Impairment of TRPC1–STIM1 channel assembly and AQP5 translocation compromise agonist-stimulated fluid secretion in mice lacking caveolin1

Saliva is required for maintaining oral health as well as for mastication, swallowing and proper speech while loss of salivary gland (SG) fluid secretion leads to conditions including xerostomia, dysphagia and opportunistic infections in the oral cavity (Baum, 1993; Kaplan and Baum, 1993). SG fluid secretion is stimulated in response to neurotransmitter activation of plasma membrane (PM) receptors leading to secretion of ions and fluid from acinar cells (Melvin et al., 2005; Putney, 1986). Receptor-stimulation leads to an increase in intracellular [Ca²⁺]_i, which regulates key ion transporters and channel proteins that control fluid-secretion. While both intracellular Ca²⁺ release and influx across PM contribute to the increase in [Ca²⁺]_i, sustained fluid secretion is primarily dependent on Ca²⁺ entry via the store-operated Ca²⁺ entry (SOCE) mechanism. Fluid secretion via the apical regions of SG acini is mediated by aquaporin 5 (AQP5) channels (Ambudkar, 2000; Borgnia et al., 1999; Lee et al., 2012.; Melvin et al., 2005). These water channels are retained in close proximity to the apical membrane and undergo an increase in surface expression in response to the elevation of [Ca²⁺]_i following receptor activation (Gresz et al., 2004; Ishikawa et al., 2005). Interestingly, mouse models with deletion in key molecular components involved in regulation of [Ca²⁺]_i, such as the M3 muscarinic receptor or the inositol (1,4,5)-trisphosphate receptors IP₃R2 and IP₃R3, display severe loss in pilocarpine (a muscarinic receptor agonist) stimulated saliva secretion (Futatsugi et al., 2005; Gautam et al., 2004; Matsui et al., 2000).

The transient receptor potential canonical 1 (TRPC1) channel and Orai1 have been shown to contribute for agonist-stimulated Ca²⁺ influx in salivary and pancreatic acinar cells, respectively (Hong et al., 2011; Liu et al., 2007; Ong et al., 2007; Singh et al., 2001). In addition, STIM1 has been established as the primary regulator of SOCE and is dynamically regulated in response to endoplasmic reticulum (ER) Ca²⁺ store depletion (Liou et al., 2007). Following Ca²⁺ store depletion, STIM1 is translocated to specific ER–plasma-membrane (PM) junctional regions where it activates Orai1 and TRPC1 channels. Importantly, STIM1 interaction with TRPC1 is shown to be essential for TRPC1 activation (Huang et al., 2006; Ong et al., 2007; Pani et al., 2009). Increased expression of STIM1 or TRPC1 in submandibular gland (SMG) acinar cells is shown to enhance Ca²⁺ entry (Morita et al., 2011; Singh et al., 2001). Further, Orai1 is expressed in SMG and is required for PM recruitment of TRPC1 (Cheng et al., 2011). Thus, all three proteins contribute to agonist-stimulated Ca²⁺ entry in SG. While evidence suggests that specific ER–PM junctional domains are involved in the interaction of STIM1 with Orai1 and TRPC1 (Carrasco and Meyer, 2011; Pani et al., 2012), precise nature of these domains and the molecular mechanisms underlying channel assembly under physiological conditions are poorly understood.

Caveolae are cholesterol enriched heterogeneous microdomains containing several receptors and ion channels that are known to regulate cellular functions such as transcytosis, protein sorting, cell adhesion and migration. These microdomains are distinct flask-shaped PM invaginations that are determined by multimeric assembly of the cholesterol binding protein Caveolin1 (Cav1) (Parton and Simons, 2007; Patel et al., 2008; Razani et al., 2002). Interestingly, agonist-stimulated Ca²⁺ signals have been shown to originate at Cav1 enriched PM regions (Isshiki and Anderson, 2003). Consistent with this, several Ca²⁺ signaling components, including TRPC channels, localize within caveolar domains (Pani and Singh, 2009). TRPC1 associates with Cav1 which serves as scaffold to retain the channel in the plasma membrane region and facilitate its interaction with STIM1 (Pani et al., 2009; Sundivakkam et al., 2009). Recent studies also suggest a role of lipid rafts domains (LRD) and Cav1 in regulation of Orai1 (Sathish et al., 2012; Yu et al., 2010). However, direct physiological evidence for Cav1 in regulation of Orai1-STIM1 channel assembly remains to be determined.

