Selective Effects of Calcium Chelators on Anterograde and Retrograde Protein Transport in the Cell*

Calcium plays a regulatory role in several aspects of protein trafficking in the cell. Both vesicle fusion and vesicle formation can be inhibited by the addition of calcium chelators. Because the effects of calcium chelators have been studied predominantly in cell-free systems, it is not clear exactly which transport steps in the secretory pathway are sensitive to calcium levels. In this regard, we have studied the effects of calcium chelators on both anterograde and retrograde protein transport in whole cells. Using both cytochemical and biochemical analyses, we find that the anterograde-directed exit of vesicular stomatitis virus G protein and the retrograde-directed exit of Shiga toxin from the Golgi apparatus are both inhibited by calcium chelation. The exit of vesicular stomatitis virus G from a pre-Golgi compartment and the exit of Shiga toxin from an endosomal compartment are sensitive to the membrane-permeant calcium chelator 1,2-bis(2-amino phenoxy)ethane-N,N,N′,N′-tetraacetic acid–tetrakis (acetoxymethyl ester) (BAPTA-AM). By contrast, endoplasmic reticulum exit and endocytic internalization from the plasma membrane are not affected by BAPTA. Together, our data show that some, but not all, trafficking steps in the cell may be regulated by calcium. These studies provide a framework for a more detailed analysis of the role of calcium as a regulatory agent during protein transport.

The modulation of cytosolic calcium levels serves as an important signaling system for cell regulation (1). Protein transport within the secretory pathway is among the processes regulated by calcium. The best characterized role for calcium in vesicular transport is in the calcium-triggered fusion of synaptic and secretory vesicles at the plasma membrane (2)(3)(4). Calcium-regulated vesicle fusion and fusion reactions in the constitutive secretory pathway have similar molecular mechanisms. For example, both types of fusion reactions are mediated or regulated by SNARE 1 proteins, as well as rab-and sec1/munc18-related proteins (5). Recent studies indicate that a regulatory role for calcium may also extend to fusion reactions in the constitutive secretory pathway.
Both major organelles of the early secretory pathway, the endoplasmic reticulum and the Golgi apparatus, are known to be cellular calcium stores (6,7). Secretory vesicles have also been shown recently to act as a dynamic calcium store (8). Two classes of ATP-dependent pump proteins maintain high lumenal calcium concentrations in these organelles. The sarco/endoplasmic reticulum calcium ATPase (SERCA)-type pumps are found both on the ER and the Golgi (9 -11). Pmr1p was identified as a Golgi calcium pump in yeast (12), and mutations in the mammalian isoform lead to Hailey-Hailey disease (13). Studies analyzing the subcellular localization of calcium indicate that calcium concentrations are particularly high in the Golgi region of cells and that calcium gradients could exist among organelles of the secretory pathway (6,14). Both the ER and the Golgi apparatus have also been shown to have inositol 1,4,5-trisphosphate receptors for the triggered release of calcium during cell signaling (1,15). Specific calcium binding or sequestering proteins have been localized to each organelle. The calcium-binding chaperone proteins calreticulin and calnexin function in the ER (16,17). At least three lumenal calcium-binding proteins, Cab45, CALNUC (nucleobindin), and p54/NEFA have been localized to the Golgi (11,18,19). The presence of calcium stores and specific calcium-binding proteins in secretory organelles has revived the notion that calcium or calcium gradients are important for regulating the constitutive secretory pathway (20).
Studies on ER to Golgi transport in yeast and in semi-intact (perforated) cells have indeed indicated a role for calcium in this trafficking step (21)(22)(23)(24). Regulation by calcium has also been implicated in homotypic vacuolar fusion (25), in late endosome-lysosome heterotypic fusion and the reformation of lysosomes from hybrid organelles (26), and in fusion between endosomal compartments (27). There are several lines of evidence showing that calcium is important for the function of the Golgi apparatus. Most directly, calcium chelators have been shown to block intra-Golgi protein transport in vitro (28). The brefeldin A-induced retrograde transport from the Golgi apparatus to the endoplasmic reticulum is also affected by calcium chelation (29). Calmodulin has been shown to mediate the effects of calcium on several of these fusion or transport reactions (25,28,30). Interestingly, the calcium chelator BAPTA is often found to be more effective than EGTA at inhibiting these membrane fusion reactions (25)(26)(27). BAPTA has a much faster on rate for calcium binding than does EGTA. This observation is frequently interpreted to indicate that calcium transients or gradients rather than steady-state calcium levels regulate these processes.
