Caveolin-1 Alters the Pattern of Cytoplasmic Ca2+ Oscillations and Ca2+-dependent Gene Expression by Enhancing Leukotriene Receptor Desensitization*

Background: The scaffolding protein caveolin-1 coordinates membrane signaling clusters, but how this affects Ca2+ oscillations is unknown. Results: Caveolin-1 accelerates the rundown of Ca2+ oscillations to the agonist leukotriene C4, which is prevented by modest inhibition of protein kinase C. Conclusion: Caveolin-1 increases receptor desensitization through Ca2+-dependent stimulation of protein kinase C. Significance: The findings reveal how caveolin-1 regulates receptor-dependent Ca2+ signaling.

tion. The kinetics of receptor desensitization vary over orders of magnitude. Kainate receptors desensitize within milliseconds (1), whereas the process develops over hundreds of milliseconds for NMDA receptors (2). By contrast, desensitization of G proteincoupled receptors develops over tens of seconds (3).
In many cell types, moderate stimulation of cell surface receptors that activate the phospholipase C pathway evokes a series of cytoplasmic Ca 2ϩ oscillations (4). Information can be encoded in the amplitude, frequency, and spatial profile of the oscillatory signal, leading to activation of selective downstream responses including mitochondrial metabolism, secretion, and gene expression (5).
In mast cells, the activation of cysteinyl leukotriene type I (CysLT1) 3 receptors with the proinflammatory agonist leukotriene C 4 (LTC 4 ) evokes cytoplasmic Ca 2ϩ oscillations. The CysLT1 receptor shows homologous desensitization through which protein kinase C, including the Ca 2ϩ -dependent ␣ isoform (6), phosphorylates three serine residues on the carboxyl terminus to uncouple the receptor from phospholipase C (7). Acute inhibition of protein kinase C, down-regulation of Ca 2ϩdependent protein kinase C isoforms, or siRNA knockdown of protein kinase C␣ all convert the oscillatory Ca 2ϩ response into a more sustained Ca 2ϩ rise, demonstrating that the oscillatory Ca 2ϩ signals are a consequence of reversible receptor desensitization (6), likely reflecting pulsatile increases in InsP 3 .
Reversible receptor desensitization enables phasic Ca 2ϩ signals to occur, thereby bypassing the deleterious consequences of a sustained Ca 2ϩ rise that include excitotoxicity and Ca 2ϩdependent inhibition of signaling molecules. Mechanisms that control the rate and extent of receptor desensitization will therefore have a profound influence on the spatiotemporal pattern of agonist-evoked Ca 2ϩ signals and the subsequent activation of downstream targets. Here we report that the scaffolding * This work was supported by a Medical Research Council grant (to A. B. P.). protein caveolin-1 enhances desensitization of CysLT1 receptors. The amplitude of Ca 2ϩ oscillations is initially increased by caveolin-1, because of enhanced coupling between the receptor and phospholipase C. However, the increased Ca 2ϩ mobilization stimulates Ca 2ϩ -dependent protein kinase C, which then terminates the oscillatory response by accelerating receptor desensitization. Our work identifies caveolin-1 as a bimodal regulator of intracellular Ca 2ϩ signals.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-The rat mast cell line RBL-1 was purchased from ATCC (via United Kingdom supplier LCG Standards). For regular maintenance, cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C with 5% CO 2 as described (8). For experiments, RBL-1 cells were transfected using the Amaxa system and then incubated overnight in medium without penicillin/streptomycin. Experiments were carried out 24 -36 h after transfection.
Plasmid Constructs-Wild type caveolin-1 tagged with EGFP was kindly provided by Dr. Suetsugu (University of Tokyo, Japan) (9), and the pleckstrin homology domain linked to GFP (GFP-PHD) was kindly provided by Dr. Meyer (Stanford University) via Addgene. Both caveolin-1-myc-RFP and the tyrosine 14 phospho-inactive form (Y14F caveolin-1-myc-RFP mutant) were kind gifts from Dr. Nabi (University of British Columbia, Canada) (10). Transfection efficiency for these constructs was similar and varied between 30 and 45%.
