BiP, a Major Chaperone Protein of the Endoplasmic Reticulum Lumen, Plays a Direct and Important Role in the Storage of the Rapidly Exchanging Pool of Ca2+ *

The activity of BiP, the major chaperone of the endoplasmic reticulum (ER) lumen, is known to be Ca2+-regulated; however, the participation of this protein in the ER storage of the cation has not yet been investigated. Here such a role is demonstrated in human epithelial (HeLa) cells transiently transfected with the hamster BiP cDNA and incubated in Ca2+-free medium, as revealed by two different techniques. In the first, co-transfected aequorin was employed as a probe for assaying either the cytosolic of the mitochondrial free Ca2+ concentration. By this approach higher Ca2+ release responses were revealed in BiP-transfected cells by experiments in which extensive store depletion was induced either by repetitive stimulation with inositol 1,4,5-trisphosphate-generating agonists or by treatment with the Ca2+ ionophore, A23187. In the second technique the cells were loaded at the equilibrium with 45Ca, and the release of the tracer observed upon treatment with thapsigargin, a blocker of the ER Ca2+ ATPases, was larger in BiP-transfected than in control cells. The latter results were obtained also when BiP was overexpressed not via transfection but as a response to ER stress by tunicamycin. These results are sustained by increases of the ER Ca2+ storage capacity rather than by artifacts or indirect readjustments induced in the cells by the overexpression of the chaperone since (a) the exogenous and endogenous BiP were both confined to the ER, (b) the expression levels of other proteins active in the ER Ca2+ storage were not changed, and (c) effects similar to those of wild type BiP were obtained with a deletion mutant devoid of chaperone activity. The specificity of the results was confirmed by parallel 45Ca experiments carried out in HeLa cells transfected with two other Ca2+-binding proteins, calreticulin and CaBP2(ERp72), only the first of which induced increases of Ca2+ capacity. We conclude that BiP has a dual function, in addition to its chaperone role it is a bona fide ER lumenal Ca2+ storage protein contributing, under resting cell conditions, to around 25% of the store, with a stoichiometry of 1–2 moles of calcium/mole of BiP.

Studies carried out during the last decade have identified BiP as a ubiquitous lumenal resident protein of the endoplasmic reticulum (ER) 1 (1)(2)(3), playing a key role in the assistance of newly synthetized proteins for folding and acquisition of their correct tertiary and quaternary structure. Such a function, defined as a chaperone, implies the direct binding of BiP to the growing chains, with a stimulation of its ATPase activity (6 -9). When this task is not accomplished, misfolded proteins remain complexed to BiP (and to other chaperones) (7,10) to be ultimately disposed of by nonlysosomal proteolytic process(es) (quality control) (7, 9 -11). These fundamental activities are carried out by BiP working in the peculiar environment of the ER lumen (12) characterized, among other properties, by a high (mM) concentration of free Ca 2ϩ ([Ca 2ϩ ] er ) (13,14). Such a property of the environment appears to be of importance for BiP function. Indeed, previous studies carried out with both recombinant BiP protein and BiP purified from cells have shown that the chaperone ATPase activity is altered by Ca 2ϩ , and BiP association with proteins in vitro is stabilized by Ca 2ϩ (8,15,16). Moreover, major changes of Ca 2ϩ homeostasis, such as those induced by prolonged applications of Ca 2ϩ ionophores, have been shown to induce major disturbances in the cells, including inappropriate release and secretion of BiP-associated proteins (17), alterations in the folding of ER proteins leading to the transcriptional up-regulation (3,18), and altered distribution of BiP and other chaperones (19). So far, however, the possibility that BiP, in addition to its chaperone function, could also participate in the intralumenal storage of Ca 2ϩ , thus contributing to the accumulation of the rapidly exchanging pool of the cation, has never been investigated.