We have recently reported that in addition to regulating membrane fusion reactions, calcium may play a role in regulating the coating or uncoating of transport vesicles. We found that treating cells with the membrane-permeable calcium chelator BAPTA-AM led to the dissociation of the COPI vesicle coat protein, coatomer, from the Golgi apparatus (31). Using cell-free Golgi binding assays, we also observed that the addition of BAPTA led to the rapid dissociation of coatomer from previously coated Golgi membranes. As with fusion, BAPTA was more effective than EGTA at mediating this effect. The addition of calcium chelators appears to affect a step in vesicle coat assembly after the recruitment of ADP-ribosylation factor, the small GTP-binding protein that triggers coat assembly.
The previous studies strongly implicate a role for calcium in regulating transport through the constitutive exocytic and endocytic protein transport pathways. These studies have utilized cell-free systems, and thus it is still not clear to what extent calcium modulates transport through the secretory system in whole cells. As a step toward clarifying the role of calcium in transport, we have analyzed the effects of membrane-permeant calcium chelators on both anterograde and retrograde protein transport through the secretory pathway. This study indicates that select transport steps are sensitive to calcium transients or gradients.

EXPERIMENTAL PROCEDURES
Materials-Tissue culture media were obtained through the Diabetes and Endocrinology Research Center, University of Iowa. The construct pT77-SLT-B-Glyc-KDEL was obtained as gifts from Drs. B. Goud and A. Girod. BAPTA-AM and EGTA-AM were purchased from Molecular Probes (Eugene, OR). The following antibodies were used in this study: anti-rat mannosidase II (Berkeley Antibody Co., Richmond, CA), polyclonal anti-VSV G (Medical & Biological Laboratories Co., Naka-Ku Nagoya, Japan), P5D4 monoclonal anti-VSV G (32), anti-BiP (Affinity Bioreagents Inc., Golden, CO), and anti-biotin (Sigma).
VSV G Protein Transport Assay-The cell line Gts-NRK (a gift from Dr. V. Malhotra) that stably expresses the temperature-sensitive ts045-VSV G was maintained by growth at a permissive temperature (32°C) in ␣-minimum essential medium plus 5% fetal calf serum. The cells were incubated for 14 -16 h at a restrictive temperature (39.5°C) to accumulate VSV G protein in the ER and processed for immunofluorescence. For VSV G protein transport assay, the cells were incubated at 39.5°C for 14 -16 h, and then VSV G protein was released from the temperature block by switching the cells to 32°C medium containing 10 g/ml cycloheximide. The cells were incubated for 15 or 90 min at (32°C) and processed for immunofluorescence. BAPTA-AM or EGTA-AM was added at the indicated times by replacing the medium with medium containing 50 M BAPTA-AM or 50 M EGTA-AM plus 10 g/ml cycloheximide. Where indicated, 2 M A23187 (Sigma), 3 mM EGTA, or 5 mM CaCl 2 were also added to the medium.
Immunofluorescence-Gts-NRK cells were plated onto glass cover slips, and after 24 h the cells were washed with phosphate-buffered saline (PBS), fixed with 4% paraformaldehyde, and permeabilized using 0.1% Triton X-100 for 4 min at room temperature. The formaldehyde was quenched with 50 mM ammonium chloride for 10 min at room temperature. The cells were washed three times with PBS and blocked with 2.5% donkey serum in PBS for 1 h at room temperature. Appropriate dilutions of the anti VSV G, anti-mannosidase II, and anti-BiP antibodies in the blocking buffer were added to the cells for 1 h at room temperature. The cells were washed three times with PBS and incubated with fluorescein isothiocyanate-conjugated anti-rabbit and Texas Red-conjugated anti-mouse secondary antibodies in blocking buffer for 1 h at room temperature. The cells were washed three times, mounted on slides, and analyzed on a confocal microscope (Bio-Rad).
