The Role of Calcium on the Activity of ERcalcistorin/Protein-disulfide Isomerase and the Significance of the C-terminal and Its Calcium Binding

ERcalcistorin/protein-disulfide isomerase (ECaSt/PDI) shows a 55% identity with mammalian protein-disulfide isomerase (PDI) (Lucero, H. A., Lebeche, D., and Kaminer, B. (1994) J. Biol. Chem. 269, 23112–23119) is a high capacity low affinity Ca2+-binding protein and behaves as a Ca2+ storage protein in the ER of a living cell (Lucero, H. A., Lebeche, D., and Kaminer, B. (1998) J. Biol. Chem. 273, 9857–9863). Here we show that recombinant ECaSt/PDI bound 26 mol of Ca2+/mol and a C-terminal truncated mutant bound 14 mol of Ca2+/mol, both with aK d of 2.8 mm in 50 mm KCl and 5.2 mm in 150 mm KCl. The percentage reduction in Ca2+ binding in the mutant corresponded with the percentage reduction of deleted pairs of acidic residues, postulated low affinity Ca2+-binding sites. 5 mm Ca2+ moderately increased the PDI activity of both ECaSt/PDI and the C-terminal truncated mutant on reduced RNase and insulin. Surprisingly, ECaSt/PDI in the absence of Ca2+prevented the spontaneous reactivation of reduced bovine pancreatic trypsin inhibitor. In the presence of 1–5 mmCa2+ (or 10 μm polylysine) ECaSt/PDI augmented the bovine pancreatic trypsin inhibitor reactivation rate. In contrast, the C-terminal truncated ECaSt/PDI augmented rBPTI reactivation in the absence of Ca2+ and 1–5 mmCa2+ further accelerated the reactivation rate, responses similar to those obtained with mammalian PDI.

The endoplasmic reticulum (ER) 1 is a multifunctional organelle involved in post-translational modification of nascent proteins, in lipid synthesis and regulation of intracellular Ca 2ϩ . Protein folding in the ER associated with disulfide bond formation is likely to be catalyzed by protein-disulfide isomerase (PDI), the identification of which stemmed from the early findings of a factor which catalyzed the renaturation of ribonuclease in vitro (1,2).
On the basis of sequence homology in the primary structure of PDI, Edman et al. (17) identified multiple domains, a, b, b and a. The a domain (at the N-terminal) and the a (a portion of the C-terminal) are homologous with thioredoxin in their sequence and folded structure and each contains the CGHC sequence, the active sites. b and b are two homologous central domains with folded structures similar to thioredoxin but with no thioredoxin sequence homology (18). Interestingly, domain b shares a limited sequence identity with calsequestrin (19). These four domains therefore resemble thioredoxin folded structures (18). A domain, between the a domain and b domain was designated e because of its homology to the ligand-binding site of the estrogen receptor (11). Whether this should be regarded as a separate domain is now being questioned (19).
A sixth domain in the C-terminal was recently labeled c as a putative Ca 2ϩ -binding domain (4). This speculation is in keeping with our findings and hypothesis on calcium binding. We found quantitatively, for the first time, that sea urchin (20) and mammalian PDI (21) are high capacity low affinity Ca 2ϩ -binding proteins. And it should be recognized that calcium binding by mammalian PDI was not quantitated previously (22). Having determined the number of calcium ions bound, we related them to the number of paired acidic residues and hypothesized that paired acidic residues constituted low affinity Ca 2ϩ -binding sites, a cluster of which resides in the tail end of the C-terminal of mammalian and sea urchin PDI (23). The Cterminal of PDI ends with a KDEL sequence characteristic of ER-resident proteins (24). However, PDI has also been detected on the cell surface by immunocytology (25) and other techniques (26), in different cell types (27,28) and platelets (29).
Mammalian PDI has been regarded for many years to behave as a homodimer on gel filtration (30,31) and to form a tetramer under certain conditions (32). However, recent reports suggest that it exists as a monomer under standard conditions (19,33). Several properties of mammalian PDI domains have been studied. Recent investigations on recombinant fragments have elucidated domains involved in folding (18,19,34), catalysis (19, 34 -36), chaperone activity (35,37), and peptide/protein recognition (38).
A purified microsomal protein from sea urchin eggs binds 23 mol of Ca 2ϩ at low affinity (20), shares certain characteristics with calsequestrin (39), is localized within the ER (40), has 55% identity with mammalian PDI and has PDI activity (23). Hence, we designated it ERcalcistorin/protein-disulfide isomerase (ECaSt/PDI) alluding to its putative dual functions within the ER (23). We subsequently showed that mammalian PDI binds 19 mol of Ca 2ϩ /mol of protein at low affinity (half-satu-ration values, derived from the Hill equation were 2.77, 4.73, and 5.20 mM in the presence of 20, 100, or 100 mM KCl plus 3 mM MgCl 2 , respectively (21)). Thus, PDI presumably acts as a low affinity Ca 2ϩ storage protein within the ER of mammals and sea urchins. We have recently obtained direct supporting evidence for such a role of ECaSt/PDI in the ER of a living cell (41). Hence PDI besides calreticulin (42) may be another major Ca 2ϩ storage protein in the ER.
