The SWELL1-LRRC8 complex regulates endothelial AKT-eNOS signaling and vascular function

The endothelium responds to numerous chemical and mechanical factors in regulating vascular tone, blood pressure, and blood flow. The endothelial volume-regulated anion channel (VRAC) has been proposed to be mechanosensitive and thereby sense fluid flow and hydrostatic pressure to regulate vascular function. Here, we show that the leucine-rich repeat-containing protein 8a, LRRC8A (SWELL1), is required for VRAC in human umbilical vein endothelial cells (HUVECs). Endothelial LRRC8A regulates AKT-endothelial nitric oxide synthase (eNOS) signaling under basal, stretch, and shear-flow stimulation, forms a GRB2-Cav1-eNOS signaling complex, and is required for endothelial cell alignment to laminar shear flow. Endothelium-restricted Lrrc8a KO mice develop hypertension in response to chronic angiotensin-II infusion and exhibit impaired retinal blood flow with both diffuse and focal blood vessel narrowing in the setting of type 2 diabetes (T2D). These data demonstrate that LRRC8A regulates AKT-eNOS in endothelium and is required for maintaining vascular function, particularly in the setting of T2D.


Introduction
The endothelium integrates mechanical and chemical stimuli to regulate vascular tone, angiogenesis, blood flow, and blood pressure (Cahill and Redmond, 2016). Endothelial cells express a variety of mechanosensitive and mechanoresponsive ion channels that regulate vascular function (Gerhold and Schwartz, 2016), including TRPV4 (Sonkusare et al., 2012;Earley et al., 2009;Zhang et al., 2009;Mendoza et al., 2010) and Piezo1 (Rode et al., 2017;Li et al., 2014;Coste et al., 2010). The volume-regulated anion channel (VRAC) current is also prominent in endothelium and has been proposed to be mechanoresponsive , to activate in response to fluid flow/hydrostatic pressure  and to putatively regulate vascular reactivity. However, the molecular identity of this endothelial ion channel has remained a mystery for nearly two decades.
Lrrc8a (leucine-rich repeat-containing protein 8A, also known as SWELL1) encodes a transmembrane protein first described as the site of a balanced translocation in an immunodeficient child with agammaglobulinemia and absent B-cells (Sawada et al., 2003;Kubota et al., 2004). Subsequent work revealed the mechanism for this condition to be due to impaired LRRC8A-dependent GRB2-PI3K-AKT signaling in lymphocytes, resulting in a developmental block in lymphocyte differentiation (Kumar et al., 2014). Thus, for~11 years, LRRC8A was conceived of as a membrane protein that regulates PI3K-AKT mediated lymphocyte function (Sawada et al., 2003;Kubota et al., 2004). Although LRRC8A had been predicted to form a hetero-hexameric ion channel complex with other LRRC8 family members (Abascal and Zardoya, 2012), it was not until 2014 that LRRC8A was shown to form an essential component of the volume-regulated anion channel (VRAC) (Qiu et al., 2014;Voss et al., 2014), forming hetero-hexamers with LRRC8b-e (Voss et al., 2014;Syeda et al., 2016). Therefore, historically, LRRC8A was first described as a membrane protein that signaled via proteinprotein interactions and then later found to form an ion channel signaling complex.
We showed previously that LRRC8A is an essential component of VRAC in adipocytes  and skeletal myotubes (Kumar et al., 2020), which is required for insulin-PI3K-AKT signaling Kumar et al., 2020) to mediate adipocyte hypertrophy , skeletal myotube differentiation (Kumar et al., 2020), skeletal muscle function (Kumar et al., 2020), and systemic glucose homeostasis Kumar et al., 2020;Xie et al., 2017;Gunasekar et al., 2019). The PI3K-AKT-endothelial nitric oxide synthase (eNOS) signaling pathway is central to transducing both mechanical stretch (Hu et al., 2013) and hormonal inputs (insulin) to regulate eNOS expression and activity, which, in turn, regulates vasodilation (blood flow and pressure), inhibits leukocyte aggregation, and limits proliferation of vascular smooth muscle cells (atherosclerosis). Indeed, insulin resistance is thought to be a systemic disorder in the setting of type 2 diabetes (T2D), affecting endothelium in addition to traditional metabolically important tissues, such as adipose, liver, and skeletal muscle (Janus et al., 2016;Kearney et al., 2008;Muniyappa and Sowers, 2013). In fact, insulin-resistant endothelium and the resultant impairment in PI3K-AKT-eNOS signaling have been proposed to underlie much of the endothelial dysfunction observed in the setting of obesity and T2D, predisposing to hypertension, atherosclerosis, and vascular disease (Janus et al., 2016;Kearney et al., 2008;Muniyappa and Sowers, 2013).