Cytoplasmic Ca 2ϩ Measurements-Cells were loaded with Fura-2/AM for 40 min at room temperature in the dark and then washed three times with a solution composed of 145 mM NaCl, 2.8 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 10 mM D-glucose, and 10 mM HEPES, pH 7.4, with NaOH as described (11). Cells were left for 15 min to allow further de-esterification. Ca 2ϩ -free solution contained 145 mM NaCl, 2.8 mM KCl, 2 mM MgCl 2 , 10 mM D-glucose, 10 mM HEPES, and 0.1 mM EGTA, pH 7.4, with NaOH. Cytoplasmic Ca 2ϩ imaging experiments were carried out using a TILL Photonics system with an IMAGO CCD camera. Cells were excited alternately at 356 and 380 nm, and images were acquired every 2 s. Images were analyzed off line using IGOR Pro for Windows. Ca 2ϩ signals are represented at a ratio of 356/380 nm. The experiments illustrated in Fig. 9 were carried out using the imaging system in the laboratory of Dr. Glitsch (Department of Physiology, Anatomy and Genetics, University of Oxford) while repair work was being carried out on our imaging system. Immunocytochemistry and Image Analysis-For immunocytochemistry, cells were transfected with caveolin-1-RFP and FLAG-tagged CysLT1 receptor and then fixed 48 h later with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100. After that, cells were incubated with blocking solution (Thermo Scientific) for 1 h and then incubated with specific primary antibody against FLAG tag (Sigma-Aldrich). Secondary antibody against rabbit IgG was conjugated with Alexa-488 purchased from Invitrogen. Images were obtained by using an Olympus confocal microscope. Relative fluorescence intensity was analyzed using ImageJ software. For cells transfected with GFP-PHD, immunofluorescence images were obtained with a Leica microscope, and the fluorescence intensity was analyzed by ImageJ software. For colocalization studies, confocal images were taken with an FV-1000 confocal microscope (Olympus, Melville, NY), and the colocalization coefficient between two different channels was assessed by the Olympus Fluoview FV1000 system. At least five representative images in each group were used for analysis and 10 different areas on the cell membrane and in the cytosol were selected to obtain Pearson's correlation coefficient.
Gene Reporter Assay-24 -36 h following transfection with an EGFP-based reporter plasmid under an NFAT promoter, cells were stimulated with LTC 4 (see text for specific times). The percentage of GFP-positive cells was measured as describe previously (8).
Statistical Analysis-All results were expressed as mean Ϯ S.E. A two-tailed Student's t test was used to compare differences between two groups, and one-way analysis of variance was used to compare differences when groups numbered three or more (GraphPad Prism). Statistical significance was set at p Ͻ 0.05, with one, two, and three asterisks denoting p Ͻ 0.05, 0.01, and 0.001, respectively.

RESULTS
Endogenous levels of caveolin-1 were virtually undetectable in Western blots from RBL-1 cells (data not shown), so we overexpressed the GFP-tagged protein to study its impact on Ca 2ϩ oscillations. In non-transfected (wild type) cells, stimulation with LTC 4 evoked a series of cytoplasmic Ca 2ϩ oscillations (Fig.  1A), which decreased slightly over time due to receptor desensitization ( Fig. 1B) (6). Expression of caveolin-1-GFP substantially altered the pattern of the Ca 2ϩ oscillations (Fig. 1A, dotted trace). The amplitudes of the initial Ca 2ϩ oscillations evoked by LTC 4 were now considerably larger than in non-transfected cells (Fig. 1, A and C), but the oscillations ran down more quickly and so were fewer in number over a 600 to 700-s recording period (Fig. 1B). Analysis of the various oscillatory parameters revealed that the total Ca 2ϩ rise associated with each oscillation (area under the spike) was significantly larger in cells expressing caveolin-1-GFP (Fig. 1D); this reflected both an increase in the amplitude of each Ca 2ϩ oscillation (Fig. 1C) as well as an increase in duration (Fig. 1E). Cytoplasmic Ca 2ϩ during each oscillation was therefore elevated for a longer time in the presence of caveolin-1-GFP.
The effects of caveolin-1-GFP were not mimicked by expression of GFP alone (Fig. 1, B and C). However, caveolin-1-RFP replicated the marked effects of caveolin-1-GFP on the pattern of Ca 2ϩ oscillations (data not shown).