In this study, we have assessed the role of Cav1 in agonist-stimulated Ca²⁺ entry and regulation of fluid secretion in mouse salivary glands. We report that abrogation of caveolin1 in mice (Cav1^−/− mice) significantly attenuated Ca²⁺ influx in acinar cells and consequently led to loss of saliva secretion. While agonist-induced STIM1 interaction with Orai1 was unaltered, plasma membrane localization of TRPC1, its association with lipid raft microdomains and interaction with STIM1 were disrupted. In addition, targeting of AQP5 to the apical region of acinar cells and its insertion into the apical membrane in response to agonist-stimulation, were also impaired. Finally, adenovirus-mediated in vivo expression of Cav1 in SMG of Cav1^−/− mice restored agonist-stimulated TRPC1-STIM1 association, store-dependent Ca²⁺ entry, apical localization of AQP5 and fluid-secretion. Together, these findings suggest that by controlling TRPC1-mediated Ca²⁺ influx and AQP5 trafficking, Cav1 has a crucial physiological role in regulating salivary gland function. Further, Cav1 is essential for the selective organization of TRPC1-STIM1 channel assembly, but not for Orai1-STIM1 channels, thereby suggesting microdomain heterogeneity in channel regulation and function.

Ultrastructural analysis of SMG showed typical caveolar structures in basal PM region of Cav1^+/+ acini but not in Cav1^−/− (Fig. 1A; supplementary material Fig. S1Ai). Caveolar microdomains were also identifiable around apical PM of Cav1^+/+ acinar cells (supplementary material Fig. S1Aii). While loss of caveolae in Cav1^−/− mice has been shown in other tissues (Razani et al., 2002), this is a novel finding in salivary glands. Importantly, TRPC1 was detected in the basolateral regions of Cav1^+/+ acinar cells whereas in Cav1^−/− acini the channel localization was diffused (Fig. 1B). Notably, STIM1 localization was not affected by the loss of Cav1 (Fig. 1B). The relative protein distribution from membrane raft fractionations of SMG showed a predominant association of Cav1 with low-density, raft fractions (fraction 3–5) in Cav1^+/+ but not Cav1^−/− (Fig. 1C). While expression of Cav2 was significantly reduced in Cav1^−/− SMG (Fig. 1C; Fig. 3D), possibly due to its decreased stability resulting from Cav1 deletion (Razani et al., 2002), Cav3 was undetectable (Fig. 3D). TRPC1 was distributed in raft and non-raft SMG fractions in Cav1^+/+ SMG, but interestingly in Cav1^−/− SMG, the raft association of TRPC1 was undetectable (Fig. 1C). In contrast to TRPC1 distribution, STIM1 retained its membrane raft partitioning in Cav1^−/− SMG, although there was relatively less protein (∼15% less) in these fractions compared to Cav1^+/+ (Fig. 1C; supplementary material Fig. S1B). Importantly, the relative distributions of Lyn, Gαq/11 and GM1 ganglioside were largely unaltered (Fig. 1C), suggesting loss of Cav1 likely did not alter the integrity of non-caveolar raft domains. Thus, the microdomain association of STIM1, in Cav1^−/− SMG, is non-caveolar, which could be attributable to its lipid binding potential via its unique C-terminal polybasic region and a putative PIP2-binding domain therein (Carrasco and Meyer, 2011).

TRPC1 and STIM1 have been reported to translocate into LRD following CCh- or Tg-stimulation (Pani et al., 2008). Thus we examined this dynamic property in vivo in SG following pilocarpine stimulation. In Cav1^+/+, but not in Cav1^−/− SMG, both TRPC1 and STIM1 were significantly recruited to the membrane-raft fractions (BF) post stimulation (Fig. 1Di; Fig. 1Dii) additionally, a reduced level of TRPC1 was seen in the Cav1^−/− non-raft fractions (HF). This prompted us to check for the relative expression (mRNA and protein) of TRPC1 between tissues. While, the levels of TRPC1 transcript were comparable between the two sets of samples (supplementary material Fig. S1C), the protein was significantly reduced in Cav1^−/− SMG (∼40% decrease vs Cav1^+/+; supplementary material Fig. S1Di,ii). This decrease in TRPC1 protein could be due to a reduced stability (increase in protein turnover) resulting from aberrant trafficking of the channel in absence of Cav1. However, further studies will be required to understand the underlying mechanism. Further, the fraction of STIM1 translocating into raft domains was significantly lower in Cav1^−/− SMG (Fig. 1Di; Fig. 1Dii). This does not reflect any regulatory alterations in STIM1 expression, as the levels of STIM1 transcript and protein were similar in Cav1^+/+ and Cav1^−/− SMG (supplementary material Fig. S1Di). Thus, it seems the presence of Cav1 partly influences STIM1 association with raft microdomains. Compelling evidence demonstrate that STIM1 gets recruited to specific ER-PM juxtaposed cellular domains. It is conceivable that a Cav1 interacting protein and/or distinct cytoskeletal elements might be involved in directing STIM1 clusters to selective microdomains.