In this study, we demonstrate that VRAC is LRRC8A-dependent in endothelium, associates with GRB2, caveolin-1 (Cav1), eNOS, and regulates PI3K-AKT-eNOS signaling and flow-mediated endothelial cell orientation -suggesting that LRRC8 channel complexes link insulin and mechano-signaling in endothelium. LRRC8A-dependent AKT-eNOS, ERK1/2, and mTOR signaling influences blood pressure and vascular function in vivo, while impaired endothelial LRRC8A signaling predisposes to vascular dysfunction in the setting of diet-induced T2D.
Endothelial cell orientation to the direction of shear flow is impaired upon LRRC8A KD To evaluate the requirement of LRRC8A for endothelial cell responses to physiological mechanical stimuli relevant to endothelium, we examined the LRRC8A dependence of HUVEC orientation to laminar shear flow. In response to laminar shear flow at 15 dynes/cm 2 for 24 hr ( Figure 5A . These LRRC8A-dependent differences were also apparent in the degree of impairment of HUVEC elongation to the direction of fluid flow ( Figure 5G) and in the velocity of movement against the direction of flow ( Figure 5H&I).
Next, we asked whether shear-flow-mediated eNOS phosphorylation is LRRC8A-dependent, as observed with stretch-mediated signaling ( Figure 4). We stimulated siScr and siLrrc8a transfected HUVECs with either no flow, or laminar shear flow for 2 or 24 hr and then immunostained for p-eNOS ( Figure 5J). We observed marked p-eNOS induction in response to shear flow after both 2 and 24 hr in siControl transfected HUVECs, and this was significantly abrogated upon LRRC8A KD ( Figure 5J). Interestingly, this shear-flow stimulated p-eNOS displayed significant nuclear localization, consistent with prior reports (Feng et al., 1999;Andersson and Brittebo, 2012;Wang et al., 2019;Gobeil et al., 2006). The same nuclear distribution was also observed using a different p-eNOS antibody (Figure 5-figure supplement 1). We quantified the intensity of p-eNOS staining in two ways: p-eNOS excluding the nuclear signal ( Figure 5K,L) and p-eNOS in the entire cell ( Figure 5M,O). In both cases, we observed significant shear-flow-mediated p-eNOS induction in siScr HUVECs, after both 2 hr and 24 hr flow, and this was reduced in LRRC8A KD HUVEC ( Figure 5K-O), consistent with LRRC8A-dependent flow-mediated p-eNOS signaling.
Endothelial-targeted LRRC8A KO mice exhibit mild angiotensin-II stimulated hypertension and impaired retinal blood flow in the setting of T2D To examine the functional consequences of endothelial LRRC8A ablation in vivo, we generated endothelial-targeted Lrrc8a knock-out mice (eLrrc8a KO) by crossing Lrrc8a floxed mice Kumar et al., 2020;Kang et al., 2018) with the endothelium-restricted CDH5-Cre mouse (CDH5-Cre; Lrrc8a fl/fl ; Figure 7A). Patch-clamp recordings of primary endothelial cells isolated from WT and eLrrc8a KO mice ( Figure 7B) reveal robust hypotonically activated currents (Hypo, 210 mOsm) in WT endothelial cells, which are DCPIB inhibited ( Figure 7C&E), while eLrrc8a KO endothelial cells exhibit markedly reduced hypotonically activated currents ( Figure 7D&E). Consistent with this, IF staining of aortic ring explants from WT and eLrrc8a KO mice confirmed LRRC8A ablation from CD31+ primary endothelial cells ( Figure 7F&G).
Based on our findings that LRRC8A regulates AKT-eNOS signaling in HUVECs in vitro, we next examined p-eNOS in aortic endothelium by IF staining in WT and LRRC8A KO mice. Similar to results in LRRC8A KD HUVEC, we found reduced p-eNOS intensity in aortic endothelium of eLrrc8a KO mice as compared to WT mice ( Figure 7H,I).