Responses in the presence of external Ca 2ϩ reflect both InsP 3 -dependent Ca 2ϩ release and Ca 2ϩ influx through CRAC channels, the latter being required to replenish the stores with Ca 2ϩ in readiness for the next oscillatory cycle. To see which of these processes was affected by caveolin-1, we separated Ca 2ϩ release from Ca 2ϩ entry by stimulating cells with LTC 4 in the absence of external Ca 2ϩ and then readmitting external Ca 2ϩ once the oscillations had run down. Because of the lack of Ca 2ϩ influx, Ca 2ϩ oscillations decreased in size over time and were lost typically within 400 s after stimulation (the control cell is shown in Fig. 1F, and aggregate data are summarized in Fig.  1G). Readmission of external Ca 2ϩ after 600 s resulted in Ca 2ϩ entry through CRAC channels (Fig. 1F). Expression of caveolin-1-GFP increased the amplitude of the Ca 2ϩ oscillations in Ca 2ϩ -free solution considerably ( Fig. 1F; aggregate data are shown in Fig. 1H), but these oscillations ran down more quickly than the corresponding control recordings (Fig. 1G). Readmission of external Ca 2ϩ led to a significantly larger rate of rise of Ca 2ϩ (Fig. 1, F and I), indicating increased store-operated Ca 2ϩ influx. Unlike the case of caveolin-1-GFP, expression of GFP alone had no effect on the number of Ca 2ϩ oscillations (Fig.  1G), the size of the oscillations (Fig. 1H), or store-operated Ca 2ϩ entry ( Fig. 1I and Ref. 12).
Caveolin-1 increases the interaction between the heterotrimeric GTP-binding protein G q and phospholipase C (13), a mechanism that could explain the increase in amplitude of the Ca 2ϩ oscillations. If so, caveolin-1 should be expressed in the plasma membrane. Immunocytochemical studies revealed the presence of both FLAG-tagged CysLT1 receptors and caveolin-1-RFP in the plasma membrane ( Fig. 2A). A significant fraction of caveolin-1-RFP was also found in the cytoplasm, likely reflecting its contribution to vesicle sorting (14). To test for colocalization, at the level of resolution provided by confocal microscopy, we merged images and measured the subcellular distribution of each protein using line scanning ( Fig. 2A, merged panel). CysLT1 receptor distribution showed two clear peaks, corresponding to plasma membrane at the two edges of the cell (Fig.  2B, green traces). Although caveolin-1-RFP was present within the cytoplasm, two peaks at the cell periphery were also resolvable, indicating a plasma membrane location. We quantified the extent of overlap of the two proteins using Pearson's correlation coefficient (Fig. 2C). Under both basal and stimulated conditions (LTC 4 exposure for 10 min), there was a much better correlation between FLAG-tagged CysLT1 receptor and caveolin-1-RFP in the membrane than in the cytoplasm, and stimulation did not change the correlation coefficient (Fig. 2C).
If caveolin-1 increases receptor-phospholipase C coupling, two predictions are that, first, InsP 3 levels should increase more following stimulation in the presence of caveolin-1 than in wild type cells, and second, less Ca 2ϩ should remain within the InsP 3 -sensitive store after the Ca 2ϩ oscillations have run down in cells expressing caveolin-1. Using the GFP-PHD construct as a means for monitoring InsP 3 levels in individual cells (15)(16)(17), we found that stimulation with LTC 4 for 5 min resulted in a modest decrease in the membrane/cytosol ratio of GFP-PHD (decrease of 24.5 Ϯ 1.7%; Fig. 3, A and B), and this was slightly more pronounced when caveolin-1-RFP was expressed (31.2 Ϯ 1.4%, p Ͻ 0.05; Fig. 3, A and B). To test the second prediction, we stimulated cells with LTC 4 in the absence of external Ca 2ϩ , and then once the oscillations had stopped, we applied thapsigargin in Ca 2ϩ -free solution to estimate how much Ca 2ϩ For D and E, the area and duration of each oscillation was measured, and then the data were pooled together. F, store-operated Ca 2ϩ influx measured following stimulation with LTC 4 in Ca 2ϩ -free solution for 600 s followed by readmission of external Ca 2ϩ was compared between the two conditions. G, the graph compares the rundown of Ca 2ϩ oscillations among WT (24 cells), GFP-expressing (29 cells), and caveolin-1-expressing cells (26 cells) when cells were stimulated with LTC 4 in the absence of external Ca 2ϩ as shown in F. H, the amplitude of the first Ca 2ϩ oscillation, evoked by LTC 4 in Ca 2ϩ -free solution, is compared. I, the rates of store-operated Ca 2ϩ entry, measured by differentiating the Ca 2ϩ rise following readmission of 2 mM Ca 2ϩ as in F, are compared for the conditions shown (each bar denotes Ͼ25 cells from three independent experiments).  remained within the store (Fig. 3C). The thapsigargin-mobilizable Ca 2ϩ pool was significantly reduced in cells expressing caveolin-1-GFP (Fig. 3D, p Ͻ 0.01).
