Ca 2 (cid:1) Uptake and Release Properties of a Thapsigargin-insensitive Nonmitochondrial Ca 2 (cid:1) Store in A7r5 and 16HBE14o (cid:2) Cells*

In a previous study we overexpressed the thapsigargin (tg)-insensitive Pmr1 Ca 2 (cid:1) pump of the Golgi apparatus of Caenorhabditis elegans in COS-1 cells and studied the properties of the Ca 2 (cid:1) store into which it was integrated. Here we assessed the properties of an endogenous tg-insensitive nonmitochondrial Ca 2 (cid:1) store in A7r5 and 16HBE14o (cid:2) cells, which express a mammalian homologue of Pmr1. The tg-insensitive Ca 2 (cid:1) store was considerably less leaky for Ca 2 (cid:1) than the sarco(endo)-plasmic-reticulum Ca 2 (cid:1) -ATPase (SERCA)-containing Ca 2 (cid:1) store. Moreover like for the worm Pmr1 Ca 2 (cid:1) pump expressed in COS-1 cells, Ca 2 (cid:1) accumulation into the endogenous tg-insensitive store showed a 2 orders of magnitude lower sensitivity to cyclopiazonic

Many cells use inositol 1,4,5-trisphosphate (IP 3 ) 1 as second messenger to generate intracellular Ca 2ϩ signals (1). IP 3 binds to the IP 3 receptor, a Ca 2ϩ channel found in the endoplasmic reticulum (ER). Ca 2ϩ uptake into the ER, which represents the major intracellular Ca 2ϩ store in most cell types, is mediated by Ca 2ϩ transporting ATPases of the sarco(endo)plasmic-reticulum Ca 2ϩ -ATPase (SERCA) family (2). Three genes, whose transcripts are alternatively spliced, give rise to a number of different SERCA proteins that all share an equal sensitivity to inhibition by thapsigargin (tg).
More recently it has become clear from studies in intact cells, permeabilized cells, or vesicle preparations that Ca 2ϩ stored within the Golgi apparatus may also be released in response to IP 3 -producing agonists (3)(4)(5)(6)(7). The functional properties of this Golgi Ca 2ϩ store and its contribution in generating intracellular Ca 2ϩ signals remain, however, largely unknown, mainly because the Ca 2ϩ uptake system of the Golgi complex is less well understood and partly because the volume of the Golgi apparatus appears to be considerably smaller than that of the ER. Ca 2ϩ accumulation by rat liver vesicles enriched in Golgi membranes (8) and in HeLa cells overexpressing the Golgiresident Ca 2ϩ -binding protein Calnuc/nucleobindin (9) was found to be inhibited by tg, suggesting that also in the Golgi compartment SERCAs mediate the sequestration of Ca 2ϩ . In contrast, about half of the Ca 2ϩ uptake in an isolated stacked Golgi fraction obtained from rat liver was reported to be insensitive to tg (10). Likewise, half of the Ca 2ϩ taken up by the Golgi compartment in intact HeLa cells was found to be resistant to tg (4). The abolition of this tg-insensitive accumulation by orthovanadate (4) strongly suggests the involvement of a classical P-type Ca 2ϩ -ATPase, characterized by the formation of a phosphorylated intermediate. The presence of a Ca 2ϩ pump belonging to the Pmr1 family of Ca 2ϩ transport ATPases differing from the SERCA Ca 2ϩ pumps by their insensitivity to tg has been reported in the Golgi apparatus (7,(11)(12)(13).