Electron Microscopy-Gts-NRK cells were incubated with or without 50 M BAPTA-AM in ␣-minimum essential medium for 90 min at 39.5°C. The cells were fixed in 2.5% glutaraldehyde, post-fixed in 1% OsO 4 , dehydrated, and embedded in Spurr's embedding medium. Ultrathin sections stained with uranyl acetate and lead citrate were examined with an H-7000 transmission electron microscope.
The recombinant Shiga toxin B-fragment was overexpressed in the BL21(DE3)pLysS bacterial strain (Stratagene) and purified using ion exchange chromatography as described (33) except that a High Q col-umn (Bio-Rad) was used in place of Mono Q column. The STB was labeled using activated Cy2 for fluorescence microscopy (Amersham Biosciences; HP7 9NA) or activated biotin for Western blot analysis according to the manufacturer's instructions. The cells were incubated for 30 min in the presence of 2 g/ml labeled STB. The cells were then washed three times with fresh medium and incubated for various times as described in the figure legends.
Western Blotting and Glycosylation Analysis-The proteins were fractionated using SDS-PAGE and blotted onto polyvinylidene difluoride membranes using standard protocol for the Bio-Rad minigel and blotting apparatuses. Following the transfer, the membranes were incubated with appropriate dilutions of the indicated primary antibodies. The signal was visualized using horseradish peroxidase-conjugated secondary antibodies (Bio-Rad) and ECL (Amersham Biosciences).
The treatment of VSV G protein with endoglycosidase H (Calbiochem) was done according to the manufacturer's instructions. For STB glycosylation analysis, the toxin was labeled with biotin (Pierce) using appropriate molar ratio of biotin to protein. STB-biotin was detected by Western blotting using anti-biotin antibodies. In some experiments, the cells were incubated in the presence of 1 g/ml tunicamycin 1 h before the addition of STB.

BAPTA Blocks Transport from an ER/Golgi Intermediate Compartment to the Golgi but Does Not Affect Exit from the ER-
We used a NRK cell line constitutively expressing a temperature-sensitive VSV G (ts045) to analyze the effects of calcium chelation on anterograde protein transport through the secretory pathway. VSV G (ts045) is trapped in the ER at the restrictive temperature (39.5°C) but is transported normally through the Golgi apparatus to the plasma membrane at the permissive temperature (32°C). The transport of VSV G was monitored using both fluorescence microscopy ( Fig. 1) and by determining the glycosylation state of the VSV G (Fig. 2). Following a 16-h incubation at the restrictive temperature, the VSV G is localized to a dispersed ER compartment (Fig. 1A). When cells were switched to a permissive temperature (32°C) for 15 min, the VSV G was relocated to the juxtanuclear Golgi compartment (Fig. 1B). After 90 min at the permissive temperature, VSV G was found dispersed in small punctate and tubular structures and at the cell surface (Fig. 1C). In some cells, a small amount of residual Golgi-like staining was also observed. ER-localized, core-glycosylated VSV G protein is sensitive to deglycosylation by endoglycosidase H (endo H), whereas complex oligosaccharides generated by processing in the Golgi are endo H-resistant. Endo H digestion of cell extracts following the incubation at the restrictive temperature confirms that the VSV G from these cells is completely endo H-sensitive and thus has not been processed by the Golgi (Fig. 2A, lanes 1 and  2). Following a 90-min incubation at the permissive temperature, all of the VSV G in cells is endo H-resistant, indicating that it has been transported through the Golgi ( Fig. 2A, lanes 3  and 4).
To determine what effect calcium chelation had on VSV G transport, we treated cells with the membrane-permeant calcium chelator BAPTA-AM at various times relative to the switch from the restrictive to the permissive temperature. When BAPTA-AM was added to the cells either 90 min (not shown) or 30 min (Fig. 1D) prior to the switch from the restrictive to the permissive temperature, we observed a marked difference in the localization of VSV G. Instead of localizing to the dispersed post-Golgi structures and the plasma membrane, VSV G was now found to localize to large, brightly labeled perinuclear punctate structures (Fig. 1, compare C and D).