Suggestions that other ER proteins may serve a Ca 2ϩ storage role have been based on their quantitation as high capacity low affinity calcium-binding proteins, e.g. CaBP2, previously designated ERp72 and characterized as containing three thioredoxin-like active site domains (43). CaBP2 binds 12 mol of Ca 2ϩ /mol of protein, CaBP4 binds 11 mol of Ca 2ϩ /mol of protein (44), and endoplasmin 8 -10 mol of Ca 2ϩ /mol of protein (45). CaBP4 (44) shows identity with GRP94 (46) and endoplasmin (45). Increased Ca 2ϩ storage is also based on increased Ca 2ϩ release by cells transfected with a putative Ca 2ϩ storage protein, e.g. BiP (47).
The approximate correspondence of the number of calcium ions bound by ECaSt/PDI (20), mammalian PDI (21), and calreticulin (48) with the number of paired acidic residues in the respective molecules, led us to postulate, as already mentioned, that pairs of acidic residues were low affinity Ca 2ϩ -binding sites, but not exclusively, since Ca 2ϩ could bind to single carboxyl group bridged to other groups (23). In support of our hypothesis we found, in this investigation, a correspondence between the percentage loss of paired acidic residues in the C-terminal truncated mutant and the percentage reduction in its Ca 2ϩ binding capacity.
One might reasonably consider whether the abundant low affinity Ca 2ϩ binding might also influence the enzyme activity of PDI. Hence, we studied the effect of Ca 2ϩ on PDI activity using concentrations within the range and somewhat above the K d values for Ca 2ϩ binding which vary depending on ionic conditions. The choice of such concentrations was based on the assumptions that the free Ca 2ϩ concentration within the intimate microenvironment of ECaSt/PDI in the ER would be in equilibrium with the K d values and that the published ER concentrations of free Ca 2ϩ in the micromolar range (for review, see Ref. 49), would not necessarily reflect the concentrations in the microenvironment of ECaSt/PDI or might not be accurate. Free Ca 2ϩ concentrations in the ER up to 400 M have been recently detected in cells transfected with a unique construct of green fluorescent protein fused with calmodulin (50). A considerably higher concentration of about 2 mM was considered to exist using Sr 2ϩ as a surrogate for Ca 2ϩ in cells transfected with an aequorin-BiP chimera (51). In any event, there are difficulties in obtaining accurate measurements of free Ca 2ϩ in the ER (49,52). The latter finding of an average Ca 2ϩ concentration of about 2 mM suggests the possibility that the concentration may well be in the range of the K d values for Ca 2ϩ binding by ER Ca 2ϩ storage proteins.
We now report that whereas 1 mM Ca 2ϩ had a negligible effect on the enzyme activity of either ECaSt/PDI or the Cterminal truncated mutant using rRNase or insulin as substrates, 5 mM Ca 2ϩ had a moderate stimulating effect. However, with respect to the reactivation of reduced BPTI, 1-5 mM Ca 2ϩ augmented the enzyme activity of ECaSt/PDI and the C-terminal truncated mutant. 5 mM Ca 2ϩ also augmented this reactivation catalyzed by mammalian PDI. Surprisingly, ECaSt/PDI, in contrast to the C-terminal truncated mutant and mammalian PDI, prevented the spontaneous reactivation of rBPTI in the absence of Ca 2ϩ suggesting that the C-terminal of ECaSt/PDI, differs from that of mammalian PDI, and has a charge distribution which enable it to complex with rBPTI and lock it in the reduced state in the absence of Ca 2ϩ .
Construction of cDNAs-Clone 8a1a1 cDNA in pBluescript encoding ECaSt/PDI (23) was amplified by the polymerase chain reaction using a sense primer in combination with reverse primers to produce the complete molecule and the C-terminal truncated (⌬ 385-475 ) mutant. The forward primer M106 (5Ј-ATATTCTAGACATATGGCAGTCGAGGTC-GAAATCGAAGAAGATGTC-3Ј) was designed to hybridize with the first 33 nucleotides encoding the mature protein at the 5Ј terminus of the ECaSt/PDI cDNA and contains a NdeI site (in bold) at the position of the initiation codon. The reverse primer M107 (5Ј-GATTGAGCGGC-CGCTTAAAGTTCATCCTTGGCT(TGATCCTC-3Ј) was used for generation of the cDNA encoding the complete ECaSt/PDI. For the generation of a cDNA encoding the mutant the reverse primer M108 (5Ј-GT-ATGAGCGGCCGCTTAAAGTTCATCCTTATAAATGGGAGCAAGCT-GTTT-3Ј) included the NotI site (in bold), a stop codon (italics), and the sequence encoding the KDEL retention signal (underlined).
Polymerase chain reaction amplification was carried out using Am-pliT transformed aq kit (Perkin-Elmer protocol). Polymerase chain reaction products were ligated into pCRII(A/T) cloning vector and Escherichia coli HB101 was transformed with the ligation products as described (Invitrogen protocol). The fidelity of the sequence of the polymerase chain reaction products was confirmed by DNA sequencing (53). The complete cDNA was ligated into the NdeI/XhoI sites of the E. coli expression vector pAED4 and the truncated gene into the NdeI/ KpnI sites of the vector (54). Ligation products were used to transform BL21DE3 pLys(S) E. coli cells, a protease-deficient strain expressing T7 DNA polymerase under control of the isopropyl-1-thio-␤-D-galactopyranoside inducible promoter. Cultures were grown at 37°C in LB broth (250 ml) supplemented with 50 g/ml ampicillin until absorbance at 600 nm reached 0.4. Isopropyl-1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM and cells were grown for an additional 3 h.