The scaffolding domain of caveolin-1, which involves amino acids between residues 82 and 101, is required for interaction with receptors, G proteins, and other signaling molecules (18,

19).
A central core of four amino acids within this region, encompassing 92 FTVT 95 , is critical for association with G proteins (20). To determine whether this central core was required for regulation of Ca 2ϩ signals generated by CysLT1 receptors, we made mutations within the site to see the effect on Ca 2ϩ oscillations. Following transfection of a GFP-tagged caveolin-1 construct in which phenylalanine (Phe-92) and threonine (Thr-95) had been mutated to alanines, several Ca 2ϩ oscillations were seen in Ca 2ϩ -free solution (Fig. 4A); these were similar in size to those obtained in wild type cells (Fig. 4B). The number of oscillations in Ca 2ϩ -free solution (data not shown) and the rate of rise of the Ca 2ϩ signal due to store-operated entry were also not significantly different from control cells (Fig. 4C). Cytoplasmic Ca 2ϩ oscillations in response to LTC 4 showed only modest rundown when transfected with the mutated caveolin-1 (Fig.  4D), which was not different from wild type cells (Fig. 4E). The size of these oscillations was also similar to that in wild type cells (Fig. 4F).
We considered the possibility that expression of F92A,T95A caveolin-1-GFP was considerably lower than caveolin-1-GFP, thereby explaining the lack of effect of mutant caveolin-1 on Ca 2ϩ oscillations. We therefore compared GFP fluorescence in cells transfected with either caveolin-1-GFP or F92A,T95A caveolin-1-GFP. There was no difference in either the profile of GFP expression between the two groups ( Fig. 4G) or the averaged GFP fluorescence between the groups (Fig. 4H). In Fig. 4I, Ca 2ϩ signals evoked by LTC 4 are compared between a cell expressing caveolin-1-GFP and one expressing F92A,T95A caveolin-1-GFP. The cells had almost identical levels of GFP expression (92 and 93 gray scale units, respectively). However, only the presence of caveolin-1-GFP altered the pattern of the Ca 2ϩ oscillations. Confocal images showed that both caveolin-1-RFP and F92A,T95A caveolin-1-RFP were expressed at the plasma membrane with FLAG-tagged CysLT1 receptors (Fig.  4J). Pearson's correlation coefficient between mutant caveolin-1-RFP and CysLT1 receptors was similar to that seen for caveolin-1-RFP and the receptors (Fig. 2C). Collectively, these results show that the scaffolding domain of caveolin-1 is important for the modulation of agonist-evoked Ca 2ϩ oscillations.
Phosphorylation of caveolin-1 on tyrosine 14 by Src family kinases potentiates growth factor signaling and is required for internalization of caveolae (21). Expression of an RFP-tagged caveolin-1 construct with a point mutation converting tyrosine to phenylalanine (Y14F) was expressed in the plasma membrane (Fig. 4J) and mimicked the effects of caveolin-1-GFP expression on agonist-induced Ca 2ϩ oscillations. The initial Ca 2ϩ transients were larger (Fig. 4K), and fewer oscillations were obtained (Fig. 4L). Internalization of caveolin-1 through phosphorylation of tyrosine 14 therefore does not contribute to the effects of caveolin-1 on LTC 4 -driven Ca 2ϩ signals.