It has been reported the TRPC1-STIM1-Orai1 ternary channel complex regulates Ca²⁺ influx in several cell types. Thus, to determine the impact of Cav1 on agonist stimulated Ca²⁺ entry, we asked whether Cav1 deletion alters Orai1 and TRPC1 function. Co-immunoprecipitation analyses showed interaction of TRPC1, Orai1 and STIM1 in pilocarpine-stimulated Cav1^+/+ SMG but not in Cav1^−/− tissues (Fig. 2Ai,ii). Of note, the recruitment of TRPC1 into this complex was impaired without any change in Orai1-STIM1 interaction (Fig. 2Aii). The loss of TRPC1-STIM1 interaction in Cav1^−/− SMG can be explained by the failure of TRPC1 to partition into membrane-raft microdomains following pilocarpine-stimulation (Fig. 1D), which is primarily due to disruption in trafficking and PM localization of TRPC1 in the absence of Cav1 (Fig. 1B). Global disruption of membrane rafts in HSG cells (which endogenously express Cav1) by MβCD resulted in the inhibition of TRPC1-STIM1-Orai1 complex (supplementary material Fig. S2A). Consistent with these results, Tg- and CCh-stimulated Ca²⁺ entry was significantly reduced in Cav1^−/− acinar cells compared to Cav1^+/+ (Fig. 2Bi, ii). Additionally, Tg-stimulated Ca²⁺ currents were decreased by >50% in Cav1^−/− acinar cells (Fig. 2C). Although Cav1 was recently shown to affect Orai1 trafficking (Sathish et al., 2012; Yu et al., 2010), our data indicate that Orai1-STIM1 interaction is not affected in Cav1^−/− acinar cells and could thus account for the residual Ca²⁺ influx. Further, channels including other TRPC subtypes could also contribute toward this residual Ca²⁺ entry in Cav1^−/− cells and thus cannot be excluded.

To determine the physiological importance of Cav1 in SG, we measured agonist induced fluid secretion. Notably, both the rate of secretion and total volume of saliva were substantially decreased in pilocarpine-stimulated Cav1^−/− mice compared to Cav1^+/+ (Fig. 3A–C). The saliva secreted from Cav1^−/− mice was also relatively hypertonic (supplementary material Fig. S2B). This reduction in fluid secretion was not as a consequence of altered expression of key GPCR mediated Ca²⁺ signaling components as these were essentially similar in Cav1^−/− and Cav1^+/+ SMG lysates (Fig. 3D). While the body weights of Cav1^+/+ and Cav1^−/− mice were similar there was a reduction in the weight of SMGs in Cav1^−/− mice (supplementary material Fig. S2C,D). Ultrastructural analysis of the gland revealed an increase in vesicles in Cav1^−/− SMG acini compared to Cav1^+/+ (supplementary material Fig. S2E). Further, quantitative analysis of stained SMG sections demonstrated that Cav1^−/− SMG had fewer, but significantly larger acini (supplementary material Fig. S2F,G). Staining of SMG sections by PAS showed a relatively darker PAS reaction in Cav1^−/− acini, demonstrating the presence of increased vesicles compared to Cav1^+/+ (supplementary material Fig. S2H). Although further studies will be required, based on these results it is tempting to speculate that vesicle exocytosis, but not biogenesis, is impaired in Cav1^−/− SG.

Water secretion from salivary glands is mediated via AQP5 and primarily driven by the osmotic gradients generated by several ion channels and transporters. Following muscarinic stimulation of SG, AQP5 is translocated to the surface of apical membranes of acinar cells where it mediates water transport into the lumen (Ambudkar, 2000; Borgnia et al., 1999; Gresz et al., 2004; Ishikawa et al., 2005). We examined AQP5 regulation in Cav1^−/− SG. AQP5 was associated with SG raft microdomains and this partitioning was significantly reduced in Cav1^−/− SMG (Fig. 4A). Further, MβCD-mediated disruption of membrane rafts In HSG cells, resulted in the complete exclusion of AQP5 from raft fractions (supplementary material Fig. S3A). A critical aspect of AQP5 regulation, following agonist-stimulation is the recruitment and insertion of the channel into acinar cell apical membrane. Our data demonstrate that in pilocarpine stimulated Cav1^+/+ SMG, AQP5 expression was primarily detected in the apical regions of the acini (shown by arrows, Fig. 4B) while in Cav1^−/− this marked apical localization was not seen (Fig. 4B, also see Z-stack reconstructions of acini in supplementary material Movies 1, 2). Additionally, surface expression of AQP5, as assessed by biotinylation following stimulation, was found to robustly increase in Cav1^+/+ acinar cells whereas no such increase was observed in Cav1^−/− (Fig. 4C). Further, in HSG cells, silencing of TRPC1 or Cav1 as well as inhibition of Ca²⁺ entry by SKF96365 significantly attenuated CCh- or Tg-induced surface expression of AQP5 (Fig. 4D; supplementary material Fig. S3B), suggesting that surface expression of AQP5 is determined by TRPC1 – mediated Ca²⁺ entry. To see if expression of other apical proteins is also altered, we looked at the staining of IP₃R3 and Ca²⁺ activated K⁺ channels (IK), which have been shown to be localized at the apical end of salivary gland cells (Almassy et al., 2012; Liu et al., 2007; Siefjediers et al., 2007). Importantly, no alteration in the localization of either IP₃R3 or the Ca²⁺-activated K⁺ channel was observed in the SG of Cav1^−/− mice compared with Cav1^+/+ mice (S3C). These findings demonstrate that association of AQP5 with lipid raft domains is disrupted in Cav1^−/− SMG which could account for the impairment of its targeting to the apical membrane region as well as its impaired microdomain association and attenuation of regulated trafficking following receptor-stimulation.

