All chemicals without annotation in brackets were purchased from Sigma-Aldrich in this study.

140 male C57BL/6 (wild type, 8–10 weeks old) mice were purchased from Jackson Laboratory and were housed 1 week for adaption before experiments. The feeding and surgical operations of mice were carried out in SPF conditions. All animal protocols were approved by the Animal Care and Use Committee of University of Nevada, Reno. This investigation conforms to the Guide for the Care and Use of Laboratory Animals (NIH).

Pressure-overload-induced CH and HF mouse models were developed through a modified aortic banding (AB) surgery (Rockman et al., 1991). Briefly, mouse was intubated and ventilated by a mouse ventilator (Harvard Apparatus) with 2% isoflurane contained air. A small midline chest incision was made to expose the aorta and branching arteries, and aortic constriction was performed on position between brachiocephalic trunk artery and left carotid artery by tying a 8–0 silk suture ligature against a 27-gauge needle to yield a narrowing 0.4 mm in diameter when the needle was removed. Sham control mice underwent chest incision surgery but the aortas were not narrowed.

To evaluate the parameters of mouse heart, echocardiographic measurements were performed using Vevo^® 2100 imaging system (Visualsonics) with MS-550D ultrasound transducer according to the manuals every week after surgery. In brief, mouse was anesthetized with isoflurane (2%) and fixed to a warming platform in a supine position. The heart rate monitored by electrocardiogram was kept from 450 to 500 per minute. Both B-mode and M-mode of parasternal long-axis and short-axis views of left ventricle (LV) were recorded at a frame rate of 200 Hz. All data analysis was done using Vevo^® LAB desktop software.

Left ventricular myocytes (LVMs) were isolated by an enzymatic dispersion technique (Xu et al., 2002). In brief, mouse was killed by cervical dislocation, and heart was rapidly excised. Aorta was cannulated to the syringe tip of Langendorff apparatus, and retrograde perfusion of the aorta with various buffer solutions gassed with 100% O₂ and maintained at 37°C was performed. Heart was firstly perfused with Tyrode solution (in mM: 125 NaCl, five MgCl₂, 4.5 KCl, 10 HEPES, one NaH₂PO₄, five pyruvic acid, 20 taurine, 10 glucose, 1.5 CaCl₂, pH 7.4 with NaOH) until the blood was washed, then perfused with Ca²⁺-free Tyrode solution for 5 min, followed by Ca²⁺-free Tyrode solution containing 0.4 mg/mL collagenase type II (Wirthington) and 1 mg/mL albumin for 10–15 min. LV was separated and soaked in Kreb’s solution (in mM: 20 KCl, one MgCl₂, 10 KH₂PO₄, 70 l-glutamic acid, 10 hydroxybutyric acid, 10 taurine, 10 EGTA, 10 glucose, 1 mg/mL albumin, pH 7.4 with KOH). LVMs can be dispersed by suction the solution repeatedly with straw.

Rod-shaped LVMs exhibited clear cross striations were used in patch-clamp experiments which were carried out at room temperature (20–25°C) using an Axopatch 200B amplifier (Axon Instruments). Data were filtered at 1 kHz and sampled at 5 kHz. Borosilicate glass electrodes had 3–5 MΩ tip resistances when filled with pipette solution (in mM: 110 potassium aspartate, 20 KCl, one MgCl₂, five ATP-Mg, 0.1 GTP, 10 HEPES, five Na₂-phosphocreatine, 0.05 EGTA, pH 7.4 with KOH). For action potential (AP) recordings, the bath solution contained (in mM) 126 NaCl, two CaCl₂, 5.4 KCl, 0.8 MgCl₂, 0.33 NaH₂PO₄, 10 dextrose and 10 HEPES (pH 7.4 with NaOH). For whole-cell current recordings, Na⁺ in bath solution was replaced with NMDG to block Na⁺-Ca²⁺ exchange current. The equilibrium potential for Cl⁻ (E_Cl) was -48 mV. Pipette resistance was compensated. Cell membrane capacitance (C_m) was determined by integrating the capacity transient elicited by a 5 mV hyperpolarizing pulse from a holding potential of -50 mV. C_m and series resistance were compensated as much as possible. LVMs with resting membrane potentials between -60 and -80 mV were used.