We designed experiments to identify the mechanism responsible for the accelerated rundown of Ca 2ϩ oscillations seen in the presence of caveolin-1. To see whether this was dependent on Ca 2ϩ release or Ca 2ϩ entry, we stimulated cells in the absence of external Ca 2ϩ but with the plasma membrane Ca 2ϩ pump blocked with La 3ϩ . Under these conditions, Ca 2ϩ release can no longer be exported out of the cell and instead is sequestrated back into the stores. Ca 2ϩ oscillations therefore continue for several minutes, reflecting regenerative Ca 2ϩ release in the absence of Ca 2ϩ influx (11,22). Stimulation with LTC 4 in wild type cells evoked a series of repetitive Ca 2ϩ oscillations that decreased slightly in number over time (Fig. 5, A and B). By contrast, in cells expressing caveolin-1-GFP, larger Ca 2ϩ spikes were obtained initially, which then ran down quickly (Fig. 5, A  and B). As with the responses in the presence of external Ca 2ϩ , the amplitude of the first oscillation (Fig. 5C), as well as the duration of the oscillations (Fig. 5D), was significantly increased in the presence of caveolin-1-GFP. Rundown of Ca 2ϩ oscillations in the presence of caveolin-1 therefore arises from Ca 2ϩ release.
Further evidence that Ca 2ϩ release from the stores in caveolin-1-expressing cells contributes to the rundown of the oscillations is shown in Fig. 6. In these experiments, we sought to partially lower the Ca 2ϩ content of the stores in order to reduce the size of each Ca 2ϩ oscillation upon stimulation. We therefore incubated control (non-transfected) cells in Ca 2ϩ -free solution for 10 min and found that this was sufficient to reduce the extent of Ca 2ϩ release by thapsigargin by ϳ 30% when compared with control cells pre-exposed to Ca 2ϩ -free solution for just a few seconds prior to stimulation with thapsigargin (Fig. 6,  A and C). We then stimulated cells in Ca 2ϩ -free solution containing 1 mM La 3ϩ to eliminate the increased Ca 2ϩ influx due to the reduced store Ca 2ϩ content from affecting the oscillatory pattern. Oscillatory Ca 2ϩ responses to LTC 4 were sustained both in cells pretreated with Ca 2ϩ -free solution acutely (Fig. 6, D and I) and in those following 10 min of pretreatment (Fig. 6, E  and I), although the size of the oscillations was smaller in the latter case (Fig. 6H), reflecting the reduced store Ca 2ϩ content. In cells expressing caveolin-1-GFP and incubated in Ca 2ϩ -free solution for 10 min, the extent of Ca 2ϩ release induced by thap- sigargin was similar to control cells treated in the same way (Fig.  6, B and C). Whereas only a few Ca 2ϩ oscillations were seen in response to LTC 4 challenge in caveolin-1-GFP-expressing cells exposed to Ca 2ϩ -free solution for a few seconds prior to stimulation (Fig. 6, F and I), preincubation for 10 min with Ca 2ϩfree external solution resulted in more prolonged oscillatory Ca 2ϩ signals following agonist stimulation (Fig. 6, G and I). The amplitude of the first Ca 2ϩ oscillation was reduced following the 10-min preincubation in Ca 2ϩ -free solution prior to stimulation (Fig. 6H). Hence, lowering the Ca 2ϩ content of the stores results in prolonged oscillatory Ca 2ϩ signals in the presence of caveolin-1-GFP. These results are consistent with the view that the enhanced Ca 2ϩ release normally seen in caveolin-1-expressing cells is responsible for the accelerated rundown of the oscillations.
One way whereby enhanced Ca 2ϩ release can increase the rundown of Ca 2ϩ oscillations is through Ca 2ϩ -dependent inactivation of InsP 3 receptors. However, the Ca 2ϩ release transient following phospholipase C-coupled P2Y receptor activation after CysLT1 receptors had been desensitized was slightly larger in caveolin-1-expressing cells (Fig. 7B) than in the corresponding controls ( Fig. 7A; aggregate data are shown in Fig. 7C). Inactivation of the InsP 3 receptor therefore plays little role in the rundown of Ca 2ϩ oscillations in the presence of caveolin-1.
