Interaction Between Pannexin 1 and Caveolin-1 in Smooth Muscle Can Regulate Blood Pressure

Objective— Sympathetic nerve innervation of vascular smooth muscle cells (VSMCs) is a major regulator of arteriolar vasoconstriction, vascular resistance, and blood pressure. Importantly, &agr;–adrenergic receptor stimulation, which uniquely couples with Panx1 (pannexin 1) channel-mediated ATP release in resistance arteries, also requires localization to membrane caveolae. Here, we test whether localization of Panx1 to Cav1 (caveolin-1) promotes channel function (stimulus-dependent ATP release and adrenergic vasoconstriction) and is important for blood pressure homeostasis. Approach and Results— We use in vitro VSMC culture models, ex vivo resistance arteries, and a novel inducible VSMC-specific Cav1 knockout mouse to probe interactions between Panx1 and Cav1. We report that Panx1 and Cav1 colocalized on the VSMC plasma membrane of resistance arteries near sympathetic nerves in an adrenergic stimulus-dependent manner. Genetic deletion of Cav1 significantly blunts adrenergic-stimulated ATP release and vasoconstriction, with no direct influence on endothelium-dependent vasodilation or cardiac function. A significant reduction in mean arterial pressure (total=4 mm Hg; night=7 mm Hg) occurred in mice deficient for VSMC Cav1. These animals were resistant to further blood pressure lowering using a Panx1 peptide inhibitor Px1IL2P, which targets an intracellular loop region necessary for channel function. Conclusions— Translocalization of Panx1 to Cav1-enriched caveolae in VSMCs augments the release of purinergic stimuli necessary for proper adrenergic-mediated vasoconstriction and blood pressure homeostasis.


INTRODUCTION
In the peripheral circulatory system, blood pressure (BP) homeostasis is largely regulated by the contractile state of the smooth muscle cells (VSMCs) in the wall of resistance arteries. Resistance arteries are small diameter (≤200 μm) arterioles that are composed of an intimal endothelial layer, a tunica media consisting of one to two VSMC layers 1 , and are functionally defined by the ability to respond to changes in intraluminal pressure to control vascular resistance and blood flow. Sympathetic nerve innervation of VSMCs is a major regulatory pathway that induces VSMC constriction, thus altering arteriolar lumen diameter and increasing the amount of vascular resistance 2,3 . While much is already known about the induction of rapid (purinergic) and potentiated (adrenergic) sympathetic nerve-derived stimuli on function of resistance arteries 4 , less is known about the identity of downstream VSMC signaling molecules that subsequently coordinate VSMC-derived stimuli in the arteriolar wall. Signaling through α-adrenergic receptors (α-ARs) and activation of plasma membrane-associated signaling molecules may be compartmentalized by caveolae near areas of sympathetic innervation 2,5-8 .
Caveolae are specialized plasma membrane domains that play an important role in intracellular signaling, cellular transport, and differentiation [9][10][11] . Caveolae are defined by their hallmark plasma membrane morphology, consisting of 80-100nm-wide membrane invaginations, and by their composition of oligomeric coat-forming proteins called caveolins 12 . Of the three caveolin isoforms, caveolin-1 is essential for caveolae formation and function [13][14][15] . Caveolin-1 has been shown to regulate vasoconstriction responses in small arterioles and to influence BP homeostasis in animal models 13,[16][17][18] . Importantly, caveolin-1 is expressed in VSMCs of small arteries and acts as a membrane-scaffold protein for both α-ARs and downstream G-protein dependent vasoconstriction signaling molecules 19,20 , suggestive of a role in adrenergic-mediated vasoconstriction.
Recently, our group and others [21][22][23][24] have elucidated in mouse and humans an α-AR signaling axis that activates Pannexin 1 channels. Pannexins are a family of transmembrane channel-forming glycoproteins that have emerged as the physiological conduit for controlled ATP release from vascular and non-vascular cell types [25][26][27] . We have previously reported that Panx1 expression is polarized within the vascular tree, with high expression levels in VSMCs of resistance arteries (e.g. mesenteric, cremasteric, thoracodorsal, and coronary), but is not present in large conduit vessels such as the femoral and carotid arteries, and the aorta 28 . This expression pattern suggests a unique role for Panx1 in regulating vascular resistance. Using Panx1 pharmacological inhibitors and inducible VSMC-specific Panx1 knockout mice, we have further demonstrated that Panx1-mediated ATP release and vasoconstriction are uniquely coupled with α-adrenergic stimulation and are crucial for maintaining BP homeostasis 21,22 . Multiple groups have confirmed the activation of Panx1 channels by α-adrenergic stimulation [29][30][31] , thus demonstrating ATP release through Panx1 as a significant physiological pathway for integrating and coordinating VSMC-derived constriction responses 4,32 .
In the present study, we hypothesized that caveolin-1 acts as a molecular scaffold that concentrates VSMC Panx1 to areas important for sympathetic nerve innervation, thus supporting α-adrenergic vasoconstriction and BP homeostasis. In response to the αadrenergic agonist phenylephrine we observed a novel interaction and co-localization of caveolin-1 and Panx1 to regions of innervation at the VSMC plasma membrane. To investigate the functional role of caveolin-1 during α-adrenergic mediated responses, we generated an inducible, VSMC-specific caveolin-1 knockout mouse model. We show that VSMC-derived caveolin-1 is required for α-adrenergic stimulated ATP release and adrenergic vasoconstriction. Deletion of VSMC caveolin-1 results in a significant reduction in mean arterial BP, particularly during the nocturnal (active) period when sympathetic drive is high. Furthermore, we suggest that caveolin-1 mediated BP effects are regulated through the Panx1 intracellular loop regulatory domain, previously identified to be indispensable for α-adrenergic vasoconstriction 22 .

Data and Materials
All data will be made available upon request by the corresponding authors at the University of Virginia.