A phenomenon known as “rundown” is the gradually disappearance of I_Ca when cells are dialyzed or perfused. Cardiac cell desires 15–30 s intervals between depolarization pulses to allow recovery of ionic currents and intracellular Ca²⁺ metabolism, but rundown is accelerated by perfusion or pulses that elicit I_Ca (Zygmunt and Gibbson, 1991). In addition, depolarization sometimes causes contraction of cardiac cell and leads to “seal-lost”. Here, in order to save time and to reduce depolarizing stimuli, whole-cell currents were elicited by depolarizing voltage steps (200 ms) from -20 to +60 mV in 20 mV increments from a holding potential of -50 mV to determin I-V relations. When comparing I_TMEM16A densities of LVMs between sham and AB mice, currents were activated by voltage steps from a holding potential of -50 to +60 mV. TMEM16A-specific potent inhibitor T16A_inh-A01, and Anti-TMEM16A antibody (ab53212, Abcam) with synthetic peptide that detects the extracellular domain amino acids 628–731 of TMEM16A (including the predicted pore-forming domain) were used separately as I_TMEM16A inhibitors. All peak current values were normalized to C_m and reported as current densities (pA/pF). APs were elicited by 5 ms current pulses (500 pA) applied via the patch pipette at stimulus frequencies of 1 Hz.

HUVECs were isolated and cultured by a modified sterile technique described previously (Jaffe et al., 1973). The experiments were approved by the medical research ethics committee of Dalian Medical University and conformed to the principles expressed in the Declaration of Helsinki. Briefly, umbilical veins were perfused by D-Hanks buffer (in g/L: 8.0 NaCl, 0.4 KCl, 0.06 KH₂PO₄, 0.35 NaHCO₃), and HUVECs were harvested from the umbilical veins digested by 0.25% trypsin (Beyotime) with 0.02% EDTA. Then cells were cultured in ECM medium (sciencell 1001, Solarbio) containing 5% fetal bovine serum, 1% ECGs, 0.1% Savelt, 100 U/mL penicillin and 100 mg/mL streptomycin at 37°C, 5% CO₂ humidified atmosphere. Cells between passages three and five were used in this study.

Three kinds of siRNA (#1, #2 and #3) against human TMEM16A were designed and constructed by GenePharma (China). siRNA transfection of HUVECs was performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. Negative siRNA (GenePharma) was used as negative control (siRNA^NC). The sequences of siRNA were as follows: siRNA #1, sense (5′- CCG​GAG​CAC​GAU​UGU​CUA​UTT -3′), antisense (5′- AUA​GAC​AAU​CGU​GCU​CCG​GTT -3′); siRNA #2, sense (5′- GGA​AAC​AGA​UGC​GAC​UCA​ATT -3′), antisense (5′- UUG​AGU​CGC​AUC​UGU​UUC​CTT -3′); siRNA #3, sense (5′- GCU​GUC​AAG​GAU​CAU​CCU​ATT -3′), antisense (5′- UAG​GAU​GAU​CCU​UGA​CAG​CTT -3′); siRNA^NC, sense (5′- UUC​UCC​GAA​CGU​GUC​ACG​UTT -3′), antisense (5′- ACG​UGA​CAC​GUU​CGG​AGA​ATT -3′). Briefly, cells at 70% confluent were starved with serum-free ECM for 1 h, and treated with siRNA/lipofectamin Opti MEM (Invitrogen) for 4–6 h. After cultured with complete ECM for 72 h, HUVECs were harvested for the following experiments.

Lentivirus which can induce mouse TMEM16A overexpression (TM^OE) was designed and constructed by Genechem (China). Lentivirus transfection of HUVECs was performed with polybrene (5 µg/mL) according to the manufacturer’s protocol. Negative lentivirus (Genechem) was used as negative control (TM^NC). In brief, cells at 30–50% confluent were treated with lentivirus/polybrene complete ECM for 8–12 h and then cultured with complete ECM for 72 h. HUVECs were harvested for the following experiments.

Total RNAs were extracted from mice LV or HUVECs using TRIzol reagent (Invitrogen), qualified and quantified by a Nanodrop 2000 spectrophotometer. Equal amounts (0.5 µg) of total RNAs were reverse transcribed with SuperScript IV RT reagent Kit (Invitrogen).