We considered that rundown of the Ca 2ϩ oscillations was a consequence of the accelerated desensitization of the CysLT1 receptor. These receptors are desensitized following Ca 2ϩ -dependent protein kinase C-mediated phosphorylation of a series of serine residues on the carboxyl terminus of the receptor, and   JUNE 20, 2014 • VOLUME 289 • NUMBER 25 we had previously found a major role for protein kinase C␣ in the desensitization process (6). Increased Ca 2ϩ release following caveolin-1-GFP expression would lead to stronger activation of Ca 2ϩ -dependent protein kinase C isoforms and thus should result in more pronounced receptor desensitization. To test this possibility, we used a low concentration of the protein kinase C inhibitor Go6983 (1 nM) to reduce but not abolish kinase activity, as substantial block of the kinase results in non-oscillatory Ca 2ϩ signals (6). The typical oscillatory Ca 2ϩ response in wild type cells induced by LTC 4 stimulation (Fig.  8A) was only weakly affected by the low concentration of Go6983 (Fig. 8, B and J). However, the rapid rundown of Ca 2ϩ oscillations in cells expressing caveolin-1-GFP (Fig. 8C) was largely prevented by the protein kinase C inhibitor (Fig. 8, D and  J). Identical results were obtained with a structurally different protein kinase C blocker, GF109203X (1 nM; Fig. 8, E-H and K). Many agonists of G protein-coupled receptors elicit responses by occupying only a fraction of the total receptors. We therefore reasoned that increasing the number of available CysLT1 receptors in the plasma membrane in cells expressing caveolin-1-GFP should lead to an increased likelihood for LTC 4 to encounter a non-desensitized receptor, which should reduce the rate of rundown of Ca 2ϩ oscillations. We therefore transfected cells with plasmids for caveolin-1-GFP and the CysLT1 receptor. Increased expression of CysLT1 receptors significantly prolonged the oscillatory Ca 2ϩ response compared with cells transfected with caveolin-1-GFP alone (Fig. 8, I and J). Despite coupling to phospholipase C via G q proteins, P2Y receptor-driven Ca 2ϩ release was unaffected by caveolin-1-GFP expression (ATP responses measured at 600 s in wild type cells and in those expressing caveolin-1-GFP were similar (Fig.  7C, black bars)). This suggests that P2Y and CysLT1 receptors might couple to phospholipase C differently, with the leukotriene receptor more prominent in caveolin-1-rich domains. Lipid rafts can be disrupted by methyl-␤-cyclodextrin (M␤CD), a compound that removes cholesterol from the plasma membrane. Treatment with M␤CD abolished LTC 4 -dependent Ca 2ϩ responses ( Fig. 9A) but had no significant effect on P2Yevoked Ca 2ϩ signals (Fig. 9B). Different agonists thus differ in their sensitivity to regulation by caveolin-1 and lipid rafts.

Caveolin-1 and Cytoplasmic Ca 2؉ Oscillations
To see whether the altered pattern of Ca 2ϩ signaling by caveolin-1 had functional relevance, we measured Ca 2ϩ -dependent gene expression using a GFP construct under a promoter driven by the Ca 2ϩ -dependent transcription factor NFAT (8,23). In non-stimulated cells, expression of GFP was low (Fig. 10A), but it increased ϳ4-fold after LTC 4 was added to the culture medium. Basal gene expression was also low in caveolin-1-RFP-expressing cells, but stimulation resulted in a relatively weaker rise (ϳ 2.5 fold, Fig. 10A; p Ͻ 0.01). Because NFAT activation is tightly linked to local Ca 2ϩ entry through CRAC channels following physiological levels of stimulation in RBL cells (8,24), we hypothesized that the larger size and longer duration of the Ca 2ϩ release transients in the presence of caveolin-1 (Fig. 1, C and E) increased Ca 2ϩ -dependent slow inactivation of CRAC channels (25,26) and thereby reduced NFAT-dependent gene expression. One way to reduce Ca 2ϩ -dependent slow inactivation of CRAC channels is to use a different stimulation protocol. Stimulation with LTC 4 for 10 min in the absence of external Ca 2ϩ fails to activate gene expression despite evoking several Ca 2ϩ oscillations (8). Readmission of external Ca 2ϩ , a few minutes after the oscillations have run down, allows for recovery from slow inactivation. Using this protocol, we found that expression of caveolin-1-RFP now failed to reduce NFAT-dependent gene expression (Fig. 10B). In fact, expression increased somewhat, in accordance with the increase in store-operated Ca 2ϩ entry that arises from the more extensive store depletion (Fig. 1F). Because Ca 2ϩ -dependent slow inactivation requires a rise in bulk Ca 2ϩ , it can be prevented by the slow Ca 2ϩ chelator EGTA (25,26). We therefore reduced the Ca 2ϩ rise by loading the cytoplasm with EGTA. EGTA had no inhibitory effect on LTC 4 -induced gene expression in control cells (Fig. 10C), but it prevented the reduction in gene expression seen in the presence of caveolin-1-RFP (Fig.   10C). The reduction in LTC 4 -driven gene expression in caveolin-1-RFP-expressing cells was not seen when F92A,T95A caveolin-1-RFP was expressed instead (Fig. 10D). Gene expression was also impaired after lipid raft disruption with M␤CD (Fig. 10D). The reduction in gene expression to LTC 4 in cells expressing caveolin-1-RFP was prevented by pretreating cells with 1 nM Go6983 (Fig. 10E), a concentration that rescued repetitive Ca 2ϩ signaling to agonist (Fig. 8).