To conclusively establish the function of Cav1 in SG we performed in vivo gene delivery. Adenovirus encoding hCav1 or GFP was introduced into Cav1^−/− SMG by retrograde instillation via the secretory ducts. Efficient expression of introduced genes was verified by imaging SMG sections for GFP or Cav1 (supplementary material Fig. S3D,E). This exogenous expression of Cav1 resulted in formation of caveolae and caveolar vesicles (shown by arrows, Fig. 5A), whereas no such structures were detected in GFP expressing Cav1^−/− acini. Further, association of TRPC1 with STIM1 and Orai1 in response to pilocarpine-stimulation was restored in these cells (Fig. 5B) together with a recovery of TRPC1 channel activity and Ca²⁺ influx (Fig. 5C,D). The amplitude of the Tg-stimulated current (Fig. 5C) was comparable to that measured in Cav1^+/+ SMG acinar cells (Fig. 2C). Importantly, AQP5 targeting to the acinar cell apical membrane was also substantially recovered (indicated by arrows, Fig. 5E) which was correlated with a significant recovery of pilocarpine-stimulated saliva secretion in these mice compared to GFP expressing Cav1^−/− mice (Fig. 5F). Overall, our findings demonstrate that Cav1 has an essential role in the functional regulation of TRPC1-STIM1 as well as AQP5 channels. Together these represent the key determinants of salivary gland fluid secretion.

The findings presented above reveal that Cav1 is a critical component for SG function. We demonstrate that PM trafficking of TRPC1 as well as its interaction with STIM1 in membrane raft microdomains is determined by Cav1. Importantly, we have shown that Cav1 has an essential role in the assembly of functional TRPC1-STIM1 channels but not in the assembly of Orai1-STIM1 channels. While disruption of raft microdomains by MβCD perturbs both Orai1-STIM1 and TRPC1-STIM1 interactions, the deletion of Cav1 affects only TRPC1-STIM1 interaction. These data suggest that both channel assemblies are likely to involve lipid raft domains, with TRPC1-STIM1 channel assembly occurring in caveolar microdomains and Orai1-STIM1 association likely in non-caveolar raft microdomains. Although Cav1 was recently shown to affect Orai1 trafficking (Sathish et al., 2012; Yu et al., 2010), we did not find evidence of any alteration in STIM1-Orai1 interaction in Cav1^−/− cells. We suggest that Orai1-STIM1 channel or other Ca²⁺ channels, e.g. TRPs, or volume regulated cation channel (Liu et al., 2006; Liu et al., 2010), could account for the residual SOCE in Cav1^−/− cells. Clearly this residual influx component does not compensate for the significant loss of TRPC1 function or fluid-secretion in Cav1^−/− SG. We suggest that the decrease in SOCE and consequent loss of fluid-secretion are due to the specific disruption of TRPC1-STIM1 channels. Our data also indicate while the expression of Orai1 and STIM1 including key regulatory molecules involved in receptor-activated Ca²⁺ regulation in SG are not altered, the aberrant trafficking of TRPC1 in Cav1^−/− SG could decrease the stability of the channel. However, further studies are required to completely understand the mechanism underlying the role of Cav1 in regulating the levels of TRPC1 protein.