In the present study we report that overexpression of exogenous hamster BiP in HeLa cells induces appreciable increases of the ER Ca 2ϩ storage capacity as consistently revealed by two experimental approaches. In one, cells were cotransfected with the cDNA of the photoprotein aequorin that had been molecularly modified to assay [Ca 2ϩ ] within two distinct intracellular compartments, either the cytosol or the mitochondria (13,20). In the other, the cells were loaded at the equilibrium with 45 Ca and release was induced by exposure to thapsigargin (Tg), a blocker of the sarcoplasmic/endoplasmic reticulum Ca 2ϩ pump (SERCA). The latter treatment is known to induce the specific discharge of the ER compartment (13,21). The Ca 2ϩ storage effects observed with BiP overexpression were not due to its chaperone activity toward other proteins, since a BiP construct that lacks the protein binding domain produced similar results. Nor were these effects due to compensatory readjustments in the levels of other ER Ca 2ϩ -binding proteins, because the latter remained unchanged in the transfected cells. Moreover, parallel studies carried out in HeLa cells exposed to a stressful treatment with the glycosylation blocker drug, tunicamycin (22), also led to an increase of the ER Ca 2ϩ storage capacity, part of which is dependent on the stress-induced overexpression of BiP. We conclude that, in addition to its classical role of chaperone, BiP participates in a second function, the storage of Ca 2ϩ within the ER lumen, and that the fluctuations of the protein induce adjustments of the cellular Ca 2ϩ homeostasis.
Eukaryotic Expression Vectors-The eukaryotic expression vectors employed for the transfection experiments have been described elsewhere. The pMT vectors containing the cDNA of the wild type hamster BiP (wt-BiP/pMT) and of the 44K amino-terminal ATPase domain (lacking the protein binding domain but retaining the carboxyl-terminal 11 amino acids required for recognition by the anti-BiP antiserum and for ER retention: 44K⌬BiP/pMT) were described elsewhere (23); both the pcDNAI vectors containing either the cDNA of aequorin targeted to either the cytosol (cytAEQ/pcDNAI) or to mitochondria (mtAEQ/pcDNAI) as well as the cDNA of CRT (tCR/pcDNAI) were described in Bastianutto et al. (13) and Brini et al. (28); the pCMV 2 vector including the cDNA of rat CaBP 2 (ERp72) (CaBP 2 (ERp72)/ pCMV2) was described in Van et al. (29).
Cell Culture and Transfection-HeLa cells were grown at 37°C in 75-cm 2 Falcon flasks, bathed in Dulbecco modified Eagle's medium supplemented with 10% fetal calf serum and 100 units/ml penicillin and streptomycin, under a 5% CO 2 atmosphere. Transient expression of BiP, CRT, CaBP 2 (ERp72), and aequorin (aequorin addressed either to the cytosol or to mitochondria) were obtained by transfection of subconfluent cells with the corresponding cDNAs, using the Ca 2ϩ phosphate technique (30). The percentage of cells expressing wild type and the 44K⌬ mutant BiP (35-50%, depending on the experiments) was determined by immunofluorescence staining using the polyclonal anti-BiP antiserum that recognizes rodent but not human BiP (23). For 45 Ca release, HeLa cells were transfected with either wt-BiP/pMT, tCR/pcDNAI, or CaBP 2 (ERp72)/pCMV 2 using the same technique described above. For the aequorin experiments, the cells were plated onto 13-mm diameter round coverslips, and cotransfection with 4 g/well plasmid DNA (cytAEQ/pcDNAI or mtAEQ/pcDNAI, alone or together with either wt-BiP/pMT or 44K⌬BiP/pMT in a 1:1 ratio) was carried out. Aequorin measurements and calibration of the aequorin signal were carried out 36 h after transfection, as described previously (28).