Purification of Recombinant ECaSt/PDI and the Mutant-The purification was according to the protocol described for the purification of recombinant human PDI (55) with some modifications. E. coli cells, expressing ECaSt/PDI and the mutant in 250 ml of culture, were collected by centrifugation at 5,000 ϫ g for 15 min, resuspended in 50 ml of 100 mM Tris-HCl (pH 7.4), 1 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, and 30 M leupeptine. Cells were disrupted by sonication and the lysate was spun down at 20,000 ϫ g for 30 min. The supernatant was dialyzed 3 times against 4 liters of 25 mM sodium phosphate (pH 6.3). Partial purification was attained by ion exchange chromatography. The dialysate was applied to a 2.5 ϫ 35-cm column of DEAE-Sephacel equilibrated with the same buffer. The column was washed with 300 ml of 25 mM sodium phosphate (pH 6.3) and proteins were eluted by a linear gradient of 0 to 0.7 M NaCl (200 ml each) in 25 mM phosphate buffer (pH 6.3). Fractions (4 ml) containing PDI activity that eluted at approximately 0.37-0.43 mM NaCl were analyzed by SDS-PAGE, combined, concentrated to 2-2.5 mg/ml by ultrafiltration, dialyzed against 100 volumes of 25 mM Tris-HCl (pH 7.4), 1 mM EDTA, 1 mM EGTA and stored frozen at Ϫ20°C. Further purification was achieved by gel filtration in a Superose 12 column equilibrated and eluted in the same buffer. Samples (100 l) were loaded onto the column and 0.5-ml fractions were collected at a flow rate of 0.5 ml/min. Protein elution was monitored by the absorbance at 280 nm and the eluate with the major absorbance peak and PDI activity was concentrated to 1.5-2 mg/ml by ultrafiltration and stored at Ϫ20°C.
Purification of ECaSt/PDI from Sea Urchin Eggs and Rabbit PDI from Liver-ECaSt/PDI from sea urchin eggs (39) and mammalian PDI from rabbit liver (56) were purified as described.
Calcium Binding-The E. coli expressed ECaSt/PDI and the mutant (0.5-0.6 mg/ml) were predialyzed and calcium binding was measured by equilibrium dialysis essentially as described (20). Briefly, protein samples (150 l) were dialyzed in a Spectropor semimicrodialysis tubing (4 mm, 12-14 kDa cut-off) against a buffer solution (20 ml) containing 20 mM MOPS (pH 7.0), 3 mM MgCl 2 , and 50 or 150 mM KCl, and various concentrations of CaCl 2 with specific radioactivity of 250 -300 cpm/nmol, at 4°C for 24 h in capped polyethylene scintillation vials. At equilibrium, quadruplicate samples (30 l) were taken from inside and outside the dialysis tubing to determine radioactivity and duplicate samples (15 l) from inside the dialysis tubing were taken to measure protein concentration (57). Calcium binding for each CaCl 2 concentration was determined in three experiments.
Reduction and Denaturation of BPTI-BPTI was reduced and denatured essentially as described (58). BPTI (4 mg) was suspended in a medium (200 l) containing 8 M guanidinium-HCl, 150 mM dithiothreitol, 200 mM Tris-HCl (pH 8.5), and stirred for 24 h at room temperature. The reduced, denatured protein was isolated by gel filtration on a Sephadex G-25 column equilibrated and eluted in 10 mM HCl. The presence of reduced BPTI in the column fractions (0.5 ml) was detected using 5,5Ј-dithiobis(2-nitrobenzoic acid) as described (59). Fractions containing rBPTI (1.5-2.0 mg/ml) were pooled and kept frozen at Ϫ20°C until used. Under these conditions the reduced, denatured inhibitor was stable for at least 3 months. BPTI protein concentration was estimated spectrophotometrically at 280 nm (60).
Reactivation of rBPTI-The reactivation of rBPTI was assayed by monitoring its stoichiometric inhibitory effect on trypsin activity (58). PDI catalysis on refolding of BPTI occurs even in the absence of thiol redox buffer (61). Since we found the rate of spontaneous reactivation in the presence of redox buffer to be too rapid to study the effects of other agents such as Ca 2ϩ , we eliminated the redox buffer and observed the reactivation over a prolonged period of time. In one assay rBPTI was incubated in a reactivation medium (125 l) containing 100 mM Tris-HCl (pH 7.4), 0.5 mM EDTA, 0.5 mM EGTA. In another assay the reactivation was measured in the same medium as for the Ca 2ϩ binding measurements (20 mM MOPS, pH 7.0, 150 mM KCl, 3 mM MgCl 2 ). Additional components included ECaSt/PDI (10 M), the C-terminal truncated mutant (10 M), mammalian PDI (10 M), polyglutamate, polylysine, Mg 2ϩ , and Ca 2ϩ as indicated in the legend of Fig. 4. The mixtures were incubated for 5 min at 22°C before adding rBPTI (20 g) to initiate the reactivation. Aliquots (10 l) of each medium were transferred at time intervals up to about 230 min in a cuvette containing trypsin assay medium (see below).