Animals
All animals were cared for under the provisions of the University of Virginia Animal Care and Use Committee and the National Institute of Health guidelines for the care and use of laboratory animals. Male C57BL/6 mice between 10-15 weeks of age were purchased from Taconic. Male smooth muscle myosin heavy chain-Cre recombinase modified estrogen receptor binding domain (SMMHC-CreER T2 ) modified mice, a kind gift from S. Offermanns 33 , were used for experimentation due to the restrictive presence of Cre recombinase on the Y chromosome. Aortae from Connexin 43 globally deficient mice (Cx43 −/− ) were harvested at birth. Mice harboring lox-P recombination sites for caveolin-1 (Caveolin-1 fl/fl ) were generated as previously described 34 . SMMHC-CreER T2 mice were mated with Caveolin-1 fl/fl mice to specifically delete caveolin-1 from vascular smooth muscle cells. Induction of Cre-mediated deletion was performed at 6 weeks of age via ten daily intraperitoneal injections (100 μl) of tamoxifen (1 mg/kg) to generate caveolin-1 null animals (SMMHC-CreER T2+ /Cav1 Δ/Δ ) or ten daily 100 μl injections of peanut oil (vehicle control) to generate control animals (SMMHC-CreER T2+ /Cav1 fl/fl ). All animal experiments were performed at the least 14-days from the final injection with tamoxifen and/or peanut oil since the Cre recombinase is located on the Y. Mice lacking the Cre recombinase allele were also used as tamoxifen controls. Please see the Major Resources Table in Supplemental Material for detail.

Cell Culture
Primary human vascular coronary smooth muscle cells (VSMCs) were purchased from Lonza (Cat# CC-2583). All cells were maintained under standard cell culture conditions (5% CO 2 at 37°C) in smooth muscle growth media (Lonza; Cat# CC-3181) supplemented with smooth muscle growth factors (Lonza; Cat# CC-3182) and 10% fetal bovine serum (FBS) (Lonza; Cat# CC-4102D). Cells were used at 8 passages or fewer for in vitro experiments.

Ultrastructure electron microscopy
Mouse arteries were processed for ultrastructure TEM as previously described 38 . Images were obtained using a Joel 1230 transmission electron microscope at the Advanced Microscopy Core at the University of Virginia.

Proximity Ligation Assay (PLA) and Immunofluorescence
Thoracodorsal arteries (TDA) were isolated as previously described 39 , incubated in Krebs-HEPES physiological saline buffer, and treated with phenylephrine (20 μmol/L) in a single well of a 96-well dish. TDAs were then placed in a 1.5 mL Eppendorf tube, fixed in 4% paraformaldehyde, and subjected to en face proximity ligation assay using the Duolink in situ PLA detection kit (Sigma) as previously described by us 40 . Sympathetic nerves were labeled using anti-mouse tyrosine hydroxylase antibody (Abcam #ab112; 1:250 dilution) and visualized using an Alexa Fluor 568 secondary antibody (Life Technologies #A-21099; 1:400 dilution). Primary antibodies for PLA labeling included anti-mouse Caveolin-1 (BD Biosciences# 610406, clone 2297; 1:400 dilution) and anti-mouse Panx1 CT395 (characterized by Penuela et al, 41 1:300 dilution). PLA detection was performed according to manufacturer's protocol. Immunofluorescence staining on aorta and TDAs was performed as previously described 28 . Primary antibodies for immunofluorescence included anti-rabbit Cx43 antibody (Sigma #C6219; 1:300 dilution) and vesicular nucleotide transport protein (anti-mouse VNUT; a kind gift from Dr. Chen Li, 1:200 dilution). All images were acquired using an Olympus Fluoview 1000 confocal microscope. PLA punctate spots were counted per 100um 2 cell area. Caveolin-1 deletion was quantified using batch-processed tissue and threshold generated images. The relative fluorescence intensity within the smooth muscle cell layer (demarcated by co-association with Acta2 staining, and between the boundary lines of the IEL and adipose tissue) was measured using ImageJ 42 and normalized to Acta2 positive area. Co-staining for smooth muscle cells (Acta2; Sigma #A2547, 1:500 dilution) and endothelial cells (PECAM-1; Santa Cruz #sc28188, 1:400 dilution) was performed. Data are presented as mean ± SEM. A students t-test was performed for statistical significance. *p < 0.05.

Live cell imaging
Confocal imaging was performed with a Leica TCS SP8 confocal microscope. Human VSMCs were cultured as indicated above and plated on 100μg/mL poly-D-lysine (PDL) on cover glass. Cells were transfected with plasmids encoding Panx1-RFP 43 and Caveolin-1-GFP using jetPRIME (Polyplus transfection/VWR) according to the manufacturer's protocol. Image acquisition and co-distribution analyses were performed double-blinded to treatment conditions with identical imaging parameters. For live imaging, baseline images were collected at 30 sec intervals for 2 min using a 20X (0.7 NA) objective. Importantly, the large cellular size (range of lengths) permitted imaging of only one cell per field of view. Treatment with phenylephrine (100 μmol/L; Sigma-Aldrich), ATP (500μmol/L; Sigma-Aldrich) or vehicle control (water) was performed by removing half the volume of control media and replacing it with the same volume of media containing 2X agonist. Images were collected at 30 sec intervals up to 5 min. The z-section containing the largest cellular area was selected for Mander's coefficient analysis of Panx1-Cav1 co-distribution using the JACoP plugin in FIJI 44 . All post-treatment data were normalized to the average obtained at baseline. Data were collected from N=5-11 cells per experimental condition and analyzed using a two-way ANOVA for time and treatment