We examined whether the VSVG-containing punctate structures observed upon BAPTA-AM treatment were a pre-or post-Golgi compartment by testing whether the VSV G was endo H-resistant under this condition. Fig. 2A shows that when BAPTA-AM was added at least 30 min prior to switching the cells to the permissive temperature, the majority of the VSV G was endo H-sensitive and thus had not reached the Golgi (lanes 5-8). Fig. 2B shows that the block in ER to Golgi transport by BAPTA-AM is reversed when intracellular calcium levels are restored by the addition of the calcium ionophore A23187 and CaCl 2 to the medium. Together, the fluorescence microscopy and biochemical analysis demonstrate that when BAPTA-AM is added prior to the switch to the permissive temperature, VSV G is blocked in a pre-Golgi compartment.
The punctate perinuclear structures could result from an alteration in ER morphology caused by the presence of BAPTA. Alternatively, in the presence of BAPTA, the VSV G might exit the ER and then become trapped in a punctate ER/Golgi intermediate compartment. We examined the morphology of both the ER and the Golgi following BAPTA treatment to differentiate between these possibilities. As shown in Fig. 3, immunofluorescence using anti-mannosidase II antibodies to label the Golgi (C and D) and anti-Bip antibodies to label the ER (A and B) revealed that BAPTA treatment had little or no effect on the morphology of either of these organelles. In particular, no punctate Bip-positive ER elements analogous to the VSV G-containing structures shown in Fig. 1D were observed. As expected from our previous studies (31), coatomer was found to be dissociated from the Golgi after treatment with BAPTA-AM (not shown). Ultrastructural characterization also reveals that the morphology of the Golgi apparatus is not overtly affected in cells following treatment with the calcium chelator (Fig. 3, E and F). The electron micrographs show that the Golgi is of similar size and retains a stacked structure after treatment with BAPTA for 90 min. Together, our results reveal that addition of BAPTA-AM and thus a decrease in cytosolic free calcium does not block the exit of VSV G from the ER. However, it prevents VSV G transport to the Golgi apparatus and causes the accumulation of VSV G in a punctate perinuclear ER/Golgi intermediate compartment.
The Exit of VSV G from the Golgi Is Blocked by BAPTA-The addition of BAPTA-AM to the cells simultaneously with the switch to the permissive temperature caused VSV G to accumulate in a juxtanuclear structure reminiscent of the Golgi apparatus (Fig. 1E). Colocalization studies revealed that indeed, under this condition, VSV G localization overlaps almost completely with the Golgi marker mannosidase II (Fig. 4). Fig.  2A shows that when BAPTA-AM is added at the same time as the temperature switch, all of the VSV G becomes endo Hresistant (lanes 9 and 10), confirming that it has now reached the Golgi. Together, the data indicate two distinct blocks in anterograde protein transport of VSV G in the presence of BAPTA-AM: one block in an ER/Golgi intermediate compartment and a second block either within or from the Golgi apparatus.
We were not able to detect any additional effects of calcium chelation on VSV G transport in post-Golgi transport steps by adding BAPTA-AM after the temperature switch. The addition of BAPTA-AM 30 min after the temperature switch caused a small amount of VSV G to accumulate in the Golgi of some cells, but there was no obvious accumulation of VSV G in any additional structure that might represent a transport intermediate between the Golgi and the plasma membrane (Fig. 1, compare C and F). This finding is consistent with a previous study showing that there are no defects in protein transport when a calcium ionophore and EGTA are added to cells after a 1-h pulse-chase (34).
Both calcium-sensitive membrane fusion reactions and COPI coat assembly share the property that they are more sensitive to BAPTA than to EGTA (25)(26)(27)31). Because BAPTA has a much faster on-rate for calcium binding, this difference has been interpreted to indicate that a calcium transient or gradient may be involved. If the effects of BAPTA-AM in the whole cell analysis reflect the effects of BAPTA in cell-free fusion and budding assays, we predicted that whole cell trafficking might also be more sensitive to BAPTA than to EGTA. Thus, we compared the effects of EGTA-AM and BAPTA-AM on VSV G transport. Fig. 5 shows that there was no obvious accumulation of VSV G in the Golgi or in a pre-Golgi compartment upon treatment with EGTA-AM. Furthermore, biochemical analysis revealed that VSV G became endo H-resistant regardless of when EGTA-AM was added relative to the switch to the permissive temperature (Fig. 6). Although 50 M BAPTA-AM was a more effective inhibitor than 50 M EGTA-AM, 3 mM EGTA, which is not membrane permeant, inhibited VSVG transport to the Golgi when added together with a calcium ionophore (Fig.  2B, lanes 5 and 6). These findings support the model that BAPTA-AM inhibits protein transport by disrupting intracellular calcium levels (see "Discussion").