Trypsin Activity Assay-Trypsin activity was assayed using the substrate N-␣-benzoyl-DL-arginine-p-nitroanilide as described (60). Briefly, trypsin (240 pmol of active trypsin) was added to a medium (500 l) containing 100 mM Tris-HCl (pH 7.4), 2 mM CaCl 2 , 0.15 g of N-␣benzoyl-DL-arginine-p-nitroanilide. The increase in absorbance at 405 nm was recorded for at least 1 min before and after the addition of the 10-l aliquot from the reactivation medium containing BPTI. The percent of remaining trypsin activity after addition of BPTI was converted to the percent of reactivated BPTI using the following equation: % reactivated BPTI ϭ [T Ϫ (% trypsin activity ϫ T/100)] ϫ 100/P, where T is the total number of picomoles of active trypsin in the trypsin assay medium and P is the total number of picomoles of BPTI transferred to the trypsin assay medium. The molarity of active trypsin was determined using the active site titrant p-nitrophenyl-pЈ-guanidinobenzoate (62).
Other PDI Activity Assays-PDI activity was assayed by the reactivation of reduced RNase or the net reduction of insulin as described (23).
Isolation of ECaSt/PDI⅐rBPTI Complex by Gel Filtration-125 I-rBPTI (20 g containing 400 cpm/g) was mixed with complete ECaSt/ PDI (140 g) in 125 l of a buffer containing 100 mM Tris-HCl (pH 7.4), 0.5 mM EDTA, 0.5 mM EGTA, with or without 5 mM CaCl 2 at 22°C for 30 min. Samples of ECaSt/PDI, 125 I-rBPTI, and a mixture of them in the presence or absence of 5 mM CaCl 2 were loaded on a Superose 12,000 column (Pharmacia) that was equilibrated and eluted (0.5 ml/min) with the corresponding buffer. Fractions were collected at 0.5-min intervals and absorbance at 280 nm and radioactivity were monitored.
Nondenaturing Gel Electrophoresis-Purified ECaSt/PDI and the ECaSt/PDI⅐ 125 I-rBPTI complex, isolated by gel filtration under condi-tions described above, were electrophoresed under nondenaturing conditions as described for the electrophoresis of acidic proteins using a high pH discontinuous buffer system (63). The lane containing purified ECaSt/PDI was cut from the gel and stained with Coomassie Blue. The presence of the ECaSt/PDI⅐ 125 I-rBPTI complex in other lanes of the gel was detected immediately after electrophoresis by exposing the wet, unstained gel to x-ray film (Kodak X-Omat) for 24 h at 4°C.
Acid Gel Electrophoresis-ECaSt/PDI (20 g) and rBPTI (7 g) were mixed in 20 l of buffer containing 100 mM Tris-HCl (pH 7.4), 0.5 mM EDTA, 0.5 mM EGTA at 22°C for 30 min. The mixture was then subjected to acidic gel electrophoresis under nondenaturing conditions as described (64). Fig. 1 are the six mammalian PDI domains (4) placed in corresponding positions in the block diagrams of the primary structures of ECaSt/PDI and its C-terminal truncated mutant. The boundaries of the domains were assigned on the basis of sequence homology. The domains a and a of ECaSt/ PDI are 32% identical, share 59 and 62% identity with the corresponding domains in mammalian PDI and each one contains the thioredoxin-like active site (CGHC). The domains b and b in ECaSt/PDI are 20% identical and share 50 and 51% identity with the corresponding domains in mammalian PDI. The e domain is 50% identical to the e domain of mammalian PDI and 20% identical to the human estrogen receptor. The c domain of ECaSt/PDI contains the last 29 residues (451-479) of the C-terminal to correspond with the 29 residues (463-491) of the c domain of human PDI (18) and they have a 28% identity. The chosen human c domain originally contained 43 residues (4). Interestingly, the more recent changed number of residues (463-491) (18,35) in the human c domain we had previously designated the "tail end" of its C-terminal (23). The C-terminal truncated mutant (⌬ 385-475 ) lacks part of the a and the c domain. It was designed to retain the thioredoxin active site at residues 374 -377 and to truncate the cluster of paired acidic residues in the C-terminal.

Apparent Molecular Weights of Recombinant ECaSt/PDI and the C-terminal Truncated Mutant-Recombinant
ECaSt/ PDI and the mutant were produced in E. coli with yields and purity similar to those reported for recombinant human PDI (55). Full-length ECaSt/PDI and the C-terminal truncated mutant migrated at 58 and 45 kDa, respectively, on SDS-PAGE (Fig. 2)  were studied in the medium described for native ECaSt/PDI (20) at two different concentrations of KCl. At 50 mM recombinant ECaSt/PDI bound 26 mol of Ca 2ϩ /mol of protein with a K d of 2.8 mM (Fig. 3A). The truncated mutant bound 14 mol of Ca 2ϩ /mol of protein with a K d of 2.8 mM (Fig. 3A). At 150 mM KCl the K d for Ca 2ϩ binding was 5.2 mM for both molecules while the maximal Ca 2ϩ binding was 25.6 and 14.2 mol of Ca 2ϩ /mol of protein for ECaSt/PDI and the truncated mutant, respectively (Fig. 3B). Therefore the C-terminal truncated ECaSt/PDI showed a 46% reduction in the binding capacity which interestingly corresponds to a reduction of 45% of pairs of acidic residues, a reduction of 9 out of a total of 20 pairs in ECaSt/PDI. This is in keeping with our previous hypothesis that carboxyl pairs are the main low affinity Ca 2ϩ -binding sites (23). The above increasing K d values for Ca 2ϩ binding to ECaSt/PDI with the increments of ionic strength are expected due to the weak ionic interactions associated with low affinity Ca 2ϩ binding.