Membrane fractionation and isolation of caveolin-1 enriched membrane domains
Human VSMCs were grown to confluence and incubated in media containing 0.2% FBS for 48 hr prior to use. VSMCs were washed with PBS and re-equilibrated for 10 min in Krebs buffer (mmol/L: 118.4 NaCl, 4.7 KCl, 1.2 MgSO 4 , 4 NaHCO 3 , 1.2 KH 2 PO 4 , 10 Hepes, 6 Glucose) containing 2 mmol/L CaCl 2 . VSMCs were treated with 100μmol/L phenylephrine or vehicle control, scraped with a cell scraper and lysed in ice-cold detergent-free lysis buffer (500 mmol/L Na 2 CO 3 , 50 mmol/L NaF, 2 mmol/L Na 3 VO 4 , pH 11, supplemented with 1 mg/mL of protease inhibitor cocktail (Sigma) and 1 mg/mL P2 and P3 phosphatase inhibitor cocktails (Sigma)). Lysates were homogenized using a dounce homogenizer (10 strokes) and sonication (25 pulses for 1 sec) on ice. Lysates were either fractioned using differential centrifugation (40,000 rpm; 1hr Beckman XL80 ultracentrifuge with Sw55Ti rotor) or using centrifugation across a sucrose gradient. To create a sucrose gradient, sucrose solutions were mixed in MBS-sodium carbonate buffer (25 mmol/L MES, 0.5 M NaCl, 250 mmol/L Na 2 CO 3 ) to 85%, 30%, and 5% by mass by adding 42.5 g, 15 g, or 2.5 g of sucrose, respectively, to 50 mL of MBS. Lysates were mixed with equal volume of 85% sucrose solution to create 42.5% layer. 1.5 mL of 42.5% layer was added to the bottom of a Sw55Ti ultracentrifuge tube, after which a 5-42.5% discontinuous sucrose gradient was formed by adding, dropwise on top of previous layers, 1 mL of 30% sucrose solution followed by 1 mL of 5% sucrose solution. Gradients were centrifuged at 42,000 rpm for 18 hr in a Beckman XL80 ultracentrifuge with Sw55Ti rotor. Ten fractions of 350 μL each were removed starting from the top of the gradient and analyzed using western blot with antibodies for rabbit antihuman Pannexin 1 (characterized by Penuela et al. 45 1:1000 dilution) and mouse anticaveolin-1 (BD Biosciences #610406; 1:1000 dilution). Co-immunoprecipitation was performed using either Pannexin1 antibodies (1:50 dilution) in conjunction with anti-rabbit IgG Dynabeads (Invitrogen) respectively. To perform co-immunoprecipitations, caveolin-1enriched fractions (4)(5) and non-enriched fractions (7)(8) were combined and total protein from each was measured by BCA assay. Equal amounts of protein from each pair of fractions was used for co-immunoprecipitation as described above. Five independent experiments were performed. The ratio of caveolin-1 signal was normalized to the amount of immunoprecipitated Pannexin 1. Data are represented as mean ±SEM. A students t-test was performed for statistical significance. *p < 0.05.

Pressure myography
Pressure myography was performed on TDAs as previously described 46 . Briefly, mice were sacrificed using CO 2 asphyxia. TDAs were microdissected, cannulated on glass pipettes in a temperature-controlled pressure arteriography chamber, and pressurized to 80mmHg. After a 30-min equilibration period in Krebs-HEPES with 2mmol/L Ca 2+ , vessels were treated with cumulative doses of phenylephrine (PE, 10 −10 -10 −3 mol/L) applied to the bath. The luminal diameter was analyzed using digital calipers in the DMT vessel acquisition software (Danish Myo Technology). Smooth muscle cell viability was assessed using serotonin (1 μmol/L) and KCl (30 mmol/L). Endothelial-dependent vasodilation was measured using cumulative doses of acetylcholine (10 −11 -10 −2 mol/L) as previously described 46 . A two-way analysis of variance (ANOVA) with Bonferonni post-hoc test was performed for multiple comparisons.
Concentration-effect curves were fitted to the data using four-parameter, non-linear regression curve fitting using GraphPad (version 7).

ATP measurements
For the measurement of extracellular ATP, intact TDAs of equal length were placed in individual wells of a 96-well plate in Krebs-Hepes physiologic solution for 15 min. The ectonucleotidase inhibitor ARL 67156 trisodium (Tocris; 100μmol/L) was added 30 min prior to treatment with contractile agonists as previously described 22 . Vasoconstrictor compounds were added to the incubation media for 5 min to allow ATP accumulation: Phenylephrine (PE; 20μmol/L), Norepinephrine (NE; 20 μmol/L), Serotonin (5-HT; 40nmol/L), and Endothelin-1 (ET-1; 40nmol/L) (all purchased from Sigma). Following stimulation, the media surrounding the vessel was collected and immediately placed into pre-chilled 1.5 mL Eppendorf tubes on ice. All samples were centrifuged at 10,000 x g for 5 min. For intracellular ATP measurements, TDAs were microdissected, cleaned of adventitia, and cut into equal 10.5mm vessel segments. Segments were individually lysed in ATP lysis buffer according to manufacturer's protocol, spun at 10,000 x g for 1 min, and samples collected. ATP concentration in the incubation media was quantified using the ATP bioluminescence assay kit HSII (Roche) using a FluoStar Omega plate reader luminometer. Extracellular ATP measurements for each sample were tested in triplicate and calculated using an ATP standard curve for all experiments. Intracellular ATP measurements were measured from three vessel segments (one TDA in triplicate). Data are presented as % change in ATP release from baseline (unstimulated) or as the concentration of ATP in the media compared to control samples. One-way ANOVA with Tukey's test was performed for statistical significance of extracellular ATP. A Kruskal-Wallis (one-way ANOVA on ranks) with Dunn's post-hoc test performed for intracellular ATP. Significance denoted as *p < 0.05.