Since eNOS signaling is central to blood pressure regulation, we next examined blood pressures in eLrrc8a KO mice compared to WT controls (Lrrc8a fl/fl mice). Male mice exhibited no significant differences in systolic blood pressure under basal conditions ( Figure 7J), while female mice were mildly hypertensive relative to WT mice ( Figure 7K). However, after 4 weeks of angiotensin-II (Ang II) infusion, male eLrrc8a KO mice developed exacerbated systolic hypertension as compared to Ang II-treated WT mice ( Figure 7L). These data are consistent with endothelial dysfunction and impaired vascular relaxation in eLrrc8a KO mice, resulting in a propensity for systolic hypertension.
As endothelial dysfunction may also result in impaired blood flow, we performed retinal imaging during intraperitoneal (i.p.) injection of fluorescein to assess retinal vessel blood flow and morphology in WT and eLrrc8a KO mice. Mice raised on a regular diet had mild, non-significant impairments in retinal blood flow, based on the relative rate of rise of the fluorescein signal in retinal vessels (  Densitometry values for GRB2 co-immunoprecipitated eNOS (eNOS/GRB2). Representative blots from three independent experiments. (C) Representative endogenous LRRC8A and eNOS immunofluorescence staining in Ad-shScr and Ad-shLrrc8a transduced HUVECs. Representative image from six independent experiments. (D-E) Quantification of LRRC8A (D, n = 6) and eNOS (E, n = 6) immunofluorescence staining upon LRRC8A knockdown (KD). Evidence of LRRC8A-eNOS co-localization (C, insets) in plasma membrane (i) and perinuclear regions (ii). Significance between the indicated groups is calculated using an unpaired two-tailed Student's t-test. p-Values are indicated on figures. Data are shown as mean ± s.e.m. ***p<0.001. The online version of this article includes the following source data and figure supplement(s) for figure 3: Source data 1. Source data for Figure 3D. Source data 2. Source data for Figure 3E.  Figure 8J). These parameters are all suggestive of diabetes-induced retinal vessel dysfunction in the eLrrc8a KO mice, consistent with the loss of eNOS activity that is expected when insulin signaling is compromised (Brooks et al., 2001;Kondo et al., 2003). Notably, both WT and eLrrc8a KO mice were found to be equally glucose-intolerant and insulin-resistant ( Figure 8-figure supplement 3), indicating that these differences in microvascular dysfunction were not due to increased hyperglycemia and more severe diabetes in eLrrc8a KO mice. However, while there were no differences in mean body weight or fat mass between WT and eLrrc8a KO males raised on HFHS diet (Figure 8-figure supplement 4), female eLrrc8a KO mice had higher body weights than WT female mice, and this was driven by an increase Figure 4. Leucine-rich repeat-containing protein 8a (LRRC8A) is required for intact stretch-induced AKT-endothelial nitric oxide synthase (eNOS) signaling. (A) Western blot of LRRC8A, pAKT1, pAKT2, pERK1/2 in response to 30 min of 0.1% and 5% static stretch in Ad-shScr (n = 3) and Ad-shLrrc8a (n = 3) transduced human umbilical vein endothelial cells (HUVECs). GAPDH is used as a loading control. (B-D) Densitometry quantification from A of pAKT1 (B), pAKT2 (C), and pErk1/2 (D). (E) Western blot of peNOS, LRRC8A, in response to 5% static stretching for 5, 60, and 180 min in Ad-shScr (n = 3) and Ad-shLrrc8a (n = 3) transduced HUVECs. (F) Densitometry quantification from E of eNOS. GAPDH is used as a loading control. Statistical significance between the indicated group is calculated using one-way analysis of variance (ANOVA), Tukey's multiple comparisons test for (B-D), and two-way ANOVA for (F) with p-value in upper left corner. Data are shown as mean ± s.e.m. *p<0.05; **p<0.01; ***p<0.001. The online version of this article includes the following source data for figure 4: Source data 1. Source data for Figure 4F. Source data 2. Source data for Figure 4B,C, D.
in total fat mass ( Figure 8-figure supplement 4). Taken together, our findings reveal that LRRC8A is highly expressed in endothelium and functionally encodes endothelial VRAC. LRRC8A regulates basal, stretch, and shear-flow-mediated ERK, AKT-eNOS signaling, forms an LRRC8A-GRB2-Cav1-eNOS signaling complex, and regulates vascular function in vivo.