Effect of Ca 2ϩ on the PDI Isomerase Activity Using Insulin or Reduced RNase as Substrates-Under the conditions optimized for mammalian PDI (23), recombinant ECaSt/PDI and its Cterminal truncated mutant displayed similar specific activities on rRNase and insulin as substrates. The refolding of reduced rRNase and the reduction of insulin by GSH were catalyzed by both ECaSt/PDI and the mutant with specific activities of 0.21 mol of RNase renatured/min/mol of protein and 0.2 mol/ min/mg of protein, respectively. Thus the deletion of the Cterminal, involving part of the a domain and c domain, does not seem to affect ECaSt/PDI activity when insulin or rRNase are substrates. Those activities were augmented moderately, 1.8-and 2.2-fold, respectively, by 5 or 10 mM CaCl 2 ; 1 mM CaCl 2 or 10 mM MgCl 2 had negligible effects (data not shown).
Effects of Ca 2ϩ or Polylysine on the Reactivation of rBPTI by ECaSt/PDI, the C-terminal Truncated Mutant and Mammalian PDI-We studied the effects of ECaSt/PDI, its C-terminal truncated mutant, and mammalian PDI on reactivation of rBPTI in the absence and presence of Ca 2ϩ or polylysine. Reactivation of rBPTI was assayed by monitoring the rate of recovery of its inhibitory activity on trypsin. The assay carried out in the absence of redox buffer resulted in a slow rate of spontaneous reactivation of rBPTI which displayed biphasic kinetics with a lag phase of about 60 min preceding a linear increment (Fig. 4, A-D). Fifty percent (t1 ⁄2 ) of spontaneous reactivation of rBPTI occurred in 220 min and was insensitive to Ca 2ϩ or polylysine (data not shown). This reaction was done at pH 7.4, slightly alkaline, to promote spontaneous disulfide bond formation and reactivation but at a rate slow enough to assess any influence of a possible activator such as Ca 2ϩ (Fig. 4).
Surprisingly, recombinant ECaSt/PDI prevented this spontaneous reactivation observed over a period of 3 days in the absence of Ca 2ϩ (Fig. 4A). Native ECaSt/PDI purified from the sea urchin eggs had a similar effect (data not shown). The addition of 5 mM CaCl 2 (or 10 M polylysine) to the solution prior to mixing of ECaSt/PDI with rBPTI accelerated the reactivation of rBPTI by ECaSt/PDI, the t1 ⁄2 was reduced from 220 to 85 min (Fig. 4A). The addition of 5 mM MgCl 2 to the solution, prior to mixing of the two proteins, had no effect on the inhibition of rBPTI reactivation (Fig. 4, panel A). Neither 5 mM CaCl 2 nor 10 M polylysine reverted the inhibitory effect of ECaSt/PDI when added after both proteins were mixed (not shown).
The C-terminal truncated mutant, in contrast to full-length ECaSt/PDI, catalyzed the reactivation of reduced BPTI with a t1 ⁄2 of 149 min in the absence of CaCl 2 (Fig. 4B). This reactivation was accelerated by 5 mM CaCl 2 or 10 M polylysine (t1 ⁄2 ϭ  80 min) and was insensitive to 5 mM MgCl 2 (Fig. 4B). We confirmed the previous finding that mammalian PDI reactivated rBPTI in the absence of CaCl 2 (3,61). Under our experimental conditions the reactivation showed a t1 ⁄2 of 63 min (Fig.  4, panel C) that was further reduced to 32 min (Fig. 4, panel C) by 5 mM CaCl 2 . The reactivation of rBPTI by ECaSt/PDI, the mutant, and mammalian PDI are compared in Table I.
These experiments were also done at pH 7.0 in the presence of 150 mM KCl, in the same medium used for Ca 2ϩ binding assay, and the effects of Ca 2ϩ at 1, 2.5, and 5 mM were tested. As expected the spontaneous reactivation took place at a much reduced rate, too slow to determine the t1 ⁄2 (Fig. 5, A and B). However, the effects of ECaSt/PDI and the C-terminal truncated mutant were essentially similar to those described in Fig.  4. ECaSt/PDI prevented the spontaneous reactivation of rBPTI in the absence of Ca 2ϩ . A progressive acceleration of rBPTI reactivation was observed with increasing concentration of Ca 2ϩ . The t1 ⁄2 for reactivation were 240, 170, and 87 min in the presence of 1, 2.5, and 5 mM Ca 2ϩ , respectively (Fig. 5A). Thus, 5 mM Ca 2ϩ further augmented the activation by 1 mM Ca 2ϩ 2.8-fold. The C-terminal truncated mutant reactivated rBPTI with t1 ⁄2 of 125 min in the absence of Ca 2ϩ . The reactivation increased progressively with t1 ⁄2 of 110, 90, and 75 min in the presence of 1, 2.5, and 5 mM Ca 2ϩ , respectively (Fig. 5B).