Blood pressure telemetry
Blood pressure was measured using telemetry equipment as previously described 22 . Briefly, telemeters (Data Sciences International [DSI]) were implanted in C57BL/6 or SMMHC-CreER T2+ /Cav1 fl/fl (SMC-Cav1 fl/fl ) mice. Under isoflurane anesthesia, the catheter of a single telemetry unit (TA11PA-C10, DSI) was implanted in the left carotid artery and the transmitter placed in a subcutaneous pouch along the right flank of the mouse. After implantation surgery, mice were allowed to recover for 7 days to re-establish normal circadian rhythms and blood pressure. For experiments using inducible Cre recombinase, mouse blood pressure baselines were continuously recorded using Dataquest A.R.T. 20 software (DSI) for 5 days after normal recovery and before starting intraperitoneal tamoxifen injections or vehicle control (peanut oil) for 10 days. Blood pressure was recorded for an additional 5 days starting 24 hr after the last tamoxifen injection. Change in MAP (ΔMAP) was calculated by subtracting the average MAP measured before tamoxifen injections to the MAP after tamoxifen injections. Diurnal (inactive period) MAP was measured during animal's light cycle: 6:00 a.m. to 5:59 p.m., and nocturnal (active period) MAP was measured during the animal's dark cycle: 6:00 p.m. to 5:59 a.m. MAPs before and after tamoxifen injections were compared with a Wilcoxon test (nonparametric paired t-test). C57BL/6 mice similarly received intraperitoneal injections of tamoxifen or vehicle control, and basal blood pressure was measured as for transgenic animals. For assessment of the blood pressure effects of the PxIL2P peptide inhibitor (formerly referred to as PxIL2P1 peptide 22 ), animals were intraperitoneally injected with saline vehicle control or peptide (20 mg/kg in a volume not exceeding 100 μL). Blood pressure was recorded for 2 hr after injection, and the MAP data was averaged and compared to the basal blood pressure. Change in MAP (ΔMAP) was calculated by subtracting the baseline MAP 30 min before injection from the MAP measured during the final 30 min of the 2 hr treatment period. Data represent mean ±SEM. Two-way ANOVA with Tukey post-hoc test was performed for assessment of MAP before and after tamoxifen/vehicle control induction and for mice treated with scrambled peptide or PxIL2P peptide inhibitor.

Cardiac magnetic resonance imaging and histology
All MRI animal studies were performed under protocols that comply with the Guide for the Care and Use of Laboratory Animals (NIH publication no. 85-23, Revised 1996) and were approved by the Animal Care and Use Committee at our institution (ACUC, UVA). Mice were positioned supine in the scanner and body temperature was maintained at 36 ± 0.5°C using thermostatic circulating water. Anesthesia used was 1.25% isoflurane in O 2 inhaled through a nose cone during imaging. A 30 mm-diameter cylindrical birdcage RF coil (Bruker) with an active length of 70 mm was used, and heart rate, respiration, and temperature were monitored during imaging using a fiber optic, MR-compatible system (Small Animal Imaging Inc., Stony Brook, NY). MRI was performed on a 7 Tesla (T) Clinscan system (Bruker, Ettlingen, Germany) equipped with actively shielded gradients with a full strength of 650 mT/m and a slew rate of 6666 mT/m/ms. Baseline LV structure and function were assessed 47 . Six short-axis slices were acquired from base to apex, with slice thickness equal to 1mm, in-plane spatial resolution of 0.2 × 0.2 mm 2 , and temporal resolution of 8-12 ms. Baseline ejection fraction (EF), end-diastolic volume (EDV), endsystolic volume (ESV), myocardial mass, wall thickness, and wall thickening were measured from the cine images using the freely available software Segment version 2.0 R5292 (http:// segment.heiberg.se). EDV and ESV were then indexed to body mass (EDVI and ESVI, respectively). Mass to volume ratio (MVR) was calculated as the ratio of myocardial mass to EDV.

Statistics
All data were analyzed using GraphPad Prism v5.0 for live cell image analysis or v7.0 software for all other analyses. Briefly, D'Agostino-Pearson tests were used to determine normality. Brown-Forsthe/Barlett's tests were used to determine equal variance for ANOVA and F-test was used to determine equal variance for t-test in GraphPad Prism v7.0 software. Data that passed normality tests and equal variance tests were analyzed by t-test for two groups or ANOVA for three or more groups. Data that were not normally distributed were analyzed by Kruskal-Wallis test (three or more groups). Post-hoc analysis for multiple comparisons were selected as appropriate to test for statistical significance; * p < 0.05, ** p < 0.01, ***p<0.001. Results are presented as mean ± SEM.