Immunofluorescence and Electron Microscopy-Cell monolayers were fixed with a mixture of 4% paraformaldehyde and 0.25% glutaraldehyde in 125 mM phosphate buffer, pH 7.4 for 1 h. After a short treatment with 1% Na-borohydrate (to eliminate glutaraldehyde fluorescence) they were washed and then exposed for 30 min to a solution containing 0.3% Triton X-100, 15% filtered goat serum, 0.45 M NaCl, and 10 mM phosphate buffer, pH 7.4. After washing, the sections were exposed (1 h at 37°C) to either one of the primary antibodies diluted in the above Triton X-100 and goat serum-containing solution. They were then washed thoroughly and treated with the appropriate rhodaminelabeled sheep anti-first antibody (30 -60 min, 37°C, purchased from Technogenetics, Milan, Italy; diluted 1:20 -1:40 in the Triton X-100, goat serum solution). After another thorough washing the sections were mounted in glycerol and then examined in a laser scanning confocal microscope (MRC 1024, Bio-Rad House, Hemel Hempstead, Hertfordshine, UK). For additional details, see Raichman et al. (31).
For electron microscopy the cell monolayers were fixed as above, however, with 1% glutaraldehyde and 2.5% paraformaldehyde. After washing with the buffer they were postfixed with 2% OsO 4 in phosphate buffer, washed, and then embedded in Epon. Ultrathin sections cut perpendicularly to the monolayers were stained with uranyl and lead citrate and then studied in a Hitachi H7000 electron microscope (31).
For electrophoresis and Western blotting analyses, protein suspensions were solubilized in SDS-sample buffer (48 mM Tris-HCl, pH 6.8, 0.8 M sucrose, 4% SDS, 8% ␤-mercaptoethanol, 0.008% bromphenol blue in a ratio 100 -70 (sample/mix solution)) and heated on a boiling water bath for 5 min. The solubilized proteins were loaded and separated in Laemmli-type SDS-polyacrylamide gel electrophoresis gels (typically 4 and 7% polyacrylamide for the stacking and the running gels, respectively) which were run as described elsewhere (26). The separated proteins were electrotransferred in a cold room onto a 0.2-m pore nitrocellulose membrane at constant 70 mA for 16 h in a buffer containing 25 mM Tris, 192 mM glycine, 20% methanol, pH 8.3. After brief staining with 0.2% (w/v) Ponceau Red in 3% trichloroacetic acid to reveal the standard molecular weight markers, the blots were blocked with 5% (w/v) low fat dried milk in PBS supplemented with 0.01% (v/v) Tween-20 (PBS-Tween) for 1 h at room temperature and then exposed to the primary antibodies either for 1 h at room temperature (for the polyclonal antibodies specific for either hamster BiP, PDI, CNX, CRT, IP 3 R, or CaBP 2 (ERp72)) or overnight at 4°C (for the rat monoclonal antibody which recognizes both the rodent and human BiP). Blots were then washed in 2.5% (w/v) milk in PBS-Tween (five times for 10 min each). For the polyclonal antibodies specific for either hamster BiP, PDI, CNX, CRT, IP 3 R, or CaBP 2 (ERp72), immunostaining was then developed by the ECL kit (for further details, see Rooney and Meldolesi (32). In the case of the rat monoclonal antibody which recognizes both the rodent and human BiP, blots were incubated with an anti-rat Ig (HϩL) rabbit antibody (diluted 1 to 1000, in 2.5% (w/v) milk in PBS-Tween) for 1 h. After washing (five times for 10 min each, with 2.5% (w/v) milk in PBS-T), the blots were labeled with 125 I-protein A, washed, dried, and finally autoradiographed at Ϫ80°C for variable periods of time. The relevant bands revealed using either ECL kit or 125 I-protein A was quantitated using a Molecular Dynamics Imagequant apparatus (31). Results shown are representative of three to five separate experiments. 45 Ca Measurements-Control and transfected HeLa cells were grown as described above, except that during the last 48 h their incubation medium was supplemented with 45 Ca (1.5 Ci/ml). At the end of this period the labeled cells were rapidly washed in Krebs-Ringer-Hepes (KRH) medium containing 125 mM NaCl, 5 mM KCl, 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 2 mM CaCl 2 , 6 mM glucose, 25 mM Hepes-NaOH, pH 7.4). The cells were then resuspended at 37°C in 3 mM EGTA-containing Ca 2ϩ -free KRH medium, and Tg (0.1 M in Me 2 SO) was added quickly. Immediately before and 4 min after the addition of Tg, aliquots of 1 ϫ 10 6 cells were removed and centrifuged. The 45 Ca recovered from the medium and the pellets were assayed in a Beckman ␤-counter (for further details, see Fasolato et al. (33)).