There is convincing evidence to demonstrate that STIM1 is recruited into specific cellular domains where the ER and PM are juxtaposed (Liou et al., 2007; Wu et al., 2006) where it anchors to the PM microdomains by associating with membrane lipids via its C-terminal domain (Calloway et al., 2011; Walsh et al., 2010). This anchoring is critical for its interaction and regulation of Orai1 and TRPC1 channels. Although the exact molecular components in the ER and PM that determine the assembly of these channels are not well established, there is sufficient evidence to suggest that association of STIM1 with lipid-rich PM microdomains, LRD, is enhanced upon ER store-depletion (Pani and Singh, 2009). Additionally, following activation of immune cells (which essentially lack the expression of Cav1 and caveolae), STIM1 localizes to and activates Orai1 at the immunological synapse (Lioudyno et al., 2008; Quintana et al., 2011) and cap-like membrane structures (Barr et al., 2009) which are lipid-rich PM signaling domains. These studies together with our present findings further suggest that both TRPC1-STIM1 and Orai1-STIM1 channel complexes are assembled at distinct membrane-raft microdomains. Thus specific PM lipid microdomains can contribute significantly to the regulation of cell function by providing a platform for channel assembly and spatio-temporal control of Ca²⁺ signals. This is evident in the exclusive regulation of NFAT by Orai1-mediated Ca²⁺ (Gwack et al., 2007) while Ca²⁺ entry via TRPC1 channels contributes to NFκB activation (Pani et al., 2009). Although a number of recently identified proteins (e.g. SARAF, Golli, POST, ERp57 and CRAC2A) have been suggested to regulate Orai1-STIM1 function (Fujii et al., 2012; Palty et al., 2012; Srikanth and Gwack, 2012), it is unclear how exactly these molecules contribute to the CRAC channel assembly and regulation.

A potentially important feature of membrane rafts is that they are highly dynamic, and exist in a constant state of flux. These mobile PM microdomains can coalesce into larger signaling compartments which could be stabilized by protein scaffolds and cytoskeletal interactions. We suggest that the PM microdomains respectively containing TRPC1 and Orai1 channel complexes are localized within close proximity to each other, since Orai1-mediated local Ca²⁺ entry is required to trigger PM-recruitment of TRPC1 (Cheng et al., 2011). Additionally, the spatial proximity of these distinct microdomains could potentially facilitate crosstalk between distinct signaling complexes and thus modulate the frequency and magnitude of intracellular Ca²⁺ signals. Nonetheless, based on the mislocalization of TRPC1 in Cav1^−/− SMG and loss of its interaction with STIM1 it is evident that Cav1 positions TRPC1 at precise PM regions enabling its functional activation by STIM1. These findings are consistent with our previous studies which showed that Cav1 acts as a PM scaffold for inactive TRPC1 and targets the channel to specific ER-PM junctional domains. Following store depletion, when TRPC1 interacts with STIM1 it dissociates from Cav1 (Brazer et al., 2003; Pani et al., 2009). Cav1-TRPC1 interaction is restored following intracellular Ca²⁺ store-refilling. We thus suggest the caveolar microdomain has a critical role in regulating the dynamic interactions between TRPC1 and STIM1 required for store-operated Ca²⁺ entry in SG acinar cells.

Salivary fluid secretion is under the direct control of parasympathetic stimulation of salivary epithelial cells via muscarinic receptors which leads to IP₃-mediated release of intracellular Ca²⁺ and activation of Ca²⁺ influx. The net increase in [Ca²⁺]_i leads to activation of ion channels such as the Ca²⁺-activated K⁺ and Cl⁻ channels as well as the Na/K/Cl co-transporter which together generate the osmotic gradient required to drive fluid-secretion from acinar cells (Ambudkar, 2000; Kaplan and Baum, 1993; Melvin et al., 2005; Putney, 1986). An important component of SG fluid-secretion is the AQP5 channels (Borgnia et al., 1999). We show that targeting of AQP5 to the apical membrane regions as well as its insertion to the surface membrane upon receptor-stimulation is abrogated in the absence of Cav1. To our knowledge this is the first in vivo demonstration identifying an obligatory role of Cav1 in SG fluid-secretion. It has been shown that muscarinic receptor activation of salivary cells, which results in an intracellular increase of [Ca²⁺]_i (Ambudkar, 2000; Putney, 1986), accounts for the regulated trafficking of AQP5 channels (Gresz et al., 2004; Ishikawa et al., 2005). Further, we have previously established that TRPC1-mediated Ca²⁺ entry is critical for agonist-stimulated saliva secretion (Liu et al., 2007; Singh et al., 2001). This suggests that TRPC1-mediated Ca²⁺ entry is critical for the surface expression of AQP5 in the apical membrane of SG acini. Disruption in membrane lipid domains could also account for the apparent change in secretory granule secretion as demonstrated by the increased accumulation of the granules in acinar cells from Cav1^−/− mice. This involvement of TRPC1 channels in regulated vesicular activity and secretory events is underscored by our finding that compared to Cav1^+/+ the Cav1^−/− SMGs have increased intracellular vesicles, and is in agreement with recent study demonstrating the direct role of TRPC1 in mast cell granule exocytosis (Cohen et al., 2012) and in the release of glutamate from astrocytes (Parpura et al., 2011). Notably, the in vivo expression of Cav1 in SMGs of Cav1^−/− mice was able to substantially recover the loss of agonist-stimulated Ca²⁺ entry and fluid-secretion. Thus, our findings establish that Cav1 is at the critical physiological nexus of receptor-activated Ca²⁺ entry and fluid-secretion wherein, Cav1 orchestrates agonist-stimulated Ca²⁺ entry via the TRPC1-STIM1 channels which could impact AQP5 trafficking and thereby salivary fluid-secretion. In addition, Cav1 can possibly have more direct effects on the targeting of AQP5 to the apical membrane as well as fusion of secretory granules with the apical membrane. We propose Cav1 is a critical, non-redundant and as of yet unrecognized molecular determinant in the regulation of salivary gland function.