The expression levels of mRNAs in LV of mice were detected on 7900 HT Real-time PCR System by RT² qPCR Primer Assays (Qiagen) according to the manufacturer’s protocol. The relative expression level of TMEM16A mRNA was calculated using the 2^-∆∆Ct method with GAPDH mRNA as a reference. Primers for mTMEM16A (Catalog no 330001 PPM26917B) and mGAPDH (Catalog no 330001 PPM02946E) were purchased from Qiagen.

The expression levels of mRNAs in HUVECs were detected on 7900 HT Real-time PCR System with TransStart Top Green qPCR SuperMix (TransGen Biotech) according to the manufacturer’s protocol. The relative expression level of TMEM16A mRNA was calculated using the 2^-∆∆Ct method with GAPDH mRNA as a reference. The sequences of primers were as follows: hTMEM16A, forward (5′- TGA​AAC​TGA​AGA​TGC​CGA​CG -3′), reverse (5′- AGG​AGA​GTC​TCT​TCA​TGG​TCT​G -3′); mTMEM16A, forward (5′- GAG​GCC​AGT​AGC​CAT​CAG​AG -3′), reverse (5′- GAG​AGC​GTG​TGA​TTG​ACG​AA -3′); hGAPDH, forward (5′- AGG​GCT​GCT​TTT​AAC​TCT​GGT -3′), reverse (5′- CCC​CAC​TTG​ATT​TTG​GAG​GGA -3′).

Total membrane protein was extracted from mice LV using Plasma Membrane Protein Extraction Kit (101 Bio) according to the manufacturer’s protocol. Total protein was extracted from HUVECs using RIPA lysis buffer (Solarbio) according to the manufacturer’s protocol.

Briefly, HUVECs were incubated to grow into full confluence on 12-well plates. A wound was made in the cell monolayer in each well using a pipette tip. Serum free ECM with vehicle (PBS) or 30 ng/mL VEGF, or 1 μM AngII was added. Images of the wounds were taken under microscope after 3 h, 6 h and 12 h of incubation. Wound area was measured using ImageJ software and wound closure percentage was calculated to assess endothelial cell migration.

Tube formation assay was performed by a modified technique as described (Shi et al., 2013). In brief, HUVECs (2 × 10⁴ in 50 μL ECM with 0.1% BSA) were seeded on Ibitreat angiogenesis slides (Ibidi, Martinsried) pre-coated with 10 μL Matrigel. After incubation with serum free ECM with vehicle (PBS) or 30 ng/mL VEGF, or 1 μM AngII for 6 h, images were taken under microscope. Total branches length was quantified using ImageJ software to assess tube formation.

Endothelial cell spheroids sprouting assay was performed by a modified technique as described (Tang et al., 2016). Briefly, cell spheroids were generated overnight in hanging-drop culture consisting of 400 HUVECs in ECM and 20% methylcellulose. Spheroids were harvested and embedded in collagen type I gels (1 × M199 medium (Gibco), 1500 μg/mL collagen type I (BD Biosciences), 0.22% NaHCO₃, pH 7.4 with NaOH). Then spheroids-containing gel was rapidly transferred into a 24-well plate and allowed to polymerize. Equal volume of 2 × MCDB medium (Gibco) with vehicle (PBS) or 60 ng/mL VEGF, or 2 μM AngII was added on top of the gel. After incubation for 24 h, images were taken under microscope, and total capillary-like sprouts length per spheroid was quantified using ImageJ software to assess capillary sprouting.

Data were presented as mean ± SEM and analyzed by GraphPad Prism software. Statistical significance was determined by Student’s t-test or by one-way ANOVA followed by Bonferroni posttest, or by two-way ANOVA followed by a post hoc test (Bonferonni) for multiple comparisons. In some instances, sample-matched repeated measures (RM) ANOVA was used, as appropriate. p < 0.05 was considered statistically significant.