DISCUSSION
Caveolin-1 is a conserved plasma membrane scaffolding protein that facilitates interaction between signaling molecules within subcompartments of the membrane. One such interaction involves enhanced coupling between G q and phospholipase C, thereby generating larger increases in InsP 3 (13). Our data add a new aspect to this role for caveolin-1, namely in  Here, cells were stimulated with LTC 4 in Ca 2ϩ -free solution for 8 min, and then external Ca 2ϩ was readmitted for 5 min before cells were placed in culture medium and left in the incubator overnight. C, aggregate data for the various conditions are compared. Stimulation with LTC 4 was carried out as in A. D, aggregate data for the conditions shown are compared. Stimulation with LTC 4 was as in A. E, the effects of a low concentration of Go6983 on gene expression induced by LTC 4 is compared between control cells and those expressing Cav1-RFP. All data are aggregates from three independent experiments with between 50 and 80 cells from each experiment. Stimulation with LTC 4 was as in A. NS, not significant; **, p Ͻ 0.01; ***, p Ͻ 0.001. triggering receptor desensitization and thus terminating Ca 2ϩ -dependent responses following physiological levels of stimulation.
Stimulation of CysLT1 receptors with LTC 4 leads to repetitive Ca 2ϩ oscillations, which reflect regenerative Ca 2ϩ release followed by transient Ca 2ϩ entry through CRAC channels (11). The Ca 2ϩ oscillations can be converted into a more prolonged non-oscillatory Ca 2ϩ rise by interfering with protein kinase C activity (6). Protein kinase C triggers CysLT1 receptor desensitization through phosphorylation of three serine residues on the carboxyl terminus of the receptor (7). Overexpression of caveolin-1 resulted in Ca 2ϩ oscillations with larger amplitude and greater duration, as expected from increased G q -phospholipase C coupling. However, the oscillations ran down more quickly and Ca 2ϩ -dependent gene expression was reduced following overexpression of caveolin-1. The rundown was not due to compromised store refilling or inactivation of the InsP 3 receptors. Rather, the increased Ca 2ϩ release in the presence of caveolin-1 led to stronger Ca 2ϩ -dependent activation of protein kinase C, which resulted in increased leukotriene receptor desensitization. Partial block of protein kinase C reversed the effects of caveolin-1 on oscillation amplitude, duration, rundown, and gene expression. The increase in size and duration of Ca 2ϩ release in the presence of caveolin-1 would lead to enhanced Ca 2ϩ -dependent inactivation of CRAC channels (25,26). Because Ca 2ϩ microdomains near these channels activate gene expression, larger or prolonged Ca 2ϩ release impairs transcription by reducing CRAC channel activity.
Our results reveal a novel mechanism for cytsteinyl leukotriene receptor desensitization involving caveolin-1. Enhanced Ca 2ϩ release due to increased coupling between the receptor and phospholipase C both activates Ca 2ϩ -dependent protein kinase C, which leads to pronounced receptor desensitization, and accelerates Ca 2ϩ -dependent slow inactivation of CRAC channels. Activation of this pathway likely involves subcompartments within the membrane, as P2Y receptor-dependent Ca 2ϩ release was unaffected by caveolin-1. By regulating desensitization, caveolin-1 is therefore an important determinant of the duration of receptor stimulation and thus of subsequent Ca 2ϩ -dependent downstream signaling.