Cell Stress Experiments-In the stress experiments monolayers of nontransfected HeLa cell were loaded with 45 Ca as above for a total of 72 h during the last 24 h of which they were exposed to tunicamycin (7 g/ml in Me 2 SO) (22). Controls received the solvent only. At the end of the incubations the cells were detached and analyzed for 45 Ca release by experiments as described above. Parallel samples were investigated by Western blotting.

Characterization of BiP-transfected HeLa Cells-The
HeLa cell system employed for the reported experiments was systematically characterized by biochemical and immunocytochemical studies. Fig. 1 shows confocal images of cells, transfected with either the wild type or the 44K amino-terminal portion of hamster BiP, that have been decorated with a purified polyclonal antibody raised against the 11-amino acid COOH-termi-nal sequence of the hamster protein. This antibody is specific for rodent BiP and does not recognize the human protein endogenous to HeLa cells (23) (see also Fig. 2A). From the immunocytochemical results, the percentage of transfected HeLa cells was established to vary in different experiments from 35 to 50%. Among positive cells, those heavily expressing the transfected protein were relatively rare, while most exhibited levels that remained well below saturation under the employed experimental conditions. Moreover, the distribution of both wild type and 44K⌬BiP proteins revealed the classical, ER-type reticular pattern (Fig. 1, B and C), similar to that given by other lumenal proteins such as CRT (not shown; see Bastianutto et al. (13). Further characterization of the BiP-overexpressing population was carried out at the electron microscope level. In these experiments thin sections of at least 20 cells group were analyzed, with special interest for the extension and distribution of the ER and for the size and shape of its cisternae. Such numbers guarantee the study of both transfected and non-transfected cells. An ultrastructural analysis was important because in a previous study the cells transfected with a BiP mutant (but not with the wild type or the 44K⌬BiP mutant employed here) were reported to exhibit fragmented cisternae (23), a typical sign of cell damage. In our samples, however, significant alterations of the ER were never observed (not shown). We conclude therefore that transfection of both wild type and 44K⌬ mutant BiP did not induce any unspecific cell damage appreciable by conventional electron microscopy.
The decoration of Western blots from transfected cells, car-ried out with the same antibody employed for immunocytochemistry, is shown in Fig. 2A. Notice that with the polyclonal antibody control cells were negative while the bands corresponding to the wild type and the 44K⌬BiP-transfected proteins were prominent. Fig. 2B shows in contrast the bands of control and wild type BiP-transfected cells revealed by the rat monoclonal, an antibody recognizing both the rodent (exogenous) and human (endogenous) protein, as well as a scale of bands obtained with increasing amounts of purified recombinant hamster BiP protein. Based on the calibration curves drawn from these (and other) results, the total level of BiP expression in the transfected cell population was estimated systematically in all experiments. On the average, such a level was found to be ϳ200% of that in controls. Fig. 2C shows blots of control and BiP-transfected cells decorated with antibodies raised against other ER proteins, the chaperones PDI, CNX and CRT, the latter being also the major Ca 2ϩ -binding protein of the system (34,35); as well as the IP 3 receptor. As can be seen (Fig. 2C), none of these components appeared appreciably different in the BiP-transfected preparations with respect to controls, suggesting that no significant changes in the expression of other ER proteins that contribute to Ca 2ϩ homeostasis had taken place. [Ca 2ϩ ] c and [Ca 2ϩ ] mt Assays with Aequorin-The effects of increased levels of BiP expression on Ca 2ϩ release from the ER compartment, as revealed by the cotransfected photoprotein, aequorin, trapped within the cytosol, are shown in Figs. 3 and 4. Control and BiP-overexpressing cells, bathed in the Ca 2ϩfree medium, were exposed to the Ca 2ϩ ionophore, A23187 (Fig.  3A). The signal generated by this treatment originates, however, not only from the ER, but also from any other stores with nonacidic lumena. Despite this limitation of the approach, in six out of six experiments the release signals from the cell populations with wild type BiP (and, to a slightly lower level, also with the 44K⌬BiP) were greater than those from controls, both in terms of maximal [Ca 2ϩ ] c peak and of total [Ca 2ϩ ] c response area (see the representative trace results in Fig. 3A and the average Ϯ S.D. of the six experiments in Fig. 3B).