Thus 1-5 mM Ca 2ϩ augments the reactivation of rBPTI by ECaSt/PDI, by its C-terminal truncated mutant and by mammalian PDI. The C-terminal truncated ECaSt/PDI behaves essentially like mammalian PDI. Both molecules enhance the reactivation of rBPTI in the absence of Ca 2ϩ and Ca 2ϩ further augments the reactivation. It appears therefore that the Cterminal portion of ECaSt/PDI differs from that of mammalian PDI and somehow allows for complex formation of the molecule with rBPTI in the absence of Ca 2ϩ , a state which prevents the spontaneous refolding of rBPTI. Once the complex is formed Ca 2ϩ or polylysine cannot dissociate the molecules. Such a complex did form and is described below.  It should be noted that there are quantitative differences between the activation by mammalian and sea urchin PDI under the same in vitro conditions. ECaSt/PDI and the mutant have less ability to reactivate rBPTI than does mammalian PDI (see Table I). Similarly, the activity of ECaSt/PDI in a previous study was less; it had 30% of the activity of mammalian PDI when insulin and rRNase were the substrates (23).
Effects of Polyglutamate on the Reactivation of rBPTI-On the supposition that certain glutamate and/or aspartate residues in the C-terminal of ECaSt/PDI were responsible for the prevention of rBPTI reactivation in the absence of Ca 2ϩ , we tested the effect of synthetic polyglutamates on this process. No measurable spontaneous reactivation of rBPTI was observed up to 3 days when poly(Glu) (average molecular mass ϭ 31.4 kDa), poly(Glu,Ala,Tyr) (average molecular mass ϭ 50 kDa), or poly(Glu,Tyr)(average molecular mass ϭ 39 kDa) were present at 10 M concentration in the reactivation medium. Thus the polyglutamates mimicked the behavior of ECaSt/PDI. Only the effect of the homopolymer polyglutamate is presented (Fig. 4,  panel D).
Isolation of an ECaSt/PDI⅐ 125 I-rBPTI Complex by Gel Filtration-The prevention of rBPTI reactivation by ECaSt/PDI (Fig. 6A) in the absence of Ca 2ϩ suggested the formation of a complex between the two molecules. To study this possibility 125 I-rBPTI and ECaSt/PDI were mixed in the presence and absence of 5 mM CaCl 2 and the mixtures were analyzed by gel filtration chromatography. Fig. 6A shows the elution peak of ECaSt/PDI detected by absorbance at 280 nm. Fig. 6B shows the overlapping radioactive and absorbance peaks of 125 I-rBPTI. A mixture of ECaSt/PDI and 125 I-rBPTI in the absence of Ca 2ϩ , eluted as a complex with overlapping absorbance and radioactive peaks, with a retention time close to that of ECaSt/ PDI alone (Fig. 6C). When 5 mM CaCl 2 was added to the incubation medium before ECaSt/PDI and 125 I-rBPTI were mixed, two distinct peaks of radioactivity were detected; a large radioactive peak that co-eluted with rBPTI and a small radioactive peak that co-eluted with ECaSt/PDI (Fig. 6D). Thus, in the absence of Ca 2ϩ , rBPTI and ECaSt/PDI form a complex that is stable during gel filtration and the presence of Ca 2ϩ prevented the complex formation.
Such complex formation was confirmed by gel electrophoresis under nondenaturing conditions and by autoradiography. An elution sample containing the ECaSt/PDI⅐ 125 I-rBPTI complex in the absence of Ca 2ϩ (see peaks in Fig. 6C) migrated electrophoretically as a complex; the autoradiogram in Fig. 7A, lane 2, shows a single radioactive band that co-migrates with ECaSt/PDI in the absence of Ca 2ϩ (lane 2 also shows that no detectable radioactive band was associated with the minor protein impurities traveling above ECaSt/PDI (lane 1)). On the other hand, a sample containing mainly ECaSt/PDI in the presence of Ca 2ϩ (see peaks in Fig. 6D) shows no radioactive band (lane 3), thus confirming that Ca 2ϩ dissociated the complex.
Redox State of BPTI in the Complex-To determine if rBPTI in the complex with ECaSt/PDI remained reduced or underwent some partial reoxidation, the complex was electrophoresed in an acidic nondenaturing gel which resolves reduced and partially or completely oxidized forms of BPTI (64). The acidity dissociates the complex and a single band is seen in Fig. 7B, lane 1, traveling with identical mobility of rBPTI (Fig. 7B,  upper band in lane 2). The lower band in lane 2 is nBPTI. The absence of protein bands below the rBPTI band in lane 1 indicates the absence of partially oxidized BPTI intermediates which might have been formed in the complex.