Discussion
Our findings demonstrate that the LRRC8 hetero-hexamer functionally encodes endothelial VRAC, whereby LRRC8A associates with GRB2 and Cav1 and positively regulates PI3K-AKT-eNOS and ERK1/2 signaling. LRRC8A depletion in HUVECs reduces pAKT2, pAKT1, p-eNOS, and pERK1/2 under basal conditions, and abrogates both stretch and shear-flow stimulated p-eNOS stimulation. We also discovered the ability of HUVECs to align along the direction of laminar shear flow is markedly impaired in LRRC8A KD cells -suggesting these cells have an impaired ability to either sense or respond to laminar shear flow (or both). These data reveal that LRRC8A-mediated PI3K-AKT signaling is conserved in endothelium, similar to previous observations in adipocytes  and skeletal myotubes (Kumar et al., 2020), and in turn positively regulates eNOS expression and activity. Consistent with this mechanism, endothelial-targeted Lrrc8a ablation in vivo predisposes to microvascular dysfunction in the setting of T2D and to hypertension in response to Ang II infusion. These results are in line with the notion that LRRC8A depleted endothelium contributes to an insulin-resistant state in which impaired PI3K-AKT-eNOS signaling results in a propensity for vascular dysfunction Muniyappa and Sowers, 2013). Insulinmediated regulation of NO is physiologically (Zeng and Quon, 1996;Zeng et al., 2000;Montagnani et al., 2002) and pathophysiologically (Steinberg et al., 1996) important, as NO has vasodilatory (Quillon et al., 2015;Palmer et al., 1987), anti-inflammatory (Kataoka et al., 2002), antioxidant (Clapp et al., 2004), and antiplatelet effects (Schäfer et al., 2004a;Schäfer et al., 2004b;Radomski et al., 1987a;Radomski et al., 1987b). Impaired NO-mediated vascular reactivity is a predictor of future adverse cardiac events (Schächinger et al., 2000) and portends increased risk of atherosclerosis (Bugiardini et al., 2004). Consistent with these NO-mediated effects, the RNA sequencing data derived from LRRC8A KD HUVECs revealed enrichment in inflammatory, cell adhesion, and proliferation pathways (GADD45, IL-8, mTOR, TREM1 signaling) that may arise from LRRC8A-mediated dysregulation of eNOS activity. Moreover, Lrrc8a mRNA is highly expressed in HUVECs relative to adipocyte and myotube cell lines examined previously Kumar et al., 2020), higher than many endothelial genes, and higher than other mechanosensitive (Piezo1) and mechanoresponsive (TRPV4) ion channels with well-established roles in endothelial biology (Sonkusare et al., 2012;Rode et al., 2017;Li et al., 2014;Cappelli et al., 2019;Dalsgaard et al., 2016;Dunn et al., 2013;Earley et al., 2005;Harraz et al., 2018;Mercado et al., 2014;Sonkusare et al., 2014).
In addition to reductions in AKT-eNOS signaling, LRRC8A depletion in HUVECs also reduced ERK1/2 signaling. This decrease in pERK1/2 suggests impaired MAPK signaling which is connected to the insulin receptor by GRB2-SOS. Indeed, we also found that LRRC8A and GRB2 interact in HUVECs, and this may provide the molecular mechanism for the observed defect in ERK signaling. Interestingly, GRB2-MAPK signaling is thought to promote angiogenesis, migration, and proliferation (Zhao et al., 2011), so reductions in ERK1/2 signaling would be predicted to inhibit these processes. One hypothesized mechanism of LRRC8A-mediated regulation of ERK1/2 and AKT2-eNOS signaling involves non-conductive protein-protein signaling via the LRRD as described previously in adipocytes Gunasekar et al., 2019). An alternative hypothesis is that LRRC8Amediated VRAC channel activity may alter endothelial membrane potential to regulate other endothelial ion channels, such as Piezo1 or TRPV4, and downstream calcium signaling to influence downstream signaling. Future studies will further delineate the molecular mechanisms of LRRC8A regulation of intracellular signaling in endothelial cells and vascular function, and also directly explore whether endothelial LRRC8A-mediated VRAC channel complexes are themselves mechanoresponsive, as suggested in prior studies in other cell types (Browe and Baumgarten, 2003;Browe and Baumgarten, 2006).
Another curious observation that warrants further exploration is the reduction in basal Akt2 and NOS2 (eNOS) mRNA expression upon LRRC8A KD in HUVECs, which is also observed at the protein level. We can only speculate that LRRC8A is somehow regulating Akt2 and eNOS gene expression. Both PI3K and ERK1/2 signaling pathways have been described to regulate total eNOS expression (Wu, 2002), and we observe reductions in both PI3K and ERK1/2 signaling in LRRC8A KD HUVECs, so reductions in eNOS expression are certainly consistent with disruptions in these signaling pathways. Consistent with this conserved biology with respect to LRRC8A signaling, we also observed significant twofold reductions in AKT2 expression upon Lrrc8a deletion in both 3T3-F442A adipocytes  and C2C12 myotubes (Kumar et al., 2020), in addition to reductions in NOS2 (iNOS) (Kumar et al., 2020) in adipocytes and NOS1 (neuronal NOS) in C2C12 myotubes (Kumar et al., 2020). Collectively, these findings across multiple cell types support a PI3K and ERK1/2 mediated mechanism of regulation of eNOS and AKT2 gene expression.
The phenotype of endothelial-targeted Lrrc8a KO (eLrrc8a KO) mice are consistent with a reduction in eNOS and p-eNOS as these mice exhibit mild hypertension at baseline (females) and exacerbated hypertension in response to chronic angiotensin infusion, suggesting a modulatory effect on vascular reactivity. Similarly, retinal blood flow is only mildly impaired in eLrrc8a KO mice raised on a regular diet, with some evidence of microvascular disease. However, both retinal blood flow and retinal vessel morphology are markedly impaired in obese-T2D, insulin-resistant eLrrc8a KO mice raised on an HFHS diet compared to controls. This is consistent with a synergistic role of endothelial Source data 1. Source data for Figure 5E. Source data 2. Source data for Figure 5F. Source data 3. Source data for Figure 5G. Source data 4. Source data for Figure 5I. Source data 5. Source data for Figure 5K. Source data 6. Source data for Figure 5L. Source data 7. Source data for Figure 5M. Source data 8. Source data for Figure 5N.     Figure 6 continued on next page eLrrc8a ablation and T2D/obesity in the pathogenesis of vascular disease. Our results suggest that reductions in LRRC8A signaling may contribute to impaired vascular function observed in humans in response to insulin and/or shear stress in the setting of obesity (Arcaro et al., 1999;Tack et al., 1998;Westerbacka et al., 1999;Williams et al., 2006) and insulin resistance (Murphy et al., 2007). Continued on next page genes. (C) Expression levels of select LRRC8 mRNA, classic endothelial markers, and mechanoresponsive ion channels in shScr and shLrrc8a-treated HUVECs. (D) Expression levels LRRC8a-e mRNA in HUVECs compared to data previously published from C2C12 myotubes (Kumar et al., 2020) and differentiated 3T3-F442A adipocytes . For (C), statistical significance between the indicated values was calculated using an unpaired two-tailed Student's t-test. Error bars represent mean ± s.e.m. n = 3, independent experiments. *p<0.05; **p<0.01; ***p<0.001. The online version of this article includes the following source data for figure 6:

Materials and methods
Source data 1. Source data for Figure 6C. Source data 2. Source data for Figure 6D.

Animals
The institutional animal care and use committee of the University of Iowa and Washington University School of Medicine approved all experimental procedures involving animals. All mice were housed in temperature, humidity, and light controlled environment and allowed water access and food. Both male and female Lrrc8a fl/fl Kang et al., 2018) (WT), CDH5-Cre;Lrrc8a fl/fl (eLrrc8a

siRNA KD
HUVECs were plated at 360,000 cell/well in six-well plates. Cells were grown for 24 hr (90-95% confluency) and transduced with either a silencer select siRNA with siLRRC8a (Cat#4392420, sense: GCAACUUCUGGUUCAAAUUTT antisense: AAUUUGAACCAGAAGUUGCTG, Invitrogen) or a nontargeting control silencer select siRNA (Cat#4390846, Invitrogen), as described previously (Koh et al., 2008). The siLRRC8a used targets a different sequence from the shRNA described above. Briefly, siRNAs were transduced twice, 24 and 72 hr after HUVEC plating. Each siRNA was combined with Opti-MEM (285.25 ml, Cat#11058-021, Invitrogen) siPORT amine (8.75 ml, Cat#AM4503, Invitrogen) and the silencer select siRNA (6 ml) in a final volume of 300 ml. HUVECs were transduced over a 4 hr period at 37˚C, using DMEM +1% FBS. After transduction, the cells Statistical significance between the indicated values for B and C was calculated using two-way analysis of variance (ANOVA) and (D-L) using a two-tailed unpaired Student's t-test. Error bars represent mean ± s.e.m. *p<0.05; **p<0.01; ***p<0.001. The online version of this article includes the following video, source data, and figure supplement(s) for figure 8: Source data 1. Source data for Figure 8B. Source data 2. Source data for Figure 8C. Source data 3. Source data for Figure 8D. Source data 4. Source data for Figure 8E. Source data 5. Source data for Figure 8F. Source data 6. Source data for Figure 8G. Source data 7. Source data for Figure 8H. Source data 8. Source data for Figure 8I. Source data 9. Source data for Figure 8J. Source data 10. Source data for Figure 8K. Source data 11. Source data for Figure 8L.           were returned to media containing M199, 20% FBS, 0.05 g heparin sodium salt, and 15 mg ECGS. Cell lysates were collected at basal conditions on day 4.