In a subsequent series of experiments, the study of the cells cotransfected with BiP and aequorin was focused on the receptor, IP 3 -mediated responses of cytosolic free Ca 2ϩ , [Ca 2ϩ ] c , triggered by release of the cation from the ER. Previous studies of HeLa cells transfected with another lumenal Ca 2ϩ binding protein, CRT (13), had shown that the signaling differences sustained by differences in the capacity of the ER store are hard to reveal in response to a single shot given to cells, and become much more clear when the cells, incubated in Ca 2ϩ -free medium to prevent the contribution of Ca 2ϩ influx, are exposed sequentially to two short (30 s) treatments with agonists specific for different receptors (to prevent desensitization), first ATP and then histamine (Fig. 4)

or vice versa (not shown).
Similarly to what is observed with CRT (13), cells transfected with either wild type BiP or 44K⌬BiP were found in five experiments to show an initial [Ca 2ϩ ] c increase only moderately greater and more persistent than that of controls. In contrast, in the second response the difference became remarkable in all the experiments because the [Ca 2ϩ ] c rise was greatly reduced in the controls and still considerable in the transfected cells (on the average ϩ150 Ϯ 12% with wild type BiP). These last results document the persistence of a considerable Ca 2ϩ pool within the ER stores of transfected cells at a time when in controls these stores were largely depleted.
The question which was then raised was whether the increased [Ca 2ϩ ] c responses such as those observed in the BiPtransfected cells were restricted to the cytosol or could have a more vast impact on cellular Ca 2ϩ homeostasis. To investigate the problem, experiments were carried out by applying the same two response protocol used in Fig. 4, but this time to cells that had been cotransfected with an aequorin-containing construct addressed not to the cytosol but to mitochondria, the organelles known to depend largely from ER release for their Ca 2ϩ uptake (20). The results obtained by this approach, concerning cells transfected with either BiP or CRT, are shown in Fig. 5. In terms of size, the rises of mitochondrial free Ca 2ϩ , [Ca 2ϩ ] mt , in the BiP-transfected cells in response to the first stimulation were hardly different from those in controls, whereas those of the CRT-transfected cells were moderately increased. Persistence of these responses was slightly prolonged in both cases. Clear differences emerged, however, with the second stimulation. In control cells the [Ca 2ϩ ] mt responses were in fact decreased with respect to the first, although less markedly than those of [Ca 2ϩ ] c (cf. Figs. 4 and 5). In contrast, these second responses were largely maintained in the cells transfected with either lumenal Ca 2ϩ binding protein (Fig. 5). We conclude that the increased capacity for Ca 2ϩ of the ER can extend its impact on signaling to mitochondria where the cation is known to regulate various dehydrogenases and through them the entire Krebs cycle (36). 