Recently we expressed the Caenorhabditis elegans Pmr1 Ca 2ϩ pump in COS-1 cells and studied the functional properties of the Ca 2ϩ store to which it was targeted (7,14). The most important results of these studies were that Pmr1 largely colocalized with the Golgi apparatus and that the large tg-insensitive Ca 2ϩ store observed in those cells was 33% responsive to IP 3 , albeit with a 3 times lower sensitivity than the tg-sensitive Ca 2ϩ store (14). The heterologous expression of a Pmr1 Ca 2ϩ pump from a different species, however, did not allow us to draw conclusions about the significance of endogenous Ca 2ϩ signaling by the Golgi compartment in normal cells. We therefore screened for the presence of a Ca 2ϩ store with similar properties as the Pmr1-induced Ca 2ϩ store in nontransfected cells. Here we report on the properties of such a store in rat aortic A7r5 smooth muscle cells and in 16HBE14oϪ human bronchial mucosal cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-A7r5, COS-1, and 16HBE14oϪ cells were cultured as described previously (7,15,16). For 45 Ca 2ϩ fluxes cells were seeded in 12-well dishes (4 cm 2 ; Costar, Cambridge, MA) at a density of ϳ10 4 cells cm Ϫ2 , and for Ca 2ϩ imaging experiments cells were seeded in Coverglass Chambers (Nunc Inc., Naperville, IL) at a density of 5 ϫ 10 4 cells cm Ϫ2 . Four days after plating, COS-1 cells were tran-* This research was financed by the Interuniversity Poles of Attraction Program-Belgian State, Prime Minister's Office-Federal Office for Scientific, Technical and Cultural Affairs Grant IUAP P4/23. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

45
Ca 2ϩ Uptake-The cells were permeabilized by treating them for 10 min with 20 g ml Ϫ1 saponin at 25°C in a medium containing 120 mM KCl, 30 mM imidazole-HCl (pH 6.8), 2 mM MgCl 2 , 1 mM ATP, and 1 mM EGTA. The nonmitochondrial Ca 2ϩ stores were loaded for 45 min in loading medium containing 120 mM KCl, 30 mM imidazole-HCl (pH 6.8), 5 mM MgCl 2 , 5 mM ATP, 0.44 mM EGTA, 10 mM NaN 3 , 10 M oligomycin, 10 M antimycin A, and 150 nM free Ca 2ϩ (23 Ci ml Ϫ1 ). Tg (10 M) was added to the loading medium if inhibition of the SERCA Ca 2ϩ pumps was needed. Efflux was performed in a medium containing 120 mM KCl, 30 mM imidazole-HCl (pH 6.8), and 1 mM EGTA. Free concentrations of Ca 2ϩ were calculated by the Cabuf program (ftp://ftp.cc. kuleuven.ac.be/pub/droogmans/cabuf.zip) and based on the stability constants given by Fabiato and Fabiato (17). At the end of the experiment the 45 Ca 2ϩ remaining in the stores was released by incubation with 1 ml of a 2% (w/v) sodium dodecyl sulfate solution for 30 min.
RT-PCR Analysis-The expression of Pmr1 in A7r5 and 16HBE14oϪ cells was verified at the mRNA level by RT-PCR on total RNA. 1 g of total RNA was reverse transcribed by Moloney murine leukemia virus reverse transcriptase. 1 ⁄20 of this mixture was used for a 26-cycle PCR. The forward and reverse primers for the rat sequence were AAACTG-GAACCCTGACGAAG (nucleotides 646 -665 in GenBank TM accession no. M93018) and TTGGCTTTCCCATCAGAGTG (nucleotides 851-870), respectively. The annealing temperature was 53°C. The forward and reverse primers for the human sequence were GGTGTGAAAGAAGCT-GTTACAAC (nucleotides 1805-1827 in GenBank TM accession no. AF189723) and GTAAAATACTGCAACCTTTGG (nucleotides 1988 -2008), respectively. The annealing temperature was 60°C. In both cases, the forward and the reverse primers were located in different exons.

RESULTS AND DISCUSSION
Demonstration of an Endogenous Tg-insensitive Nonmitochondrial Ca 2ϩ Store in A7r5 and 16HBE14oϪ Cells-All members of the SERCA family of Ca 2ϩ pumps are irreversibly inhibited by tg with similar affinity (2). We loaded the Ca 2ϩ stores of permeabilized A7r5 (Fig. 1A) and 16HBE14oϪ cells ( 1A). The values were 89 and 11%, respectively, for the 16HBE14oϪ cells (Fig. 1B). Mitochondrial Ca 2ϩ uptake was prevented by the presence of 10 mM NaN 3 , 10 M oligomycin, and 10 M antimycin A and could therefore not account for the tg-resistant fraction.