Isolation of mouse lung endothelial cells
Isolation of mouse lung endothelial cells was performed according to the following protocol: Day 1 -Incubate sheep anti-rat IgG Dynabeads (Invitrogen) overnight with PECAM (Sigma, #SAB4502167) and VEGFR2 (R and D Systems, #BAF357) antibodies at 4˚C in PBS with gentle agitation. Day 2 -Lungs were removed from the mice, washed in 10% FBS/DMEM, minced into 1-2 mm squares, and digested with Collagenase Type I (2 mg/ml, Gibco) at 37˚C for 1 hr with agitation. The cellular digest was filtered through a 70 mm cell strainer, centrifuged at 1500 rpm, and the cells immediately incubated with the antibody coated Dynabeads at RT for 20 min. The bead-bound cells were recovered with a magnet, washed two times with PBS, and plated overnight on collagen type I (100 mg/ml) coated coverslips. The endothelial cells were maintained in a growth media of M199, 20% FBS, 0.05 g heparin sodium salt, 50 mg/ml ECGS, and 1Â anti-anti.

Cell culture
HUVECs were purchased from ATCC and were grown in MCDB-131-complete media overnight. HUVECs for basal condition collection were grown in growth media of M199, 20% FBS, 0.05 g heparin sodium salt (Cat#9041-08-1, Alfa Aesar), and 15 mg ECGS (Cat#02-102, Millipore Sigma). Cells were routinely cultured on 1% of gelatin-coated plates at 37˚C at 5% CO 2 . For insulin stimulation (Cat#SLBW8931), cells were serum starved for at least 13 hr in 1% FBS (Atlanta Bio selected, Cat#S11110) or without FBS using endothelial cells growth basal medium (Lonza Cat#cc-3121) instead of MCDB-131-complete media. Insulin stimulation was used for the times indicated at 100 nM.

Immunoblotting
Cells were harvested in ice-cold lysis buffer (150 mM NaCl, 20 mM HEPES, 1% NP-40, 5 mM EDTA, pH 7.5) with added proteinase/phosphatase inhibitor (Roche). Cells were kept on ice with gentle agitation for 20 min to allow complete lysis. Lysate scraped into 1.5 ml tubes and cleared of debris by centrifugation at 14,000 Â g for 20 min at 4˚C. Supernatants were transferred to fresh tube and solubilized protein was measured using a DC protein assay kit (Bio-Rad). For immunoblotting, an appropriate volume of 1Â Laemmli (Bio-Rad) sample loading buffer was added to the sample (10 mg of protein), which then heated at 90˚C for 5 min before loading onto 4-20% gel (Bio-Rad). Proteins were separated using running buffer (Bio-Rad) for 2 hr at 150 V. Proteins were transferred to PVDF membrane (Bio-Rad) and membrane blocked in 5% (w/v) BSA in TBST or 5% (w/v) milk in TBST at RT for 2 hr. Blots were incubated with primary antibodies at 4˚C overnight, followed by secondary antibody (Bio-Rad, Goat-anti-mouse #170-5047, Goat-anti-rabbit #170-6515, all used at 1:10,000) at RT for 1 hr. Membranes were washed three times and incubated in enhanced substrate Clarity (Bio-Rad) and imaged using a ChemiDoc XRS using Image Lab (Bio-Rad) for imaging and analyzing protein band intensities. ß-Actin or GAPDH levels were quantified to correct for protein loading.

Immunoprecipitation
Cells were seeded on gelatin-coated 10 cm dishes in complete media for 24 hr. Adenoviruses, Ad5-mCherry-U6-hLrrc8a-shRNA or Ad5-U6-Scr-mCherry were added to cells for 12 hr. After 4 days cells were serum starved for 16 hr with basal media containing 1% serum before stimulation with insulin (10 nM/ml). Cells were harvested in ice-cold lysis buffer (150 mM NaCl, 20 mM HEPES, 1% NP-40, 5 mM EDTA, pH 7.5) with added proteinase/phosphatase inhibitor (Roche) and kept on ice with gentle agitation for 15 min to allow complete lysis. Lysates were incubated with anti-Grb2 antibody (20 mg/ ml) or control rabbit IgG (20 mg/ml, Santa Cruz sc-2027) rotating end over end overnight at 4˚C. Protein G Sepharose beads (GE) were added to this for a further 4 hr before samples were centrifuged at 10,000 Â g for 3 min and washed three times with RIPA buffer and re-suspended in laemmli buffer (Bio-Rad), boiled for 5 min, separated by SDS-PAGE gel followed by the Western blot protocol.

Stretch assay
Equal amounts of cells were plated in each well in six well plated BioFlex coated with Laminin (BF-3001CCase) culture plate and seeded to~90% confluence. Plates were placed into a FlexCell Jr. Tension System (FX-6000T) and incubated at 37˚C with 5% CO 2 . Prior to stretch stimulation, basal media of 1% FBS was added for 16 hr. Cells on flexible membrane were subjected to static stretch with the following parameter: a stretch of 0.1% and 5% with static strain. Cells were stretched for 5, 30, 60, or 180 min. Cells were then lysed and protein isolated for subsequent Western blots.