45 Ca Release-Further information about the functioning of the ER Ca 2ϩ stores in the BiP-transfected cells was obtained by assaying 45 Ca release after cell equilibrium loading with the isotope and exposure not to a ionophore but to the specific SERCA blocker, Tg. In this case the analyzed 45 Ca release is expected to originate from the ER store of all cells, and thus from both transfected and untransfected cells. Despite this limitation in the experimental approach, the Tg-induced release observed from the BiP-transfected population was distinctly greater as compared with controls as shown in Fig. 6. The same figure shows that even bigger increases could be obtained from HeLa cells transfected with CRT (total level of expression, 250%; see also Bastianutto et al. (13)). Parallel experiments were carried out with another Ca 2ϩ -binding protein, CaBP 2 , the rat homologue of the murine ERp72, a member of the chaperone PDI family (29,37,38). Blots of control and CaBP 2 (ERp72)-transfected HeLa cells, revealed by the antibody raised against the rat protein, showed a weak signal in control cells. In contrast, a prominent band was detected in the CaBP 2 (ERp72)-transfected cells (Fig. 2D), demonstrating that expression of the protein was considerably increased. Despite this result, the capacity of the Tg-sensitive store, as revealed by 45 Ca experiments, was not increased, but rather slightly lower in CaBP 2 -transfected cells with respect to controls (Fig. 6). Thus, appreciable increases of ER capacity for Ca 2ϩ do not necessarily follow the overexpression of any ER lumenal Ca 2ϩ -binding protein.
Increased expression of BiP can also occur in cells independent of cDNA transfection, being a well known aspect of the cell stress response. In view of the observed consequences of BiP overexpression on Ca 2ϩ homeostasis we were lent to investigate whether ER stress was also able to appreciably modify the Ca 2ϩ capacity of the ER and if so, whether BiP was involved. To carry out these experiments the cells were exposed to the N-glycosylation blocker, tunicamycin (22), applied after equilibration of Ca 2ϩ pools with 45 Ca. Compared with other stressful treatments, such as those with A23187 and Tg, treatment with tunicamycin was found to be particularly appropriate for our purposes since by itself the drug does not appear to have any direct effect on Ca 2ϩ homeostasis, whereas it does induce considerable increases of the ER chaperones, in particular of those that also serve as lumenal Ca 2ϩ binding proteins, such as BiP (22) and CRT (39). Increases of these proteins, appreciable as early as 6 h, were found to reach plateaus after 24 h, with average values in our experiments of 58% for CRT and 100% for BiP (Fig. 7A). Not surprisingly, these increases in expression were paralleled by marked increases (68%) of the 45 Ca release responses induced by Tg (Fig. 7B).
Calculations, carried out as described in Bastianutto et al. (13), based, on the one hand, on the levels of expression of the two proteins as established by Imagequant analysis of the Western blots developed from fractions of the same cell lysis used for the 45 Ca release experiments, and on the other hand on the resting Ca 2ϩ storage data in the ER of both control and transfected cells (Fig. 6) (13), revealed the contribution of CRT to account for ϳ29% and that of BiP for ϳ25% of the increased ER Ca 2ϩ capacity induced by Tg treatment.  Fig. 6 except that 1) all cells analyzed were nontransfected, and 2) loading with 45 Ca was for 72 h, during the last 24 h of which half of the cells were exposed to tunicamycin (7 g/ml in Me 2 SO) while the others received the solvent only. The Western blots of BiP and CRT in stressed and control (C) cell populations are shown to the left. 100% release corresponds to 550 cpm/mg of protein. Results shown are averages Ϯ S.D. of two highly consistent experiments performed in triplicate.