Cav1^tm1Mls/J (Caveolin1 knockout, hereafter referred as Cav1^−/−) and the related control B6129SF2/J (Caveolin1 wild type, hereafter referred as Cav1^+/+) mice were obtained from Jackson Laboratories (Bar Harbor, ME) and were housed in a 12-hour light-dark cycle with ad libitum accesses to lab chow and water. Animals were maintained in accordance to the guidelines established by the University of North Dakota Institutional Animal Care and Use Committee and the National Institutes of Health. All animals used were between 8 and 12 weeks of age. Unless mentioned otherwise, all reagents used in the study were of molecular biology grade obtained from Sigma chemicals (St Louis. MO).

Human submandibular gland (HSG) cells were cultured in MEM medium supplemented with 10% FBS, penicillin (50 U/ml) and streptomycin (50 µg/ml) and cells were maintained at 37°C with 95% humidified air and 5% CO₂ and were passaged as described earlier (Pani et al., 2009). HSG cells, at about 70% confluency, were transfected with AQP5 expression plasmid using standard procedure (Invitrogen). For RNAi experiments, siRNA duplexes targeting the coding sequence of human Cav1 (Pani et al., 2009), TRPC1 (pool of four siRNA duplexes obtained from Dharmacon), and a non-targeting NT-siRNA (Qiagen, Valencia, CA) were respectively transfected using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA) in Opti-MEM medium as per supplier's instructions. Cells were typically used 48 hours post-transfection.

Respective SMGs were extracted and pulverized using a mortar and pestle over dry ice. Total RNA, from the powdered tissues, were extracted using TRIzol reagent, reverse transcribed, and PCR amplified using standard procedures as described earlier (Pani et al., 2006). A 5 µl aliquot of the amplified product was resolved in a 1.5% agarose gel and the EtBr stained bands were visualized using the Bio-Rad imager. Sequence of gene-specific primers used in the study will be made available on request.

Adenovirus-mediated gene delivery into SMGs was achieved by retrograde ductal instillation (Zheng et al., 2011). Cav1KO animals were first anesthetized with an intramuscular injection (i.m.), containing a 1∶1.5 mixture of Xylazine (20 mg/ml) and Ketamine (100 mg/ml), administered at 1 µl/g body-weight. Following sedation the orifice of the excretory ducts, at the floor of the oral cavity, was located under a surgical microscope (Leica MZ75), and custom-made capillary tubing was carefully inserted to each ductal opening. The capillary was secured along with the duct by applying krazy glue. To prevent salivation following cannulation, the animals were given atropine (0.5 mg/ml) by an i.m. injection at 1 µl/g body-weight. Thereafter, retrograde delivery of respective adenovirus (expressing either Cav1-AdCav1 of GFP-AdGFP, Vector Biolabs) to each SMG was done via the capillary. Each gland received 2.5×10⁸ pfu of respective adenovirus in a total volume of 50 µl. The capillaries were removed by a gentle tug and the animals were then placed on a temperature-regulated pad to recover. Following, recovery the animals were returned to the cages with abundant access to food and water. Saliva secretion experiments were typically performed 4–5 days following gene delivery.

To initiate neurotransmitter-induced saliva secretion, the animals were first anesthetized (as described above), and were injected with pilocarpine (0.5 mg/ml) at 1 µl/g body-weight. Whole saliva was collected into pre-weighed tubes using micro-hematocrit capillaries (Drummond Scientific, Broomall, PA). The secreted saliva was collected in every 2-minute interval for a total length of 10 minutes, and was quantified gravimetrically. Saliva osmolarity was measured using a Wescor 5500 vapor pressure Osmometer (Liu et al., 2007).