DISCUSSION
The correspondence of a 46% reduction of Ca 2ϩ binding with a 45% reduction in the number of paired acidic residues in the C-terminal truncated mutant compared with the complete ECaSt/PDI supplements our recent findings which show that the maximal Ca 2ϩ uptake by Chinese hamster ovary cell microsomes containing the full-length ECaSt/PDI is 41% higher than the maximal uptake in microsomes containing the Cterminal truncated ECaSt/PDI (41). These findings lend support to our hypothesis that paired acidic residues constitute low affinity calcium-binding sites, but not to the exclusion of other low affinity sites formed by single carboxyls paired with other groups (23). The designated c domain, the tail end of the C-terminal of ECaSt/PDI contains 7 pairs of acidic residues out of a total of 20 and would constitute according to our hypothesis 35% of the total binding capacity. Applying the same approximations, the c domain in human PDI would bind 37% of the total binding capacity. On the other hand, in calreticulin the defined c domain which contains 19 pairs of acidic residues apparently binds all 20 molecules of Ca 2ϩ (48).
Calsequestrin does not have sufficient carboxylic pairs in its primary structure to account for the number of low affinity Ca 2ϩ bound but has 110 acidic residues. If our hypothesis is correct, the necessary carboxyl pairs, or a single one in association with another group, would be formed in the folded molecule (23). The crystal structure of calsequestrin from rabbit skeletal muscle has now been determined and the authors describe the presence of three folded thioredoxin domains each having a hydrophobic core with acidic residues having 13-36 negative charges on the surface generating an electronegative potential (65). This together with clefts formed in the dimeric state constitute the sites for low affinity Ca 2ϩ binding. The number of pairs of negatively charged groups that may have been formed in these configurations is not given (65). MacLennan and Reithmeier (66), on reviewing the significance of this article, succinctly place these and the other structural findings into the context of the function of the sarcoplasmic reticulum and the ER in intracellular Ca 2ϩ regulation. With regard to calcium storage in relation to its release, calsequestrin is better understood than any of the calcium storage proteins in the ER. For example, information is evolving on its coupling to the ryanodine Ca 2ϩ release channel involving conformational changes (67) and connections with the junctional membrane (68), triadin (69), and junctin (70).
On determining that both sea urchin and mammalian PDI could bind an abundant number of calcium ions, we supposed that Ca 2ϩ would play a role in their isomerase activities at concentrations in the range of and above their K d values for Ca 2ϩ binding. And indeed, 5 mM Ca 2ϩ augmented the activity of ECaSt/PDI and the C-terminal truncated mutant about 2-fold, a moderate degree, using RNase and insulin as substrates. Primm et al. (71) using the same concentration of Ca 2ϩ and denatured RNase as a substrate, obtained a negligible effect on the activity of mammalian PDI. However, the chaperone and anti-chaperone activities of PDI were modulated by Ca 2ϩ in the millimolar range of concentration, with a maximal effect at around 5 mM (71).
The effect of Ca 2ϩ was more dramatic on the activity of ECaSt/PDI when rBPTI was the substrate (Fig. 4). In the absence of Ca 2ϩ , ECaSt/PDI prevented the spontaneous reactivation of rBPTI due to a complex formation between the two molecules which was determined by gel filtration and electrophoresis under nondenaturing conditions. The presence of 5 mM Ca 2ϩ prevented complex formation with ECaSt/PDI and resulted in a 2.6-fold augmentation of the spontaneous reactivation of rBPTI at pH 7.4. At pH 7.0, 1 mM Ca 2ϩ reduced markedly the immeasurable t1 ⁄2 of spontaneous reactivation to 240 min which was further accelerated by increasing concentrations of Ca 2ϩ ; 5 mM Ca 2ϩ further augmented the reactiva- The C-terminal truncated mutant, in contrast to ECaSt/PDI, displayed enzymatic activity in the absence of Ca 2ϩ and the activity was stimulated about 2-fold in the presence of 5 mM Ca 2ϩ at pH 7.4. At pH 7.0 a progressive acceleration occurred with increasing concentrations of Ca 2ϩ ; in 5 mM Ca 2ϩ the reactivation was increased 1.7-fold. Therefore, regions in part of the a domain and the c domain of the C-terminal end of ECaSt/PDI, the sequence truncated in the mutant, are involved in stable interactions and complex formation with rBPTI in the absence of Ca 2ϩ . The binding sites in these regions of ECaSt/ PDI are presumably acidic negatively charged residues which in the presence of Ca 2ϩ were shielded and were not available for complexing with cationic sites in rBPTI. These negatively charged residues were apparently similarly neutralized by polylysine which, like Ca 2ϩ augmented the reactivation of rBPTI (Fig. 4A). The suggestion that negatively charged residues in the C-terminal of ECaSt/PDI are responsible for the complex formation and consequent inhibition of the spontaneous reactivation of rBPTI is supported by a similar effect of polyglutamate on rBPTI (Fig. 4D). The apparent shielding of negative charges by polylysine at low concentrations (10 M) suggests the possibility that ER proteins with a cluster of basic residues could shield negative charges in ECaSt/PDI resulting in a modulation of its interactions with other proteins.