Passive Ca 2ϩ Leak from the Tg-insensitive Nonmitochondrial Ca 2ϩ Store in A7r5 and 16HBE14oϪ Cells-After loading the stores to steady state, their passive permeability to Ca 2ϩ was assessed by switching to an efflux medium containing 2 mM EGTA with no added Ca 2ϩ or ATP. The Ca 2ϩ efflux that occurred under these conditions can be considered as unidirectional since the calculated free [Ca 2ϩ ] in the efflux medium (Ͻ10 nM) was below the threshold to stimulate the Ca 2ϩ pumps, and no ATP was present to fuel the pumps. Fig. 2 shows for A7r5 ( Fig. 2A) and 16HBE14oϪ cells (Fig. 2B) the decrease in Ca 2ϩ content as a function of time for both the tg-insensitive compartment (closed circles, full line) and the SERCA-containing Ca 2ϩ store (open circles, dotted line). It is clear that the rates of Ca 2ϩ loss from the tg-insensitive compartment in A7r5 cells and especially in 16HBE14oϪ cells were significantly smaller than those of the SERCA-containing Ca 2ϩ store. These differences were not a consequence of the different initial Ca 2ϩ content of the two stores since the initial level of store loading has no effect on the passive Ca 2ϩ leak (18). The data in Fig. 2   FIG. 3. RT-PCR of Pmr1 in A7r5 and 16HBE14o؊ cells. Gel electrophoresis of RT-PCR products using primers specific for the rat (A7r5 cells) and the human Pmr1 sequence (16HBE14oϪ cells (HBE)). Single bands of the predicted length were amplified. The gel was stained with Vistra Green. M, molecular marker.  Fig. 1). Ca 2ϩ uptake by the tg-insensitive Ca 2ϩ store (q, full line) in A7r5 and 16HBE14oϪ cells was measured in a medium containing 10 M tg as the difference in Ca 2ϩ uptake in the presence and absence of 10 M A23187 (arrows a in Fig. 1). Ca 2ϩ uptake by Pmr1 in COS-1 cells (q, full line) was taken as the difference in Ca 2ϩ uptake between Pmr1-overexpressing and control COS-1 cells in a medium containing 10 M tg. Cyclopiazonic acid was dissolved in dimethyl sulfoxide, the concentration of which was constant in all experiments (1%). are in agreement with our earlier report that also the Pmr1induced Ca 2ϩ store in COS-1 cells was less leaky as compared with the ER in these cells (see Fig. 2B in Ref. 14).

FIG. 5. Effect of 2,5-di-(tert-butyl)-1,4-benzohydroquinone on Ca 2؉ uptake by the tg-sensitive and tg-insensitive Ca 2؉ stores in permeabilized A7r5 and 16HBE14o؊ cells and on Ca
A7r5 and 16HBE14oϪ Cells Express Pmr1-Pmr1 is a tginsensitive Ca 2ϩ pump present in the Golgi apparatus (7,13), making it a likely candidate for the Ca 2ϩ uptake mechanism in the presence of tg. Fig. 3 shows that Pmr1 could indeed be demonstrated in A7r5 and 16HBE14oϪ cells at the mRNA level. Another argument for the presence of Pmr1 is that the pharmacology of the Ca 2ϩ uptake mechanism of the tg-insensitive Ca 2ϩ store in A7r5 and 16HBE14oϪ cells and that of the overexpressed Pmr1 Ca 2ϩ pump in COS-1 cells were similar as discussed in the following paragraphs.
Ca 2ϩ uptake by the tg-insensitive nonmitochondrial Ca 2ϩ store was inhibited by the mycotoxin cyclopiazonic acid with an IC 50 of 165 M in A7r5 cells (Fig. 4A, closed symbols and full line) and 337 M in 16HBE14oϪ cells (Fig. 4B, closed symbols  and full line). These values were 2 orders of magnitude higher than the IC 50 value found to inhibit SERCA-mediated Ca 2ϩ uptake in A7r5 cells (1. Fig. 4B). The inhibition curves for the tg-insensitive Ca 2ϩ uptake were also steeper than those of SERCA. The IC 50 values for SERCA inhibition were higher than the previously reported value of 10 -20 nM (19), probably as a consequence of the presence of 5 mM ATP in the uptake medium since ATP protects the enzyme in a competitive manner against inhibition by cyclopiazonic acid (19). Fig. 4C shows how cyclopiazonic acid affected exogenous Pmr1 in COS-1 cells overexpressing this Ca 2ϩ pump (7,14). In this assay system, Pmr1-induced Ca 2ϩ pump-ing could be specifically measured as the difference in Ca 2ϩ uptake between Pmr1-overexpressing COS-1 cells and control cells in a medium containing 10 M tg. The closed circles and full line in Fig. 4C illustrate that cyclopiazonic acid inhibited Pmr1 with an IC 50 of 294 M, while SERCA in these cells was half-maximally inhibited at 0.7 M (Fig. 4C, open circles and dotted line). Pmr1 was therefore 2 orders of magnitude less sensitive to cyclopiazonic acid than was SERCA. The inhibition curve for Pmr1 was also steeper than that of SERCA. It is evident that the values obtained for the exogenous Pmr1 Ca 2ϩ pump in COS-1 cells are in excellent agreement with the values found for the endogenous tg-insensitive Ca 2ϩ pump in A7r5 and 16HBE14oϪ cells.
