT cycle has a well-established role in productive glycoprotein folding and quality control of nascent glycoproteins undergoing folding (High et al., 2000; Ellgaard and Helenius, 2003; Kostova and Wolf, 2003; Schrag et al., 2003). Our observations indicate that these ER chaperones interact with a mature ER resident glycoprotein (SERCA 2b). Association/dissociation of other ER chaperones with mature proteins has also been reported during the unfolded protein response (Bertolotti et al., 2000). We speculate that the classical CLNX–CRT cycle and the system we have characterized here exist in parallel and that the action of CRT–CLNX on SERCA 2b provides the homeostatic feedback necessary to maintain Ca²⁺ conditions that promote productive glycoprotein folding. Without it, Ca²⁺ depletion could result in the well-known unfolded protein response, which can bring about cellular apoptosis under certain stress conditions (Kaufman, 1999; Welihinda et al., 1999; Patil and Walter, 2001; Harding et al., 2002). Maintaining optimal Ca²⁺ concentrations in the ER is clearly important for nascent protein folding (Hardman and Wetmore, 1996; Chen et al., 1997; Stevens and Argon, 1999; Ramsden et al., 2000). The ability of the lectin chaperones to “talk” to a mature protein in the Ca²⁺ signaling machinery provides the mechanism by which the optimal Ca²⁺ environment is maintained for protein folding.

Oxidant treatment inactivates SERCA 2b to a greater extent than other pump isoforms (Grover et al., 1997; Barnes et al., 2000). Our observations imply that SERCA 2b is in fact a redox sensor. In the oxidized state of L4, this pump exhibits lower activity. On the other hand, the pump favors rapid Ca²⁺ uptake in the L4 mutants that are constitutively reduced by cysteine mutagenesis. Interestingly, the ryanodine receptor, a Ca²⁺ release channel located on the SR–ER is also a redox sensor (Zable et al., 1997; Xia et al., 2000; Sun et al., 2001). The oxidized channel correlates with the open state, whereas the reduced form correlates with channel closure. The IP₃R may also sense the redox potential because it contains cysteines in the ER lumenal facing loop that line the channel pore. Consequently, modulation of redox potential of reactive thiols in the ER might be a general mechanism by which SERCA 2b, IP₃R, and ryanodine receptor control ER Ca²⁺. In terms of Ca²⁺ homeostasis, our observations with ERp57 fit the view that ER redox modulates Ca²⁺ release and uptake in coordinate and opposite directions. In particular, oxidization appears to favor Ca²⁺ release by opening Ca²⁺ channels and inhibiting Ca²⁺ pumping. On the other hand, reduction seems to favor Ca²⁺ uptake by channel closure and increasing Ca²⁺ pumping. The combined effect of these two actions helps to minimize the loss of ER Ca²⁺ and maximize refilling of the stores, thereby protecting the physiological functions of the ER such as glycoprotein folding.

The Xenopus expression vectors for rat SERCA 2a and SERCA 2b have been described previously (John et al., 1998). L4 cysteine–deficient mutants of SERCA 2b were generated as follows: cysteines 875 and 887 were mutated to serines (mutant SERCA 2b-C1SC2S) or to alanines (mutant SERCA 2b-C1AC2A). A two-step mutagenesis protocol was performed. To generate the SERCA 2b-C1SC2S mutation primers 5′-GCTGAGTCATTTCCTGCAGTCAAAGGAGGACAACCC-3′ and 5′-CCAGACTTCGAAGGAGTGGATTCAGCAATCTTTGAGTCCC-3′ with their respective reverse complement primers were used. Primers 5′-GCTGAGTCATTTCCTGCAGGCTAAGGAGGACAACCC-3′ and 5′-CCAGACTTCGAAGGAGTGGATGCTGCAATCTTTGAGTCCC-3′ and their respective reverse complement primers were used for the SERCA 2b-C1C2A mutant. Plasmid pHN-SERCA 2b-C875G was generated using the forward primer 5′-GCTGAGTCATTTCCTGCAGGGTAAGGAGGACAACCC-3′ and its reverse complement. Plasmid pHN-SERCA 2b-N1036A has been generated and published previously (John et al., 1998). Mutant SERCA 2b-S1038A (plasmid pHN SERCA 2b-S1038A) was generated by PCR using the forward primer 5′-AGCACAGACACTAACTTTGCTGATATGTTCTGG-3′ and the reverse primer 5′-CCAGAACATATCAGCAAAGTTAGTGTCTGTGCT-3′.