Unlike ECaSt/PDI, mammalian PDI catalyzes the refolding of rBPTI in the absence of Ca 2ϩ (Fig. 4C, also see Refs. 3, 61, and 72) and others found that the deletion of its c domain had no effect on the catalysis of disulfide bond rearrangement in BPTI intermediates as substrates (36). Furthermore, a fragment of BPTI could not be cross-linked to the isolated a-c domains of mammalian PDI in the absence of Ca 2ϩ , suggesting that these mammalian PDI domains do not interact strongly with BPTI (38). Hence, the C-terminals of ECaSt/PDI and mammalian PDI which have a 44% identity must differ in charge distributions and electrostatic potential on their surfaces even though they have a similar number of acidic residues (23). ECaSt/PDI and mammalian PDI also differed in the degree of isomerase activities. ECaSt/PDI has 30% of the activity of mammalian PDI on substrates rRNase and insulin (23) and in this investigation similarly augmented to a lesser extent the reactivation of rBPTI than did mammalian PDI (see Table I).
These in vitro results demonstrate a role of Ca 2ϩ on mammalian PDI and ECaSt/PDI isomerase activity. In cells in culture, Ca 2ϩ also appears to play a role in maintaining the structure of ER proteins. Lodish and co-workers (73) have shown, by the use of Ca 2ϩ ionophores, an inhibition of the exit of secretory proteins, to different extents depending on the protein, probably due to a disruption of their folding associated with the depletion (or reduction) of Ca 2ϩ in the ER. In another study, Ca 2ϩ ionophores and thapsigargin impaired the maturation and proper disulfide bonding of an expressed "model protein," the H1 subunit of the asialoglycoprotein receptor, containing 8 cysteine residues in its exoplasmic domain (74). The authors speculate that the unfolded state of the H1 subunit, probably due to the reduced PDI activity in the absence (or in reduced concentration) of Ca 2ϩ , may be the cause preventing its exit from the ER. However, ionophores and thapsigargin had a marginal effect on secretion of albumin (73,74), a protein containing many cysteine residues, 35 in human serum albumin. If the H1 subunit and albumin are substrates of PDI in vivo then their PDI-catalyzed folding differs in response to the decrease of Ca 2ϩ concentration in the ER. Similarly, our in vitro results also show differences in Ca 2ϩ -dependent responses of the substrates RNase, insulin, and rBPTI. Furthermore, mammalian PDI and ECaSt/PDI respond differently to Ca 2ϩ for a given protein substrate. The depletion (or reduction) of Ca 2ϩ in the ER also facilitates the dissociation of variants of another expressed protein, the T-cell antigen receptor ␣ chain from the immunoglobulin heavy chain binding protein (BiP) (75). In summary, proper folding (73), including disulfide bonding (74) and protein-protein interactions (75) in the secretory pathway of the cell seem to be effected by the Ca 2ϩ content in the ER and interestingly, different secretory proteins are affected differently.
Our in vitro experiments show that the effect of Ca 2ϩ , in the millimolar range of concentrations, on ECaSt/PDI activity was dependent on the protein substrate. With insulin or RNase the activity was insensitive to 1 mM Ca 2ϩ and moderately stimulated at 5 mM Ca 2ϩ . However, ECaSt/PDI activity on rBPTI was absolutely dependent on Ca 2ϩ which produced a substantial effect at 1 mM concentration, by dissociating ECaSt/PDI from rBPTI and then accelerating the activity of ECaSt/PDI. Our studies also show activation of mammalian PDI by 5 mM Ca 2ϩ . FIG. 7. Identification of ECaSt/PDI⅐rBPTI complex by native gel electrophoresis and the redox state of BPTI in the complex by acidic gel electrophoresis. Mixtures of ECaSt/PDI and 125 I-rBPTI in the absence and presence of Ca 2ϩ were subjected to gel filtration. Panel A, aliquots (20 l) of elution samples corresponding to the absorption peak of ECaSt/PDI in the absence of Ca 2ϩ (see Fig. 5C) or presence of 5 mM CaCl 2 (see Fig. 5D) were electrophoresed under nondenaturing conditions, dried, and autoradiographed as described under "Experimental Procedures." Lane 1, gel containing 2 g of ECaSt/ PDI was excised after completion of the electrophoresis and stained with Coomassie Blue to identify the mobility of ECaSt/PDI. Lane 2, autoradiography of sample from ECaSt/PDI⅐ 125 I-rBPTI complex peak in Fig. 5C in the absence of Ca 2ϩ shows a 125 I-rBPTI band at the site of ECaSt/PDI indicative of a complex. Lane 3, autoradiography of sample from ECaSt/PDI peak in Fig. 5D in the presence of Ca 2ϩ shows no complex with 125 I-rBPTI. Note that the minor protein impurities traveling above ECaSt/PDI (Lane 1) did not show a corresponding radioactive band (Lane 2) suggesting that they did not form a complex with rBPTI. B, acidic gel electrophoresis of the ECaSt/PDI⅐rBPTI complex. Lane 1 shows rBPTI released from the complex traveling with identical mobility of rBPTI in lane 2. Lane 2 shows the mobilities of rBPTI (5 g) and native BPTI (3 g). Lane 1 shows no bands below the rBPTI band which would be due to partial oxidation.
Whether the modulations of Ca 2ϩ concentrations around the K d values for Ca 2ϩ binding to PDI play a role in regulating PDI activity in the ER of a cell remains to be determined. Also worthy of further considerations is whether Ca 2ϩ affects any particular substrate directly.
The continuing studies on the domains of mammalian PDI and ECaSt/PDI and the possible future elucidation of their tertiary structures will lead to a better understanding of their roles as isomerases and Ca 2ϩ storage proteins and the relationship between these two functions within the endoplasmic reticulum.