Laminar flow studies
HUVECs were grown in 1Â M199 media (Gibco, Cat#11043-023) supplemented with 20% FBS (Gibco, Cat#16140-071), 50 mg bovine hypothalamus extract, and 0.01% heparin until full confluence in six-well plates. The cells were transfected twice across 3 days with 100 nM of siRNA for the Lrrc8a gene (siLrrc8a) (Invitrogen, Cat#4392421, assay #:s32107) and a negative siRNA control (siControl) (Invitrogen, Cat#43900846) using siPORTAmine transfection agent (Invitrogen, Cat#AM4503) following the manufacturer recommendations (Koh et al., 2008). Twenty-four hours after the final siRNA treatment, cells were transferred to m-Slide 0.4 I Luer slide (iBidi, Cat#80176) and incubated overnight at 37˚C, 5% CO 2 , and 95% humidity. The slides were then cultured for 24 hr in a microscope stage incubator (Invitrogen, EVOS7000) under conditions of constant laminar flow at 15 dynes/cm 2 /s (Flocel Inc, Cat#QPL1010) and images acquired every 20 min. The peristaltic pump was converted to be laminar through the utilization of dampers on the outflow tracts of the pump. Efficacy in the generation of laminar flow was confirmed through the imaging of Q-Dots in the circulating media prior to the experiment (Invitrogen, Cat#Q21721MP). At the termination of the experiment, the cells were rinsed with 1Â PBS and fixed with 4% paraformaldehyde (PFA) for downstream immunostaining applications.

Immunofluorescence imaging
Cells were plated on gelatin-coated glass coverslips. Cells on coverslip were washed in PBS and fixed with 2% (w/v) PFA for 20 min at RT. PFA was washed three times with PBS and permeabilized in PBS containing 0.2% Triton X-100 for 5 min at RT. Cells on coverslips were washed in PBS and blocked for 30 min at RT with TBS containing 0.1% Tween-20% and 5% BSA. Cells on coverslip were incubated overnight at 4˚C with primary antibody (1:250) in TBS containing 0.1% Tween-20% and 1% BSA. Cells were then washed in PBS five times and incubated for 2 hr at RT with 5% BSA in TBST. Cells were washed three times in TBST and then incubated with secondary antibodies at 1:1000 dilution (Invitrogen, Anti-Rabbit 488, A11070; Anti-mouse 568, A11019; Anti-mouse 488, A11017) for 1 hr at RT. Coverslips were then incubated for 10 min with Topro 3 (T3605, Thermofisher) or mounted with mounting media containing DAPI (Invitrogen) to visualize nuclei. Images were taken using Axiocam 503 Mono Camera controlled by Zeiss Blue using a Plan-Apochromate 40Â oil immersion objective.

Immunohistochemistry of thoracic aorta sections
Mouse thoracic aortas were dissected and immediately fixed in 10% formalin. The formalin fixed aortas were embedded in paraffin and 5 mm sections were cut by automated microtome and mounted on positively charged slides. Slides were deparaffinized with Histoclear and hydrated with 100% ethanol, 75% ethanol, 50% ethanol, and then distilled water. To achieve the antigen retrieval, slides were steamed in Citrate buffer at pH 6 (DAKO Target retrieval buffer S1699, DAKO) for 30 min and subsequently cooled for 30 min. Furthermore, to inhibit the endogenous peroxides, slides were quenched with 3% H 2 O 2 for 10 min. Slides were washed with distilled water, then 1Â PBS and blocked with 1% BSA in 1Â PBS for 1 hr. After washing with 1Â PBS, slides were incubated with p-eNOS antibody (1:500 dilution) overnight in the humidified chamber at 4˚C. Slides were washed two times with 1Â PBS and incubated with secondary antibody (AF488, Invitrogen, 1:1000 dilution) for 1 hr. Subsequently slides were washed with 1Â PBS three times and mounted with DAPI added mounting media with coverslip. The slides were imaged by Zeiss LSM700 confocal microscope (10Â objective).