RESULTS
Caveolae are specialized plasma membrane domains that facilitate interactions between signaling proteins (i.e. Gq, PLC, Src, etc 12,48,49 ) and membrane receptors (i.e. α-adrenergic receptors 49 ). These structures are found in endothelium and smooth muscle; however, their signaling functions are much less studied in smooth muscle. In electron micrographs, we consistently observe caveolae contained in arteriolar smooth muscle in proximity to sympathetic nerves (Supplemental Figure I). For this reason, we hypothesized that the caveolin-1 protein, the main component to caveolae, may associate with Panx1, which is activated in response to adrenergic stimulation.
In initial experiments, we performed live-cell confocal microscopy using in vitro human VSMC culture systems to measure the distribution and co-localization of exogenously expressed RFP-tagged Panx1 and GFP-tagged caveolin-1 following phenylephrine stimulation ( Figure 1A). Baseline fluorescence measurements were recorded for 2 min prior to stimulation with phenylephrine (100 μmol/L), vehicle control, or high concentration ATP (500 μmol/L) to promote Panx1 internalization. A strong and significant co-distribution was observed between caveolin-1 and Panx1 at 30 seconds of phenylephrine stimulation at the plasma membrane ( Figure 1A-B), which persisted above control fluorescence, although not to statistically significant levels ( Figure 1B-C). Conversely, treatment with highconcentration ATP caused a significant and continuous reduction in co-localized signal, consistent with internalization and loss of Panx1 on the cell surface 50,51 that is independent of caveolin-1 mediated endocytosis 43 . Vehicle control-treated cells showed no changes in fluorescence co-localization from baseline.
To confirm observations from live-cell imaging experiments, we next performed in vitro cell fractionation and co-immunoprecipitation assays using in vitro VSMC culture systems to probe for endogenous interactions between Panx1 and caveolin-1 at the plasma membrane. We found that caveolin-1 and Panx1 localize to membrane-associated fractions ( Figure 1D) and specifically overlap in caveolin-1-enriched sucrose-gradient fractions, suggesting partial localization of Panx1 to lipid microdomains containing caveolin-1 ( Figure 1E). To determine if caveolin-1 and Panx1 interact following adrenergic stimulation, we acutely treated VSMCs with phenylephrine (20 μmol/L). Fractions enriched or deficient in caveolin-1 were isolated and Panx1 was precipitated using a protein specific antibody 45 . We found that caveolin-1 from enriched fractions significantly co-precipitated with Panx1 after adrenergic stimulation ( Figure 1F), suggesting that plasma membrane complexes containing caveolin-1 and Panx1 form following adrenergic stimulation and may act to facilitate channel function.
Panx1-mediated ATP release and subsequent vasoconstriction are specifically mediated through α-AR activation 21,22,[29][30][31] . Based on our initial observations and the in vitro biochemical findings herein ( Figure 1F), we first tested if caveolin-1 and Panx1 similarly interact in VSMCs of ex-vivo isolated mouse resistance arteries. We performed proximity ligation assays (PLA) between Panx1 and caveolin-1 on TDAs and assessed the focal plane where sympathetic nerves innervate VSMC (Supplemental Figure II). Sympathetic nerves were specifically labeled with tyrosine hydroxylase. In control experiments using PLA secondary antibodies alone or IgG controls, we could not detect PLA punctate signals (red puncta indicate protein associations when PLA probes <40nm in apposition) -only sympathetic nerves could be viewed (Figure 2A-B). Next, we performed PLA for caveolin-1 and Panx1. Here we observed relatively few positive red punctate signals under control conditions ( Figure 2C); however, following acute (1 min) phenylephrine stimulation (20 μmol/L), we observed an induction of PLA signal, which predominantly localized at the VSMC plasma membrane near areas of sympathetic nerve innervation ( Figure 2D-E). As a control, TDA smooth muscle and sympathetic nerves were also analyzed for Cx43 expression, but it was not detected (Supplemental Figure III). The neuronal vesicular nucleotide transporter (VNUT) was also analyzed, but was only observed in sympathetic nerves, and not VSMCs as anticipated (Supplemental Figure III). These data demonstrate the formation of potential signaling microdomains where caveolin-1 and Panx1 are recruited together following adrenergic stimulation.
To investigate the functional and physiological role of VSMC caveolin-1 during α-AR vasoconstriction, we generated an inducible, VSMC-specific caveolin-1 knockout mouse model (SMC-Cav1 fl/fl ), which upon induction of Cre recombinase deletes exon 2 of caveolin-1 (SMC-Cav1 Δ/Δ ) ( Figure 3A-B). Caveolin-1 deletion was specific for VSMCs of resistance arteries, not affecting caveolin-1 expression in CD31 positive endothelial cells ( Figure 3C-D). Due to the well-established contribution of Panx1-mediated ATP release during adrenergic stimulation 21,22,29,52 , we measured vasoconstrictor-dependent ATP release from isolated resistance arteries. Using a luminescence-based assay, we observed a caveolin-1 dependent response, whereby ATP released following adrenergic stimulation (norepinephrine 20 μmol/L or phenylephrine 20 μmol/L) was significantly reduced in SMC-Cav1 Δ/Δ mice compared with controls ( Figure 3E). No significant effect on ATP release was observed in any of the genotypes tested in response to agonists for other potent vasoconstriction pathways (e.g., ET-1 or 5-HT) ( Figure 3E). Intracellular ATP concentration was also unchanged in any of the genotypes tested (Supplemental Figure IV).
In line with our ATP findings, we reasoned that blunted adrenergic-mediated ATP release in SMC-Cav1 Δ/Δ mice would concomitantly impair vascular responses to phenylephrine. To directly measure vasoconstriction responses in ex vivo resistance arteries, we assessed vascular responses to increasing concentrations of phenylephrine using pressure myography. SMC-Cav1 Δ/Δ mice exhibited significant reductions in phenylephrine-stimulated vasoconstriction (red line) compared with control animals (black line) ( Figure 4A). Unlike global caveolin-1 knockout mice, which exhibit reduced expression of endothelial cell caveolin-1 16,53 , no significant differences were observed in endothelium-dependent vasodilation (acetylcholine responses) in SMC-Cav1 Δ/Δ mice ( Figure 4B). Thus, VSMC-specific deletion of caveolin-1 impairs adrenergic vasoconstriction without altering endothelial-mediated responses.
Previously we have shown that VSMC-specific deletion of Panx1 channels results in significantly reduced BP due to blunted adrenergic-stimulated ATP release and vasoconstriction. Thus, we used telemetry to test whether SMC-Cav1 Δ/Δ mice exhibited a similar BP phenotype. BP was assessed in individual animals before (baseline) and after induction of caveolin-1 deletion. A significant reduction in 24-hour mean arterial pressure (total MAP= −3.8 mmHg) was observed only after tamoxifen injection in SMC-Cav1 Δ/Δ mice ( Figure 5A-B). Moreover, a significant and greater BP reduction (ΔMAP= −6.8 mmHg) was observed during the active period, when sympathetic drive to resistance arteries is higher ( Figure 5C) compared to the inactive period ( Figure 5D). No significant difference in baseline MAP was observed in any other tested genotype before induction with tamoxifen, vehicle control, or saline control ( Figure 5C; Supplemental Figure V). There was also no change in plasma renin concentration after caveolin-1 deletion, which was tested as a metric of altered basal renal-vasculature function (Supplemental Figure V).
To ensure that BP reductions in SMC-Cav1 Δ/Δ mice were not influenced by changes in cardiac function, we performed functional MRI analysis on SMC-Cav1 Δ/Δ and control animals ( Figure 6). No significant changes in cardiac function were observed in any of the tested genotypes (Table 1). This includes cardiovascular changes due to heart rate, stroke volume, cardiac output, left ventricular mass, or left ventricular wall thickness-all of which can directly influence MAP. Lastly, mice lacking the Cre allele, but maintaining the loxP genotype and injected with tamoxifen, showed no changes in expression of caveolin-1 in arteries, phenylephrine dose-responses, or MRIs (Supplemental Figure VI; Supplemental Table I).
To functionally assess whether BP reductions involving VSMC caveolin-1 are mediated through a Panx1-dependent pathway, we acutely treated animals with the Panx1 intracellular loop mimetic peptide-inhibitor PxIL2P (20 mg/kg) or scrambled control peptide (20 mg/kg). We have previously shown that PxIL2P significantly blunts phenylephrine-stimulated Panx1channel currents, ATP release, vasoconstriction, and MAP 22 . Following acute injection of PxIL2P, significant BP reduction was restricted to vehicle treated SMC-Cav1 fl/fl (vehicle treated: ΔMAP= −8.83 mmHg) and C57BL/6 (vehicle treated: ΔMAP= −12.26 mmHg; tamoxifen; ΔMAP= −9.81 mmHg) control animals-both of which contain VSMC caveolin-1. Conversely, no significant changes in MAP were observed in SMC-Cav1 Δ/Δ animals, which are deficient in VSMC caveolin-1 (tamoxifen: ΔMAP= 0.73 mmHg; Figure  7). Administration of a scrambled PxIL2P peptide of the same amino acid composition did not influence BP in any genotype tested. The resistance of VSMC-specific caveolin-1 deficient mice to BP lowering by PxIL2P supports the idea that VSMC caveiolin-1 regulates BP homeostasis through changes in Panx1 localization and channel function.