Distinct Steps in Retrograde Protein Transport Are Sensitive to BAPTA-Although the majority of proteins move through the secretory pathway in an anterograde direction, retrograde protein transport also plays an important role in the function of the secretory pathway. Retrograde transport is necessary for the retrieval of escaped resident organelle proteins, for the recycling of proteins that function in transport, and for quality control and protein degradation. Some bacterial toxins such as cholera toxin and Shiga toxin have proven to be good tools for studying retrograde protein transport through secretory organelles (35)(36)(37)(38). These toxins are endocytosed into cells and are then transported from endosomes through the Golgi apparatus to the ER (35,39,40). The toxins are transported from the lumen of the ER into the cytosol where they then exert their toxic effect (40).
We have used the Shiga-like toxin B subunit to analyze the effects of BAPTA-AM on retrograde protein transport. For these experiments, recombinant STB was purified and co- valently labeled with Cy2 (see "Experimental Procedures"). Vero cells were incubated with the labeled toxin for 30 min to allow the toxin to internalize. The noninternalized toxin was then washed away with fresh medium, and the cells were allowed to grow for an additional 0.5-16 h. The location of STB in the cell was monitored by fluorescence microscopy (Fig. 7). In addition, we used a recombinant form of Shiga toxin that was engineered to contain an N-glycosylation site (33,41) to detect the arrival of the STB in the ER biochemically (Fig. 8). Fluorescence microscopy shows that the internalized STB first appeared in punctate endosomal compartments and then within 2 h became concentrated in the Golgi (Fig. 7). Diffuse ER-like staining was also observed by microscopy after 2-h incubations.
As a more precise indicator of STB arrival in the ER, we examined its glycosylation state. Consistent with previous studies (33), we observed three distinct STB bands on Western blots following retrograde transport of STB through the secretory pathway (Fig. 8). We confirmed that the uppermost band corresponds to the core-glycosylated form of STB by treating cells with tunicamycin to block core glycosylation (Fig. 8A,  lanes 6 -10). Fig. 8A shows that STB starts to arrive in the ER at about 2 h after the addition of the toxin (lane 2). We observed a peak of STB glycosylation at about 4 h after addition of the toxin to the cells (Fig. 8A, lane 3).
We determined the effects of calcium chelation on the retrograde transport of STB by adding BAPTA-AM to the cells at different times relative to the addition of the STB. We found that when BAPTA-AM was added at the same time or prior to the addition of STB, the toxin was internalized but appeared to accumulate in punctate endosomal compartments (Fig. 7). Although the Golgi had normal morphology under these conditions (Fig. 3), we never observed colocalization of STB with the Golgi when BAPTA-AM was added early (not shown). Determination of the glycosylation state of STB confirmed that it remained unglycosylated and thus did not reach the ER when BAPTA-AM was added to the cells at the same time as STB (Fig. 8B, lane 2). The results indicate that although internalization of STB into an endosomal compartment is not affected by calcium chelation, its transport from an endosomal compartment to the Golgi is blocked by the presence of the calcium chelator. Previous studies indicate that STB is transported directly from early endosomes to the Golgi, bypassing the late endosomes (42). Our results indicate a requirement for calcium in this direct trafficking step.