A construct encoding the L4 of SERCA 2b was generated by PCR amplification spanning methionine 814 and arginine 922 with primers 5′-ACTGGGATCCATGAACAAACCCCCACGGAACCCA-3′ and 5′-ACTGAAGCTTTTACCTCAGCAGGGACTGGTTTTC-3′. The PCR product was ligated into the BamHI and HindIII sites of the similarly digested vector pHN (plasmid pHN-SERCA 2b-L4). To generate plasmid pHN-GFP-L4, a BamHI fragment from pHN-GFP-S65T-ΔTAA, as described previously (John et al., 1998), was subcloned into plasmid pHN-SERCA 2b-L4 that was digested with BamHI and treated with calf intestinal phosphatase. Plasmid pHN-GFT S65T was generated previously (John et al., 1998). The Xenopus expression vector of DsRed-IP₃R was generated by a three-step PCR protocol to create a fusion of DsRed to the NH₂-terminal sequence of the IP₃R. First, the full-length DsRed was amplified by PCR from the template of pDSRed 1-N1 (CLONTECH Laboratories, Inc.). The forward and reverse primers were 5′-ACTGGAATTCATGGTGCGCTCCTCCAAGAACGTC-3′ and 5′-AAGAAAGCTGGACATTTCATTCATCAGGAACAGGTGGTGGCGGCC-3′. Second, the NH₂ terminus of IP₃R was amplified by PCR from template pHN-IP₃R with the forward and reverse primers as 5′-GGCCGCCACCACCTGTTCCTGATGAATGAAATGTCCAGCTTTCTT-3′ and 5′-GTGCACATGTAGCATCAGGTGGCAGAATGA-3′. The purified PCR products served as templates to create the final product DsRe-IP₃R. In this PCR reaction, the forward and reverse primers were 5′-ACTGGAATTCATGGTGCGCTCCTCCAAGAACGTC-3′ and 5′-GTGCACATGTAGCATCAGGTGGCAGAATGA-3′. The final PCR fusion product was digested with EcoRI and NsiI and subcloned into the EcoRI and NsiI sites of PHN-IP₃R.

The Xenopus expression vector of ERp57 (plasmid pHN-ERp57) was generated in a three-step PCR protocol. First, the mature human ERp57 was amplified from the template pET9-ERp613 (a gift from D.Y. Thomas, Biotechnology Research Institute, Montréal, Canada) with primers 5′-CTGCTCGGCCTGGCCGCCGCCTCCGACGTGCTAGAACTCACG-3′ and 5′-ACGTAAGCTTTTAGAGATCCTCCTGTGCCTTCTT-3′. A second PCR generated the signal peptide of CRT from the template pHN-CRT (Camacho and Lechleiter, 1995) using primers 5′-TAATACGACTCACTATAGGG-3′ and 5′-CGTGAGTTCTAGCACGTCGGAGGCGGCGGCCAGGCCGAGCAG-3′. The purified PCR products served as templates to generate a fusion of the CRT signal peptide and mature ERp57. In this PCR reaction, the forward and reverse primers were 5′-TAATACGACTCACTATAGGG-3′ and 5′-ACGTAAGCTTTTAGAGATCCTCCTGTGCCTTCTT-3′. The final PCR fusion product was subcloned into the HindIII and SSTI sites of the Xenopus expression vector pHN. Three mutants of ERp57 were generated. The first mutant had cysteines 57 and 60 mutated to serines in the NH₂-terminal thioredoxin motif (mutant ERp57-T1) and was generated by PCR using primer 5′-CGAGTTCTTCGCTCCCTGGTCTGGACACTCCAAGAGACTTGCACC-3′ and its reverse complement. The second construct had cysteines 406 and 409 in the COOH-terminal thioredoxin motif mutated to serines (mutant ERp57-T2) and was generated using primer 5′-GAATTTTATGCCCCTTGGTCTGGTCATTCTAAGAACCTGGAGCCC-3′ and its reverse complement. The third construct was a double mutant generated in two PCR steps with the above primers (mutant ERp57-T1T2).

Rat PDI cDNA was cloned from a rat liver cDNA library (Invitrogen) by PCR using primers 5′-ACTGGGATCCATGCTGAGCCGTGCTTTGCTGTGC-3′ and 5′-ACTGTCTAGACTACAGTTCATCCTTCACGGC-3′ based on the GenBank/EMBL/DDBJ accession no. X02918. The PCR product was subcloned into the BamHI and XbaI sites of pHN and fully sequenced.

To obtain GST-ERp57 fusion proteins including wild type and mutants as well as GST-PDI, we amplified the corresponding mature cDNA fragments by PCR and subcloned them into pGEX-4T-2 vector (Amersham Biosciences). For ERp57 and its mutants, the primers were 5′-TCGAGGATCCATGTCCGACGTGCTAGAACTCACG-3′ and 5′-ACTGCTCGAGTTAGAGATCCTCCTGTGCCTTCTT-3′, whereas for PDI, the primers were 5′-ACTGGGATCCATGGACGCTCTGGAGGAGGAGGAC-3′ and 5′-ACTGCTCGAGCTACAGTTAATCCTTCACGGC-3′. The PCR products were ligated into the BamHI and XhoI sites of pGEX-4T-2.