DISCUSSION
Sympathetic-mediated vasoconstriction plays a central role in controlling BP homeostasis. This process occurs in part due to the release of norepinephrine from sympathetic nerves, subsequent activation of VSMC α-ARs, and coordinated constriction between VSMCs, thus producing a unified vasoconstriction response. Recent work from our group revealed that VSMC Panx1 channels 21 are a primary mediator of α-AR vasoconstriction, and using VSMC-specific Panx1 knockout mice and genetic rescue experiments we demonstrated that ATP release from tehse channels was necessary for proper adrenergic vasoconstriction 21,54 .
In this way, ATP acts as an autocrine/paracrine signaling molecule that initiates vasoconstriction responses in resistance arteries. However, less is known about the intracellular signaling molecules responsible for supporting Panx1 channel function following adrenergic stimulation. Here we have identified a novel interaction between Panx1 and the caveolae structural protein caveolin-1. Caveolin-1 and Panx1 localize to areas of the plasma membrane innervated by sympathetic nerves and associate with each other following stimulation of α-ARs with phenylephrine. Using a novel, inducible, VSMC-specific caveolin-1 knockout mouse, we also demonstrate that caveolin-1 functionally regulates adrenergic-mediated ATP release, vasoconstriction, MAP, and Panx1 dependent BP responses.
Plasma membrane caveolae influence vascular homeostasis and assist in the localization of key VSMC vasoconstriction signaling molecules such as α-AR and Gq coupled effector molecules in small arteries 19,49,55 . From electron microscopy observations (Supplemental Figure I), we often observe caveolae localized near sympathetic nerves, and thus predicted that this unique vascular feature may beget vascular function. Our live cell, fluorescent tracking experiments, allowed us to examine caveolin-1 and Panx1 interactions in response to stimulation with phenylephrine. Activation of α-AR resulted in a rapid co-association of the two proteins at the plasma membrane ( Figure 1A-B). These temporal effects are consistent with previous constriction recordings following adrenergic stimulation in resistance arteries 56,8 . Moreover, we treated VSMCs with a high concentration of ATP (500 μmol/L), a manipulation that promotes active internalization of membrane associated Panx1 50 , to determine if caveolin-1 association correlates with Panx1 endocytosis. Here we observed a significant decrease in co-localization between Panx1 and caveolin-1 following ATP stimulation ( Figure 1A-C), indicating that Panx1 endocytosis occurs independently of a caveolin-1 association.
Recent studies by Boyce et al. 50 and Gehi et al. 43 report similar co-distributions of Panx1 with caveolin-1 in cell lines. In contrast to the phenylephrine-induced Panx1/caveolin-1 clustering observed herein, a similar decrease in overlap with caveolin-1 was observed following ATP application in N2a cells. 50 It is important to note in this context that caveolin-1-enriched caveolae are only one specialized variety of cholesterol-enriched membrane microdomain or lipid raft 5 . In fact, ATP-mediated Panx1 internalization is cholesterol-dependent 50,51 but dynamin-and clathrin-independent 43,50 , suggestive of a noncanonical endocytosis mechanism. Our novel results then suggest that inclusion of Panx1 in VSMC caveolae may alter channel activity (such as increased open probability [57][58][59] ) rather than trafficking, although this remains to be explicitly tested. In light of this work, and because of the strong association between caveonlin-1 and Panx1 following adrenergic stimulation, we tested whether an endogenous interaction occurs in VSMCs. The same culture model also showed a specific protein interaction using immunoprecipitation of membrane fractions after phenylephrine stimulation ( Figure 1D-F), which was absent in non-caveolar Panx1-containing fractions. In these experiments, we observed multiple glycosylation species of Panx1 60,61 (between 37-55kDa) after immunoprecipitation, perhaps indicative of the recruitment of more Panx1 channels from intracellular stores following stimulation. Panx1 channels are oligomers of Panx1 subunits. It is unknown how many units within the channel must be glycosylated to allow for the appropriately trafficking of plasma membrane channels. It is plausible that Panx1 subunits are differentially regulated, which has been strongly suggested in other published work 29 . It also remains to unknown whether binding of Panx1 to caveolin-1 after phenylephrine stimulation requires a direct interaction, which has been observed for other membrane channels [57][58][59] , or if other membrane associated effector molecules are required, such as Src family kinases 62,63 .
Panx1 plays a vital role in α-AR vasoconstriction, but not in other constriction pathways 22 . We reasoned that caveolae might engender the formation of α-AR membrane microdomains near the VSMC membrane innervated by sympathetic nerves. Using intact mouse resistance arteries in conjunction with PLA, we observed an adrenergic-induced interaction of caveolin-1 and Panx1 preferentially localized around sympathetic nerves ( Figure 2D-E). These observations suggest that caveolae may support the formation of a signaling microdomain important for Panx1 activation 22 . Moreover, resistance arteries are characterized by multiunit neural innervation to VSMCs, whereby small patches of VSMCs are contacted by sympathetic nerves to allow for finer individual control of vasoconstriction 64 . This innervation pattern differs from the unitary innervation observed in visceral organs and large arteries, which features a single VSMC neural input and relies on gap junction connectivity to synchronize constriction responses. In this study as in previously published findings 21 , we detect a scarcity of gap junction connectivity between VSMCs in our models by Cx43 immunostaining herein (Supplemental Figure III) or electron microscopy 21 . These results suggest that resistance arteries may preferentially utilize autocrine/paracrine-mediated signals to couple VSMC constriction responses, and may further utilize purinergic signaling pathways mediated through Panx1 to facilitate this function.
The idea that spatially localized signals support vascular function has been extensively characterized in the endothelium, where regulation of tyrosine kinases and endothelial nitric oxide synthase is dependent on caveolin-1 5 ; however, less is known about these processes in VSMCs, which utilize sympathetic innervation to adjust vascular resistance for proper blood pressure control 65 . Perturbations to this adrenergic signaling axis may underlie clinical pathologies in patients suffering from treatment-resistant hypertension who present with enhanced sympathetic drive, norepinephrine spillover, and excessive vascular resistance [66][67][68] .
To explore the functional role of caveolin-1 in VSMCs we generated an inducible VSMCspecific caveolin-1 knockout mouse ( Figure 3A-B) to selectively removed caveolin-1 from the vascular media ( Figure 3C). Importantly, caveolin-1 expression remained present in the vascular endothelium (CD31-positive cells) between control and knockout mice. No gross alterations in the medial wall thickness or cell number were observed in TDAs (data not shown) as was described for pulmonary arteries in the constitutive global knockout 13 . It is likely that the utilization of an inducible Cre-recombinase system in adult mice curtailed any negative compensatory effects seen in constitutive systems.