Respective SGs were extracted and fixed with 1.5% gluteraldehyde and 4% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.2) containing 5% sucrose. 1 mm tissue pieces were post-fixed in 1% OsO₄ for 1 hour. Tissues were subsequently en bloc stained with 1% uranyl acetate in 10% ethanol for 1 hour, dehydrated through a standard ethanol series and infiltrated overnight with LX112. Individual tissue blocks were embedded, polymerized and sectioned with Reichert Ultramicrotomes. Orientation thick sections were stained with Toluidine Blue and transmission electron micrograph (TEM) grids were stained with both uranyl acetate and Pb nitrate prior to examination on JEOL 1200 EX-11 with Gatan digital camera (Tsutsumi et al., 2008).

SMGs were extracted and fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2) containing 5% sucrose. Tissues were further placed for overnight in the same buffer containing 20% sucrose. Tissues were then embedded in Tissue-Tek OCT compound and 10–12 µm thick cryosections were obtained. Periodic acid-Schiff (Zheng et al., 2011) or hematoxylin and eosin (H&E) staining was performed on the sections using standard procedure (Sigma, St Louis, MO). Light images were obtained at 40× or 63×magnifications. For obtaining the relative number of acini in Cav1^+/+ and Cav1^−/− SMG, individual acini from several sections were counted manually in a blinded manner. Acini area was determined using ImageJ software (NIH). For fluorescent confocal imaging, SMG sections from respective animals were permeabilized at room temperature with 0.1% Triton X-100 in PBS (pH 7.4), blocked (10% donkey serum and 5% BSA in PBS), and probed overnight with respective primary antibodies in a hydrated chamber maintained at 4°C. Following incubation with primary antibodies the slides were washed and reacted with fluorophores-conjugated respective secondary antibodies for 1 hour at RT. Thereafter the slides were washed and coverslip mounted using vectashield hardest mounting media with DAPI (Vector Laboratories, CA). Images were acquired with 63× (1.4 NA) or 100× (1.45 NA) objectives using a confocal laser-scanning LSM 510 Meta microscope (Carl Zeiss, Thornwood, NY).

Isolation of detergent-resistant caveolar rafts was essentially performed as described earlier (Pani et al., 2008). Animals were either left untreated of stimulated with pilocarpine in vivo (for a 5-minute salivation period), and respective SMGs were extracted, washed with pre-chilled PBS (pH 7.4), triturated and homogenized on ice in pre-chilled TNE buffer (1% v/v Triton X-100, 25 mM Tris-HCl, 150 mM NaCl, and 5 mM EDTA, pH 7.5) supplemented with 1× protease and phosphatase inhibitors. Lysates were incubated on ice for 30 minutes and passed through a 28 gauge needle several times prior to a brief centrifugation step (1200 rpm, 10 minutes). The resulting supernatant was mixed with an equal volume of 80% sucrose, overlaid with a discontinuous sucrose gradient, and centrifuged at 34,000 rpm for 18 hours at 4°C as described earlier (Pani et al., 2008). A total of ten fractions were collected from the top of the tube and equal volume of individual fractions was analyzed by western blotting. To study agonist-induced protein translocations, the density gradients fractions corresponding to caveolar raft designated as buoyant fractions (BF, fractions 3–5) and non-raft/soluble fraction designated as heavy fractions (HF, fractions 8–10) were pooled, quantified, and equal amount of proteins were analyzed by western blotting. Alternatively, as described previously (Pani et al., 2008), detergent-resistant membrane rafts (R) and soluble (S) fractions were isolated from HSG cells following MβCD treatment and were analyzed by western blotting.

Co-immunoprecipitations and western analyses were carried out as described earlier (Pani et al., 2009; Pani et al., 2008). For in vivo stimulations, mice were anaesthetized and injected with Pilocarpine to salivate for 5 minutes. Following stimulation SMGs were isolated and snap frozen in liquid nitrogen and used for making protein lysates. The frozen SMGs were individually crushed using a pestle and mortar and lysed in pre-chilled RIPA buffer. Cleared lysates were quantified and equal amount of protein was used for immunoprecipitations with anti-STIM1 or anti-TRPC1 antibody, respectively. Additionally, whole tissue lysates of SMGs were analyzed to determine the relative expression of proteins (primarily involved in SG ion homeostasis) in Cav1^+/+ vs Cav1^−/− animals. Stimulations of HSG cells without or with MβCD treatments and immunoprecipitations thereafter were essentially preformed as described earlier (Pani et al., 2008). Band intensity analysis of proteins was performed by using Quantity One 4.6.5 1D-analysis software (Bio-Rad Richmond, CA). The calculated densitometry values for individual experiments are expressed as arbitrary units. A list of antibodies used in this study is provided as supplementary material Table S1. Additionally, TRPC1 (Liu et al., 2007) and NKCC1 (Nezu et al., 2009) were used as described earlier.