The Xenopus vectors encoding CRT, CRT-NP, CRT-NC, and CRT-N domains have been published previously (Camacho and Lechleiter, 1995). The construct for expression of the P domain (plasmid pHN-CRT-P) was generated by PCR mutagenesis designed to remove the N domain from template pHN-CRT-NP (Camacho and Lechleiter, 1995). The primers used had sequence 5′-GCCGCCGAGCCCGATGACTGGGACTTCCTACCCCCCAAGAAGATAAAGGACCCA-3′ and 5′-TGGGTCCTTTATCTTCTTGGGGGGTAGGAAGTCCCAGTCATCGGGCTCGGCGGC-3′. The resulting construct contains the CRT signal peptide followed by the P domain (amino acids 181–260) and a KDEL sequence for ER retention.

Automatic sequencing of all cDNA constructs was performed at the UTHSCSA Advance Nucleic Acid Core Facility. All primers were purchased from Operon Technologies (QIAGEN). Unless otherwise specified, all chemicals were purchased from Sigma-Aldrich and all restriction enzymes were purchased from Invitrogen Life Technologies.

Synthetic mRNA was prepared as described previously (Camacho and Lechleiter, 1995). In vitro translations were performed as described previously (Roderick et al., 2000).

Oocytes extracts were prepared as described previously (Camacho and Lechleiter, 1995). In brief, 5–10 oocytes were pooled and homogenized in buffer containing 20 mM Tris-HCl, pH 7.6, 140 mM NaCl, 250 mM sucrose, and protease inhibitors in a mix containing 0.2 mM AEBSF, 10 μM leupeptin, 1 μM pepstatin A, and 0.8 mM benzamidine (Calbiochem). After centrifugation at 4,500 g for 15 min at 4°C, the supernatant was ultracentrifuged at 125,000 g for 20 min at 4°C. Microsomal pellets were resuspended in solubilization buffer (10 mM Tris-HCl, pH 7.6, 140 mM NaCl, 1% Triton X-100, and protease inhibitors). A second centrifugation (15,000 g for 10 min) was performed to discard the insoluble material. SDS-PAGE was performed by loading 1–2 oocyte equivalents per lane and transferred to nitrocellulose. A polyclonal rabbit anti-SERCA antibody ( NI, 1:8,000 dilution, a gift from J. Lytton, University of Calgary, Calgary, Canada) was used to detect SERCA 2b. A rabbit anti-ERp57 polyclonal antibody (1:4,000 dilution, a gift from S. High, University of Manchester, Manchester, UK) was used to detect ERp57. HRP-conjugated donkey anti–rabbit IgG secondary antibodies (Jackson ImmunoResearch Laboratories) in a dilution of 1:10,000 were used in all Western blots and visualized by ECL (PerkinElmer Life Sciences, Inc.).

BL21 bacteria transformed with pGEX-4T-2, pGEX-4T-2-PDI or pGEX-4T-2-ERp57 (and mutants) were grown to OD 600 nm 0.6. Isopropyl-β-D-thiogalactopyranoside (Research Products International Corp.) was added to a final concentration of 1 mM to induce protein expression for 4 h at 37°C. Bacteria were lysed by sonication in 1× PBS containing 100 mM EDTA and a mix of protease inhibitors including (200 μM AEBSF, 10 μM leupeptin, and 1.5 μM pepstatin A) and 0.8 mM benzamidine. Bacterial lysates were centrifuged at 22,000 g and the supernatant collected. Binding of GST or GST fusion proteins to glutathione-Sepharose 4B (Amersham Biosciences) was performed at 4°C for 1 h followed by three washes with equilibration buffer (0.5 M Tris-HCl, pH 8.0, 4 mM EDTA, 0.1% β-ME, 5% glycerol). Elution of bound protein was performed in this buffer supplemented with 15 mM glutathione. Proteins were dialysed against 0.05 M Tris-HCl, pH 7.5, containing 1 mM EDTA, 0.2% β-ME, and stored at –80°C.