Using ATP bioluminescence assays, we measured phenylephrine-stimulated ATP release from ex vivo arteries, a process that our laboratory has shown uniquely couples with Panx1 activation 22 . In all cases, caveolin-1 deletion significantly reduced phenylephrine-and norepinephrine-stimulated ATP release, similar to observations in Panx1 VSMC-knockout mice 22 ( Figure 3E). From our immunofluorescence analyses, it seems likely that Panx1 is the primary conduit for ATP release from VSMCs, as VNUT and Cx43 were not detected in VSMCs of our vessels (Supplemental Figure III). These data suggest that caveolin-1 is involved upstream of Panx1 activation and influences the ATP release typically ascribed to Panx1 function. It remains to be demonstrated if a direct interaction between caveolin-1 is sufficient to induce Panx1 channel opening.
In the vasculature, a primary role for caveolin-1 is ascribed to negative regulation of endothelial nitric oxide synthase (eNOS) activity in the endothelium 69,70 . Global deletion of caveolin-1 increases eNOS activity, resulting in higher concentrations of cGMP, increased NO release, and alterations in myogenic tone and dilation responses to acetylcholine 16,53 . However, the role of VSMC-derived caveolin-1 has not been specifically examined in resistance arteries. Using pressure myography, we found that VSMC caveolin-1 deletion significantly blunts phenylephrine-stimulated vasoconstriction responses, but does not change vasodilation responses to acetylcholine ( Figure 4A-B), indicating that reduced vasoconstriction is not dependent on endothelial-derived mechanisms observed in global caveolin-1 knockout models 13 . ATP release from the vascular wall exhibits dual activities in resistance arteries dependeing on which cell type release ATP (vasoconstriction in smooth muscles and vasodilation in endothelial cells) The subtypes of purinergic receptors that are activated by ATP can also influence responses within the vascular wall with the largest contribution in smooth muscle being mediated by P2X1 ionotropic receptors 71 . Our laboratory has demonstrated the existence of purinergic component to adrenergic mediated vascular events whereby incubation of vessels with the ATP degrading enzyme apyrase or pharmacological blockers of purinergic receptors (P2X) prevents adrenergic vasoconstriction 21 . Thus, we conclude that caveolin-1 in our model functionally regulates the initial and upstream release of ATP utilized to activate downstream purinergic receptors after adrenergic stimulation.
VSMC-specific deletion of caveolin-1 results in reduced adrenergic ATP release and vasoconstriction, a similar vascular phenotypic observed when Panx1 is specifically deleted from VSMCs. We predicted that a similar dysregulation in BP would then occur after caveolin-1 deletion as observed in Panx1 knockout models: specifically, a reduction in MAP predominantly during the animal's active period (night MAP), when sympathetic drive is high. Indeed, in vivo BP monitoring revealed a significant reduction in 24 hour MAP ( Figure   5A-B) with a greater reduction occurring at night ( Figure 5C) in caveolin-1-deficient animals, but not during the day (inactive period) ( Figure 5D). The effect size of BP lowering due to caveolin-1 deletion (~4 mmHg, 24 hour MAP; ~6 mmHg, Night MAP) was nearly identical to BP reductions in VSMC-Panx1 knockout animals 53 . To further confirm that alterations in MAP were due to vascular changes (reduced vascular resistance), we assessed cardiac function using MRI. No measurable morphological differences were detected in our analysis and all animals were functionally similar (e.g., heart rate, stroke volume; Table 1). Therefore, alterations in vascular resistance due to VSMC caveolin-1 deletion are likely responsible for observed reductions in MAP.
To determine if caveolin-1 dependent BP responses utilize in part a Panx1-dependent pathway, we acutely treated animals with PxIL2P peptide to pharmacologically lower blood pressure. PxIL2P has previously been shown to significantly reduce adrenergic stimulated Panx1 channel opening and ATP release, and also potently lowers blood pressure in mice. In the current analysis, wild-type and control mice dramatically respond to PxIL2P treatment (i.e reduced blood pressure lowering), with no effects due to scrambled control peptide 22 . However, conditional deletion of VSMC caveolin-1 are completely protected to BP lowering by PxIL2P treatment. This strongly suggests that caveolin-1 is a key intermediary necessary for normal Panx1 function (Figure 7). These data further highlight the Panx1 intracellular loop as an important target for Panx1 activation. Previous work from our group has demonstrated that the region mapping to the PxIL2P peptide contains a regulatory motif, which upon genetic mutation negatively influences Panx1 channel function and adrenergic mediated vasoconstriction 22 . Future studies are needed to determine if caveolin-1 directly influences the Panx1 intracellular loop, or if associative factors bound to caveolin-1 are required for proper channel function.
In this analysis, we explore a novel adrenergic vasoconstriction pathway that has previously been shown to coordinate constriction responses through the release of ATP by Panx1. We describe the impact of caveolin-1 deletion in the peripheral vasculature and on systemic BP regulation using a novel smooth muscle-specific mouse model. We have limited changes due to compensatory deletion-effects by utilizing an inducible Cre-lox system in adult mice. Based on our analyses we found no additional phenotypes outside of the cardiovascular phenotypes reported in this study. Although, direct experimental evidence and future studies are required to determine if additional phenotypes exist. Here we show that the localization of Panx1 to plasma membrane caveolae and the scaffold protein caveolin-1 promotes a novel interaction important for regulating BP. It remains to be observed whether other scaffold proteins or cytoskeletal proteins contribute to the localization and formation of this unique membrane domain near sympathetic nerve terminals in VSMC. Panx1 has been shown to directly interact with the actin cytoskeleton 72 and its modulator actin-related protein 3 73,74 , and treatment with cytochalasin B destabilized Panx1 plasma membrane distribution in vitro 72 . It is also unknown which intracellular mediators are mechanistically required for activation of Panx1 by α-AR in VSMCs, which will be especially important to determine.
Overall, our data demonstrate that Panx1 and caveolin-1 functionally couple with each other in VSMCs. Using immunofluorescence co-localization and co-immunoprecipitation, we show that caveolin-1 and Panx1 interact at distinct areas of VSMC plasma membrane innervated by sympathetic nerves, suggesting the existence of an adrenergic micro-signaling domain. This interaction was induced by α-AR stimulation and is necessary for adrenergicmediated vasoconstriction and ATP release from resistance arteries. We found that VSMC caveolin-1 is necessary to control systemic BP responses through modulation of Panx1 function and may facilitate appropriate channel function through the Panx1 intracellular loop region.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