We found that addition of BAPTA-AM up to 60 min after the addition of STB completely blocked the transport of STB to the ER as revealed by its failure to be glycosylated (Fig. 8B, lanes  2, 4, and 6). When BAPTA-AM was added 90 or 120 min after STB, glycosylated STB was observed, indicating that at least some STB had moved beyond a BAPTA-sensitive transport step and arrived in the ER (Fig. 8B, lanes 8 and 10). Interestingly, by comparing the localization of STB with its glycosylation state, we found that although STB is not glycosylated when BAPTA is added 60 min after the addition of STB to the cells (Fig. 8B, lane 6), some of the STB is localized to a juxtanuclear Golgi compartment (Fig. 7). We interpret this to indicate that some of the STB has arrived at the Golgi by 60 min and that BAPTA has inhibited a Golgi to ER retrograde transport step. Thus, as was the case for anterograde transport, it appears that some retrograde transport steps, such as the internalization (endocytosis) of STB, are insensitive to calcium chelation, whereas other steps, such as endosome to Golgi apparatus and Golgi to ER transport, are inhibited in the presence of the calcium chelator BAPTA. DISCUSSION Previous studies have shown that both membrane fusion reactions and vesicle assembly reactions are inhibited by the calcium chelator BAPTA. Because these studies have relied on the cell-free reconstitution of these processes, we have set out to determine whether steps in constitutive anterograde and retrograde protein transport are affected by calcium chelators in whole cells. The use of the membrane permeant analogs, BAPTA-AM and EGTA-AM, is particularly effective for these studies because the analogs do not bind calcium until they have entered the cell and the acetoxymethyl ester moiety has been hydrolyzed. Because normal extracellular calcium levels are maintained, the cells remain attached to their substrate, allowing relatively normal cell morphology to be maintained. This has allowed the combined immunohistochemical and biochemical analysis we report in this paper.
This combined approach has revealed that at least four resolvable steps are inhibited by treating cells with BAPTA-AM. The BAPTA-sensitive steps during anterograde transport of VSV G through the secretory pathway include intermediate compartment to Golgi transport and exit from the Golgi apparatus. During retrograde transport through the secretory pathway, we find that transport of STB from the endosome to the Golgi is inhibited as well as transport from the Golgi apparatus to the ER. At least two transport steps were found to be resistant to BAPTA-AM. These include endocytosis of STB from the plasma membrane and the exit of VSV G from the ER.
A surprising aspect of this and many other studies examining the effects of calcium chelators on reactions involved in protein transport is that BAPTA acts more effectively as an inhibitor than EGTA. One possibility is that the fast initial binding rate of BAPTA is able to buffer calcium transients that are not buffered by EGTA. A second possibility is that under biological conditions (or within certain biological microenvironments), the calcium-binding constant for BAPTA is in fact smaller than for EGTA. A third explanation is that BAPTA acts in some manner other than by chelating calcium. For example, BAPTA has been shown to act as an inhibitor of phospholipase D (43).
Our results, in conjunction with other studies, are most consistent with BAPTA disrupting protein transport by acting as a calcium chelator. First, for the cell-free studies (28,31) and for ER to Golgi transport in whole cells (Fig. 2), the addition of calcium reverses the effects of BAPTA. Thus, if BAPTA is acting directly on a protein, it is only the non-calcium-bound form that is able to do so. Second, we find that disrupting intracellular calcium levels in a BAPTA-independent manner with an ionophore and extracellular EGTA also disrupts ER to Golgi transport (Fig. 2B). Similarly, in cell-free studies on membrane fusion and Golgi transport where BAPTA is more effective than EGTA, calmodulin has been shown to be involved (25,28,30). This provides independent evidence of a role for calcium in reactions disrupted specifically by BAPTA. Finally, we showed previously that although BAPTA is more effective than EGTA at uncoating COPI vesicles, EGTA did have an effect when added at a higher concentration (31).
Our results indicate that several transport steps in the cell are affected by chelating calcium with BAPTA. Because both vesicle fusion and vesicle formation are potentially regulated by calcium, an important remaining question is which of the transport blocks result from an effect on membrane fusion and which ones result from an effect on vesicle formation. Regulating vesicle formation or vesicle fusion independently could lead to an accumulation or a depletion of vesicles. For example, increasing the rate of fusion without a concomitant increase in the rate of vesicle formation would lead to vesicle depletion. An interesting possibility is that both processes are regulated simultaneously at calcium-sensitive transport steps. By coordinating the rate of vesicle fusion and the rate of vesicle formation and uncoating through similar calcium-mediated regulatory mechanisms, the problem of vesicle accumulation or depletion might be avoided. Future studies on the role of calcium on fusion reactions and in vesicle coating and uncoating reactions in the secretory pathway should provide additional insight into this regulation.