In vitro translation of L4 was accomplished in a rabbit reticulocyte lysates supplement with canine pancreatic microsomes (both from Promega) and L-[³⁵S]methionine (PerkinElmer Life Sciences). Microsomes were isolated and resuspended in solubilization buffer (10 mM Tris-HCl, pH 7.6, 140 mM NaCl, 1% Triton X-100 and a mixture of the protease inhibitors described in Western blots). Binding of in vitro–translated L4 to GST-ERp57 fusion protein(s) was performed at 4°C overnight in the presence of glutathione Sepharose 4B (Amersham Biosciences) in binding solution containing in mM 10 Tris-HCl, pH 8.0, 70 KCl, 2 MgCl₂, 50 μM EGTA, 5% BSA, and protease inhibitors supplemented with experimental CaCl₂ concentrations (60, 100, 200, and 350 μM). Proteins bound to glutathione Sepharose beads were washed three times in buffer containing 0.2 M Tris-HCl, pH 8.0, 0.1% Triton X-100, 70 mM KCl, 2 mM MgCl₂, 50 μM EGTA, protease inhibitors, and corresponding CaCl₂ concentrations. Proteins were resolved by 15% SDS-PAGE and visualized by autoradiography. Ca²⁺ concentrations were calculated according to existing algorithms (Fabiato and Fabiato, 1979).

The insulin turbidity assay was performed as described previously (Holmgren, 1979; Hirano et al., 1995). In this assay, oxidized insulin was used as substrate to measure thiol-dependent reductase activity of ERp57. Under reducing conditions, the two interchain disulphide bonds of insulin are cleaved, resulting in the formation of a white insoluble precipitate. 10 mg/ml of insulin (Sigma-Aldrich) stock solution was prepared as described previously (Holmgren, 1979). In brief, 100 mg of insulin was first resuspended in 8 ml 0.05 M Tris-HCl, pH 8.0. Subsequently, the pH was adjusted between 2.0 and 3.0 by adding 1 M HCl and rapidly titrated to pH 8.0 with 1 M NaOH. The final volume of the solution was adjusted to 10 ml with H₂O. The clear stock solution was stored at –20°C. On the experimental day, the insulin stock solution was diluted to 1 mg/ml with a buffer containing in 100 mM KAC, pH 7.5, and 2 mM EDTA, supplemented with CaCl₂ concentrations (2.44, 2.28, 2.17, and 2.12 mM) to yield free [Ca²⁺] of 300, 150, 50, and 10 μM. The final reaction volume was adjusted to 1 ml after purified ERp57 or its mutants were added. The reaction was initiated by adding 3 μl of 100 mM DTT. Abs_{650 nm} was measured every 5 min in a BioSpec-1601 spectrophotometer (Shimadzu). Enzyme activity was defined by measuring the slope of the linear portion of the absorbance curve. Ca²⁺ concentrations were calculated according to known algorithms (Fabiato and Fabiato, 1979).

Stage VI–defolliculated oocytes were injected a bolus of 50 nl of 1 μg mRNA using a standard positive pressure injector (Drummond Scientific). Oocytes were cultured in 50% L-15 media (Invitrogen) for 5–7 d at 18°C. Intracellular Ca²⁺ was imaged as described previously (Roderick et al., 2000). Oocytes were injected with fluorescent Ca²⁺ indicator Oregon green II (12.5 μM final; Molecular Probes) 30–60 min before imaging. Ca²⁺ release was initiated by injecting a 50 nl bolus of 6 μM IP₃ (Calbiochem) to yield ∼300 nM final concentration. Imaging was performed in ND96 buffer containing 1 mM EGTA. Images were acquired at a rate of 0.75 s /frame on a confocal laser scanning microscope (model PCM2000; Nikon) using a 10× objective (NA = 0.45) at zoom 1.

To characterize the ER localization of L4, GFP-L4 was coexpressed in oocytes with DsRed-IP₃R. Fluorescence was monitored using a confocal microscope (model Fluoview 500; Olympus). GFP-L4 as well as GFP (as a negative control) were detected using a 488-nm laser line for excitation and a combination of a 510-nm long pass and a 550-nm short pass barrier filter for signal emission. DsRed fluorescence was obtained with a 568-nm laser line for excitation and a 585-nm long pass barrier filter for emission. Emission signals were collected using a 60× oil (NA = 1.4) objective (Olympus) at zoom 5.

The analysis of Ca²⁺ images was performed using the public domain NIH ImageJ program (developed at the National Institutes of Health and available at http://rsb.info.nih.gov/ij/).

Statistical significance was determined by t test or one-way ANOVA as appropriate and accepted at P < 0.05.

We wish to thank David Y. Thomas for the cDNA encoding mature human ERp57, Stephen High for rabbit αERP57 antibody, and Johnathan Lytton for rabbit SERCA 2 antibody. We also thank James Lechleiter, Da-Ting Lin, and Marianna Bollo for critical reading of the manuscript.

This work was funded by National Institutes of Health grant R01 GM55372 to P. Camacho.