PERSPECTIVES
A strong medical need exists for an effective, safe, and innovative pharmacological therapy that promotes BP regulation and reduces the vasoconstriction that accompanies excessive sympathetic stimulation. Our observations demonstrating an adrenergicmediated interaction between VSMC caveolin-1 with Panx1, and the subsequent caveonlin-1 dependent control of vascular responsiveness and blood pressure homeostasis, are important for expanding our knowledge of a novel adrenergic signaling pathway that may underlie treatment-resistant hypertension. In the future, our goal is to elucidate signaling partners involved forming functional adrenergic signaling microdomains and determine their applicability as targets for therapeutic intervention.

HIGHLIGHTS
• Presents a novel interaction between the ATP release channel Pannexin1 and the caveolae scaffolding protein caveolin-1, which is mediated by adrenergic stimulation.

•
Demonstrates that conditional genetic deletion of caveolin-1 from smooth muscle cells recapitulates Pannexin1 vascular phenotypes.

•
Demonstrates that smooth muscle cell caveolin-1 regulates adrenergic stimulated vascular constriction and mean arterial pressure.         Table 1 Cardiac function is unaltered in VSMC caveolin-1 deficient mice MRI was performed on a 7 Tesla (T) Clinscan system (Bruker, Ettlingen, Germany) equipped with actively shielded gradients with a full strength of 650 mT/m and a slew rate of 6666 mT/m/ms. Baseline LV structure and function were assessed. Six short-axis slices were acquired from base to apex, with slice thickness equal to 1mm, in-plane spatial resolution of 0.2 × 0.2 mm 2 , and temporal resolution of 8-12 ms. Baseline EF, end-diastolic volume (EDV), end-systolic volume (ESV), myocardial mass, wall thickness, and wall thickening were measured from the cine images using the freely available software Segment version 2.0 R5292 (http://segment.heiberg.se). EDV and ESV were then indexed to body mass (EDVI and ESVI, respectively). Mass to volume ratio (MVR) was calculated as the ratio of myocardial mass to EDV. A Student's t-test (two tailed) was performed for significance; * p<0.