Ex vivo sprouting angiogenesis
Following Avertin injection and cervical dislocation, aortas were dissected and connective tissue removed, and then washed with PBS with 50 mg/ml penicillin and streptomycin. Using iris scissors, the aorta was cut into aortic rings of 1-2 mm cross-sectional slices; 50 ml of Matrigel was used to coat the center of coverslips in 24-well plates for 2 hr at 37˚C in the incubator to solidify the Matrigel. Aorta rings were then seeded and transplanted on Matrigel (BD Biosciences, Cat#356231) on coverslips. After seeding the aortic rings, plates were incubated at 37˚C without medium for 10 min to allow the ring to attach to the Matrigel. Complete medium was added to each well and incubated at 37˚C with 5% CO 2 for 48-72 hr. Phase contrast photos of individual explants were taken using a 10Â/0.75 NA objective Olympus IX73 microscope (Olympus, Japan) fitted with camera (Orca flash 4.0+, Hamamatsu, Japan). Cells were incubated with LRRC8A (1:250) and CD31 (1:250) primary antibodies in 0.1% Tween-20% and 1% BSA overnight, and then incubated with secondary antibodies (Invitrogen, Anti-Rabbit 488, A11070; Anti-Mouse 568, A11019). Cells were then incubated for 10 min with Topro 3 (T3605, Thermofisher) at RT.

Retinal imaging
Ketamine (Akorn Animal Health, 100 mg/ml) was prepared and mixed with xylazine, then stored at RT. Animals were anesthetized with 87.5/12 mg/kg body weight via i.p. injection. Eyes were topically anesthetized with proparacaine and dilated with tropicamide. Fluorescein (100 mg/ml, Akorn Inc) diluted with sterile saline was administered by i.p. injection (50 ml), and mice were positioned on Micron imaging platform. Images of the eyes were taken from the start of fluorescein infusion with the Micron camera with a 450-650 nm excitation filter and 469-488 nm barrier filter at 30 frame/s using Micron software for 30 s. Data were converted into tiff image for further analysis.

Ang II infusion
Infusion studies were carried out using Azlet osmotic minipumps (Model 1004). Ang II (BACHEM) dissolved in saline was filled in the minipumps and were prepared to maintain infusion rate of 600 ng/kg/min for 4 weeks. The mice were anesthetized under 2% isofluorane and the minipumps were implanted subcutaneously on the dorsal aspect of the mice.

Blood pressure recordings
Systolic tail-cuff blood pressure (BP) measurements were carried out using computerized tail-cuff system BP-2000 (Visitech Systems) at the same time of day. Mice were first acclimated to the device by performing 3 days of measurements (20 sequential measurements/day) and then mean blood pressure readings were obtained by averaging 3-5 days of measurements (not inclusive of the 3 acclimation days).

RNA sequencing
RNA quality was assessed by Agilent BioAnalyzer 2100 by the University of Iowa Institute of Human Genetics, Genomics Division. RNA integrity numbers greater than eight were accepted for RNAseq library preparation. RNA libraries of 150 bp PolyA-enriched RNA were generated, and sequencing was performed on a HiSeq 4000 genome sequencing platform (Illumina). Sequencing results were uploaded and analyzed with BaseSpace (Illumina). Sequences were trimmed to 125 bp using FASTQ Toolkit (Version 2.2.0) and aligned to Homo sapiens HG19 (refseq) using RNA-Seq Alignment (Version 1.1.0). Transcripts were assembled and differential gene expression was determined using Cufflinks Assembly and DE (Version 2.1.0). Ingenuity Pathway Analysis (QIAGEN) was used to analyze significantly regulated genes which were filtered using cutoffs of >1.5 fragments per kilobase per million reads, >1.5-fold changes in gene expression, and a false discovery rate of <0.05. Heatmaps were generated to visualize significantly regulated genes. Data have been deposited in GEO (acces-sion# TBD).

Metabolic phenotyping
Mice were fasted for 6 hr prior to glucose tolerance tests (GTTs). Baseline glucose levels at 0 min timepoint (fasting glucose) were measured from blood samples collected from tail bleeds using a glucometer (Bayer Healthcare LLC). D-Glucose was injected i.p. (0.75 g/kg body weight) and glucose levels were measured at 7, 15, 30, 60, 90, and 120 min timepoints after injection. For insulin tolerance tests (ITTs), the mice were fasted for 4 hr. Similar to GTTs, the baseline blood glucose levels were measured at 0 min timepoint and 15, 30, 60, 90, and 120 min timepoints post-injection (i.p.) of insulin (HumulinR, 1.25 U/kg body weight). Mouse body composition was performed as previously described .

Statistics
Data are represented as mean ± s.e.m. Two-tail unpaired Student's t-tests were used for comparison between two groups. For three or more groups, data were analyzed by one-way analysis of variance (ANOVA) and Tukey's post hoc test. For time-series data in Figure 5 between two groups, data were analyzed using a two-way ANOVA followed by Bonferroni post hoc test. For GTTs and ITTs, two-way ANOVA was used. A p-value<0.05 was considered statistically significant. *, **, and *** represent a p-value less than 0.05, 0.01, and 0.001, respectively.