2,5-Di-(tert-butyl)-1,4-benzohydroquinone, another inhibitor of the SERCA Ca 2ϩ pumps (20), was only a very weak inhibitor of the tg-insensitive nonmitochondrial Ca 2ϩ store since even very high concentrations (1 mM) only induced a partial inhibition in A7r5 cells (closed circles and full line in Fig. 5A), in 16HBE14oϪ cells (closed circles and full line in Fig. 5B) A7r5 cells were loaded to steady state with Ca 2ϩ in the presence of 10 M tg and a mixture of mitochondrial inhibitors and then incubated in efflux medium and stimulated with 100 M IP 3 (Fig. 6A, circles and full line) or 10 M A23187 as Ca 2ϩ ionophore (Fig. 6A, squares and dashed line). IP 3 induced a partial Ca 2ϩ release from this compartment. Inositol 1,3,4,5tetrakisphosphate (100 M), cyclic ADP-ribose (100 M), and caffeine (20 mM) all failed to release Ca 2ϩ under these conditions (data not shown). Nicotinic acid adenine dinucleotide phosphate (100 M), which has been reported to release Ca 2ϩ from a tg-insensitive nonmitochondrial Ca 2ϩ store in other cell types (21) and which was also reported to release Ca 2ϩ in A7r5 cells (22), also was unable to release Ca 2ϩ under these conditions (data not shown). In 16HBE14oϪ cells, no significant release was observed upon addition of 100 M IP 3 (Fig. 6B, circles and full line) or 100 M inositol 1,3,4,5-tetrakisphosphate, 100 M cyclic ADP-ribose, 20 mM caffeine, and 100 M nicotinic acid adenine dinucleotide phosphate (data not shown).
To compare the properties of the IP 3 receptors in the tginsensitive Ca 2ϩ store with those in the SERCA-containing Ca 2ϩ store in A7r5 cells, both types of stores were loaded with 45 Ca 2ϩ and then challenged with IP 3 in efflux medium. The open circles and dotted line in Fig. 6C illustrate the Ca 2ϩ release from the SERCA-containing Ca 2ϩ store as a function of the [IP 3 ]. The closed circles and full line are the values for the tg-insensitive Ca 2ϩ store. The EC 50 was 1.2 M IP 3 for the SERCA-containing Ca 2ϩ store (dotted arrow) and 5.2 M IP 3 for the tg-insensitive Ca 2ϩ store (solid arrow). A maximal [IP 3 ] released 83% of the ionophore-releasable Ca 2ϩ from the SERCA-containing Ca 2ϩ store but only released 11% from the tg-insensitive Ca 2ϩ store. A similar incomplete Ca 2ϩ release at the highest [IP 3 ] and a higher EC 50 for IP 3 were previously also observed for the Pmr1-induced Ca 2ϩ store in COS-1 cells (14).
Ca 2ϩ Signals in Intact A7r5 Cells-The addition of 10 M arginine-vasopressin to control A7r5 cells produced an immediate [Ca 2ϩ ] i rise (Fig. 7A). This immediate response was abolished when the cells were pretreated with 10 M tg (Fig. 7B). Under these conditions, a delayed abortive [Ca 2ϩ ] i rise occurred in 15% of the cells. This finding indicates that the tg-insensitive Ca 2ϩ store in some cells was large enough to be discharged by external stimulation of the cell. Interestingly the Ca 2ϩ spikes in Pmr1-overexpressing COS-1 cells often also occur after a long latency (14). None of the 16HBE14oϪ cells responded to 100 M ATP in the presence of tg.
Conclusions-We demonstrated in A7r5 and 16HBE14oϪ cells a tg-insensitive Ca 2ϩ store that was less leaky for Ca 2ϩ than the ER. Based on our findings that (i) Pmr1 was present in A7r5 and 16HBE14oϪ cells and that (ii) the Ca 2ϩ uptake mechanism of the tg-insensitive Ca 2ϩ store in A7r5 and 16HBE14oϪ cells and the overexpressed Pmr1 Ca 2ϩ pump in COS-1 cells had the same sensitivity to cyclopiazonic acid and 2,5-di-(tert-butyl)-1,4-benzohydroquinone, we propose that the Pmr1 Ca 2ϩ pump was responsible for loading up the tg-insensitive Ca 2ϩ store in A7r5 and 16HBE14oϪ cells. Since Pmr1 is expressed in the Golgi apparatus (11), this tg-insensitive Ca 2ϩ store in A7r5 and 16HBE14oϪ cells probably corresponds to the Golgi complex. IP 3 released 11% of the Ca 2ϩ accumulated in this compartment in A7r5 cells with an EC 50 that was 5 times higher than for the ER in these cells. This store could also be released in intact cells during agonist stimulation. Heterogeneous nonmitochondrial Ca 2ϩ stores therefore exist in A7r5 and 16HBE14oϪ cells.