DISCUSSION
The present results, obtained by two independent techniques, consistently reveal increased capacity of the rapidly exchanging ER Ca 2ϩ store in the well characterized system of HeLa cells transiently transfected with either wild type BiP or its deletion mutant, 44K⌬BiP. The two techniques employed should be considered complementary of each other. Cotransfection of the cytosol-targeted aequorin (28) is advantageous not only because it provides the time course of the induced Ca 2ϩ release responses but also because the origin of the Ca 2ϩtriggered signal is restricted to cells expressing the photoprotein, which largely coincide with those overexpressing the cotransfected protein, i.e. the wild type or the 44K⌬BiP. This property is important because in our transfected preparations only 35-50% of the cells expressed exogenous BiP on top of their endogenous complement of the protein. However, the aequorin signals depend not only on the Ca 2ϩ capacity, but also on other properties of the ER store (pumps, channels) as well as on homeostatic equilibria established within the cell. Therefore, they cannot be easily converted into quantitative data. 45 Ca release, on the other hand, occurs from all the cells of the preparation (transfected and non transfected) which were first loaded at the equilibrium with the tracer. The advantage here is that, by using Tg, a blocker specific for the SERCAs, the results obtained are due to Ca 2ϩ release only from the ER (13).
The significance of our findings deserves attention. As in all other overexpression experiments, results could in fact be due not (or not only) to the increased amounts of the investigated protein but (also) to indirect consequences induced by protein transfection. In the case of BiP, this concern appears particularly cogent in view of the chaperone function of this protein, with positive action on structure, and thus also on function, of other proteins, some of which contribute to ER Ca 2ϩ homeostasis. However, the control results we have obtained do not seem to support these possibilities. The level of other proteins of the ER store, including CRT, which is known to be the major Ca 2ϩ binder (34,35), was in fact not significantly different in BiPtransfected and control cells. Changes of the ER store capacity due not to BiP but to BiP-induced changes of expression of other components appear therefore unlikely. Likewise, the consistence of the reported results obtained not only with BiP but also with its 44K⌬BiP mutant, which is devoid of the peptide binding domain and thus of any chaperone function (23), excludes the involvement of a chaperone-induced improved function of the other Ca 2ϩ -binding proteins. It appears therefore that the effects observed are most likely due to direct, specific Ca 2ϩ binding of the transfected protein. This conclusion is reinforced by the additional 45 Ca data obtained using cells transfected not with BiP but with either CRT or CaBP 2 -(ERp72). The results obtained on the one hand confirm that the contribution of CRT to the HeLa cell rapidly exchanging Ca 2ϩ store amounts to ϳ50% (13) and on the other hand fail to show any detectable effect of CaBP 2 . In the cells overexpressing the latter protein, in fact, the apparent capacity of the ER in terms of Ca 2ϩ storage was not increased but, if anything, slightly reduced.
A role of BiP as one of the Ca 2ϩ storage proteins of the ER lumen was proposed 10 years ago by Macer and Koch (40) who first observed a polyacrylamide gel electrophoresis band with an apparent molecular mass of ϳ78 kDa positively labeled by 45 Ca overlay. Subsequent studies by Van et al. (41), carried out by a more ample approach, revealed, however, that the 45 Capositive band of that size was not BiP but CaBP 2 (ERp72). In the meantime, evidence was accumulated demonstrating that for its chaperone function, in particular for the binding of proteins and ATPase activity (8,15,16), BiP needs Ca 2ϩ to be present in the medium. So far, however, the possibility that BiP contributes to the storage of the cation within the ER lumen, together with other proteins such as CRT, had never been envisaged. By the use of the aequorin approach in HeLa cells we demonstrate here that the contribution of the chaperone to the capacity of the typical rapidly exchanging Ca 2ϩ pool is not insignificant. In fact, when investigated according to a twostimulation protocol (13), BiP overexpression was found to induce detectable increases of the Ca 2ϩ release triggered by IP 3 generation following receptor activation. This type of difference is not easy to evidentiate because it is extensively buffered by cytosolic and organelle mechanisms (13,42). Moreover, extension of the approach to mitochondria, carried out by studying cells expressing an aequorin construct addressed to those organelles (20), revealed that the BiP contribution to the ERsegregated Ca 2ϩ pool has a wide impact within the cell. In fact, similar to the cells overexpressing CRT, those transfected with BiP were able to show considerable [Ca 2ϩ ] mt transient also in response to the second stimulation, when the responses of control cells were clearly reduced. In view of the importance of Ca 2ϩ in mitochondrial physiology (36) these observations further document the importance of the BiP storage function in cell physiology.