Labeling of cell surface proteins by biotinylation was essentially done as described earlier (Pani et al., 2009), with minor modifications. Single acinar cells were prepared from SMGs of respective animals and were stimulated with pilocarpine (5 minutes at 37°C) in HBSS buffer containing Ca²⁺ and Mg²⁺ (Invitrogen) supplemented with 1% Ig-free BSA (pH 7.4). Following stimulation the cells were washed with HBSS (pH 8.0 adjusted by NaOH), and labeled with 2 mg/ml Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL) at 4°C for 20 minutes. Additionally, AQP5-expressing HSG cells were stimulated in serum free MEM media with Tg (2 µM for 5 minutes at 37°C) or CCh (100 µM for 5 minutes at 37°C), without or with SKF96365 pretreatment (20 µM for 15 minutes at 37°C) or under TRPC1 or Cav1 silenced conditions. Following stimulations, cells were washed once with 1× PBS (pH 8.0) and incubated at 4°C, with gentle rocking, for 20 minutes with 1.5 mg/ml Sulfo-NHS-LC-Biotin made in 1× PBS (pH 8.0). Following biotin labeling HSG or SMG cells were washed with respective buffer containing 100 mM glycine, and lysed with RIPA buffer. Lysate were cleared by centrifugation (13,000 rpm for 10 minutes at 4°C) and biotinylated proteins were pulled down by incubating with NeutrAvidin-linked beads (Pierce, Rockford, IL) for 2 hours at 4°C with an end to end rotation. The beads were then washed and the bound proteins were eluted by boiling in 2× SDS dye. The eluted material was analyzed by western blotting.

For single cell preparation form SMGs, freshly extracted SGs were triturated, and digested in SES buffer (pH 7.4) containing 0.02% soybean trypsin inhibitor and 0.1% BSA and collagenase P (2.5 mg/8 ml of buffer) for 15–20 minutes at 37°C (Liu et al., 2007; Pani et al., 2006). Following digestion the cells suspension was washed twice with SES buffer, re-suspended in SES buffer containing 2 µM Fura-2AM and placed in collagen and poly-d-lysine coated 35 mm glass-bottomed culture dishes (MatTek, Ashland, MA) for 45–60 minutes at 37°C. Prior to performing Ca²⁺ measurements the culture dishes were washed with and placed in Ca²⁺-free SES buffer. Fluorescence measurements were performed by imaging the Fura-2AM-loaded acinar cells using the Olympus IX50 microscope, with the excitation light provided by a Polychrome 4 (TILL Photonics) (Liu et al., 2007; Pani et al., 2006). Images were acquired using a Photometrics CoolSNAP HQ camera (Photometrics) and the MetaFluor software (Molecular Devices). Fluorescence traces shown represent [Ca²⁺]_i values that are averages from at least 30–40 acinar cells and are a representative of results obtained in at least 3–4 individual experiments.

All electrophysiological experiments were performed using previous protocol (Liu et al., 2007; Singh et al., 2001). Coverslips with freshly isolated SMG acinar cells were transferred to the recording chamber and perfused continually, through a custom-designed gravity-driven speed-controlled system at a rate of 5 ml/minute, with an external Ringer's solution of the following composition (mM): NaCl, 145; KCl, 5; MgCl₂, 1; CaCl₂, 1; HEPES, 10; glucose, 10; pH 7.4 (NaOH). The patch pipette had resistances between 3 and 5 mΩ after filling with the standard intracellular solution that contained the following (mM): cesium methane sulfonate, 145; NaCl, 8; MgCl₂, 10; HEPES, 10; EGTA, 10; pH 7.2 (CsOH). Osmolarity for all solutions was adjusted with D-mannitol to 305±5 mmol/kg using a vapor pressure osmometer (Wescor). Patch clamp experiments were performed in the tight seal whole-cell configuration at room temperature (22–25°C) using an Axopatch 200B amplifier (Axon Instruments). Voltage ramps ranging from −90 to 90 mV over a period of 1 second were imposed every 4 seconds from a holding potential of 0 mV and digitized at a rate of 1 kHz. A liquid-junction potential of less than 8 mV was not corrected, and capacitative currents and series resistance were determined and minimized. For analysis, the first ramp was used for leak subtraction for the subsequent current records.

Data analysis was performed using MicroSoft Excel or Origin 7.0 (OriginLab). Statistical comparisons were made using one-way ANOVA. Experimental values are expressed as means ± s.d. Differences in the mean values were considered to be significant at P-values <0.05.

We acknowledge The Edward C. Carlson Imaging and Image Analysis core facility, and thank Sarah Abrahamson, John A. Watt and his lab members for their excellent technical advice and helpful discussions. We thank R. James Turner and Ted Begenisich for providing anti-NKCC1 and calcium activated K+ channel (IK) antibodies.