Interesting conclusions can be drawn also from the 45 Ca results. The ϳ25% increases of the Tg-induced release obtained from the transfected preparations, in which the overall BiP expression was twice as large as in the controls, suggest that the contribution of the endogenous chaperone corresponds to ϳ25% as well. To reconstruct the whole ER segregated Ca 2ϩ pool the BiP value should be added to that previously estimated for CRT, ϳ50%, with the rest probably contributed by other components, not only proteins (lumenal and also of the membrane, such as CNX) but also nucleotides (42). Taking into account that in HeLa cells the concentration of BiP is considerable (ϳ1% of the protein, ϳ5-fold higher than CRT), the above results suggest that the stoichiometry of ER lumenal Ca 2ϩ binding, under resting conditions, is relatively low, i.e. between 1 and 2 moles of calcium/mole of protein. Since in the BiP sequence typical Ca 2ϩ binding domains are not evident, it appears likely that the observed binding is due to the abundance of acidic amino acids, many of which arranged in doublets or triplets (43,44). Because of its role in Ca 2ϩ accumulation and release, and by analogy with other storage proteins such as CRT and calsequestrin (42), this binding is expected to be of low affinity, although so far conclusive results with the purified protein have not been reported.
A final comment concerns the results obtained with cells that, instead of being transfected, were exposed to tunicamycin to induce stress (27). Under these conditions, overexpression of endogenous BiP, due to stimulated transcription, is a classical result (22). Recently, a similar response has been reported for CRT (39). To our knowledge, however, neither the consequences of these events in terms of Ca 2ϩ accumulation within intracellular stores nor the possibility of modulations in the ER Ca 2ϩ capacity had so far ever been envisaged. Here we show that the capacity of ER stores, calculated on the basis of the 45 Ca results, as described previously for CRT-overexpressing cells (13), was considerably increased (68%). Based on the observed increases of both BiP and CRT, as revealed by Western blotting carried out in parallel, and on the ER resting Ca 2ϩ storage levels established by 45 Ca experiments with Tg, the two proteins were calculated to participate similarly to the overall increase, contributing ϳ25 and 29%, respectively, above controls. Thus, the role of these chaperones during stress is not limited to the assistance and quality control of the ER proteins but includes also an increase of the Ca 2ϩ buffering within the endomembrane system, a process that could contribute to the protection of the cell from the Ca 2ϩ -dependent damage.
In conclusion, the present results have added a new function to those already recognized for BiP, not only the classical chaperone, but also a Ca 2ϩ -binding protein playing an important role in the control of the ER lumenal Ca 2ϩ homeostasis. Although typical not only of BiP but also of other ER proteins, such as CRT and CNX (7,45), the duality of function does not appear to be the rule since overexpression of another ER lumenal protein, CaBP 2 (ERp72), remained without appreciable consequences on cell Ca 2ϩ homeostasis. This latter negative result with a protein that in vitro is known to bind Ca 2ϩ (38,41) emphasizes the need for Ca 2ϩ storage to be investigated also in living cells. At variance with CRT, where Ca 2ϩ binding and chaperoning are located in distinct domains (34,35), in BiP the sites for the two functions might be intermixed, as suggested by the lack of obvious segregation in the polypeptide sequence (43) and also by the Ca 2ϩ regulation of the chaperone activity (8,16,17). The present studies have been carried out only in HeLa cells, which, however, are known to express, in their ER, levels of both BiP and CRT similar to those present in the ER of various other lines (7,34,42). We conclude therefore that the two proteins together probably account for the bulk of the rapidly exchanging Ca 2ϩ storage also in those other cells. The latter function, which is already known to play key roles in cell physiology, is shown here to be modulable, due to the changes of expression of the chaperones taking place during exposure to stressful stimuli.