To evaluate the development of CH and HF, echocardiographic measurements were performed weekly until 12 weeks after surgery. As shown in Supplementary Figure S1, compared with sham control, AB mice showed significant increases in LV wall thickness, LV mass and LV mass index 1 week or 3 weeks postoperatively, suggesting that there was a compensatory increase in LV muscle responding to pressure-overload and CH occurred. From 1 to 12 weeks after surgery, both LV mass and LV mass index in AB mice increased continuously with time, and were significantly higher than those in sham group, indicating CH developed continuously. There was a slight decrease in dimension of the LV cavity in AB group at first; from 4th week, LV cavity dimension continued increasing and was significantly larger than sham group from 7th week, suggesting pathologic growth of the myocardium induced concentric remodeling of the ventricle transitioned to an eccentric remodeling (dilation). The average value of LV wall thickness in AB group reached a maximum at 8th week, and then decreased in the next few weeks. As shown in Supplementary Figure S2, in AB group, both LV short-axis shortening fraction and ejection fraction significantly decreased from 3th week, suggesting that LV systolic function significantly decreased. The LV chamber enlargement with loss of wall as well as destroyed systolic function are clinical features of HF.

There was no significant difference in TMEM16A mRNA expression level or TMEM16A protein expression level in LV between sham and AB mice within 12 weeks after surgery (Figure 1). The molecular weight (MW) of TMEM16A is usually supposed to be 114 kDa, but in simple western blot assay, the peak with MW of 186 kDa was considered as TMEM16A (Figure 1C). It is reasonable according to the following facts. 1) In simple western assay, LV lysates mixed with different antibodies were separately loaded in different capillaries. Only one peak (186 kDa) emerged in the lane mixed with Anti-TMEM16A antibody, no peak existed in lane without antibody mixing. 2) The MW values marked in simple western assay may be higher than real values, since the peak represented for GAPDH here was marked 41 kDa, however 36 kDa was generally reported as the MW of GAPDH. 3) The MW of mTMEM16A overexpressed in HUVECs is nearly 180 kDa (Figure 5B). 4 Two MW values (85 kDa and 146 kDa) of TMEM16A exist in colon ECs from guinea pigs (He et al., 2011). The MW of TMEM16A in neonatal mice cardiac vascular ECs is between 130 and 170 kDa (Wu et al., 2014). 5) TMEM16A has different splicing isoforms, has species-specific and tissue-specific characteristics, and has post-translational modifications such as glycosylation as a functional membrane protein.

In this study, whole-cell I_to-voltage (I-V) relations were determined using a voltage-clamp protocol (shown in the inset of Figure 2) to save time and to reduce depolarizing stimuli. In cardiac cells, I_to responsible for repolarization during phase one of AP is composed of I_Cl.Ca and 4-aminopyridine (4-AP) sensitive I_K. As shown in Figure 2A, I_to were partly inhibited after adding T16A_inh-A01 (30 μM), and were greatly inhibited after adding T16A_inh-A01 along with 4-AP (5 mM) to the bath solution. The inhibition of I_to by Anti-TMEM16A antibody (1:200, adding extracellular) was similar to that by T16A_inh-A01 (Figures 2A–C). Both T16A_inh-A01 sensitive current and Antibody sensitive current were voltage-dependent, and their densities were much smaller than 4-AP sensitive current densities under the same depolarizing voltage (Figures 2C,D).

As shown in Figure 3A, the superimposed current tracings were activated by voltage steps from a holding potential of -50 to +60 mV in the absence (control) and presence of T16A_inh-A01 (30 μM) in the same cell. The peak I_to difference was measured as the T16A_inh-A01 sensitive current (I_TMEM16A) value. The I_TMEM16A densities were not significantly different in LVMs isolated from sham or AB mice within 12 weeks after surgery (Figure 3B).

Inhibition of TMEM16A by T16A_inh-A01 (30 μM) didn’t significantly alter the AP shape or APD₉₀, whereas inhibition of K⁺ channels by 4-AP (5 mM) significantly prolonged APD₉₀ in LVMs (Figures 4A,B). LVMs of AB mice showed significant prolongation of APD₉₀ 4 weeks postoperatively compared with sham LVMs (Figure 4C). TMEM16A inhibition by T16A_inh-A01 insignificantly influenced the APD₉₀ in LVMs of sham or AB mice within 12 weeks after surgery (Figure 4C).

Equal proportion mixtures of three kinds of siRNA against TMEM16A were labeled as siRNA^Mix. As shown in Figures 5A,B, siRNA against TMEM16A (#1, #2, #3 and Mix) transfection significantly decreased the TMEM16A mRNA and protein expression level, whereas TM^OE transfection significantly increased the TMEM16A molecular expression levels in HUVECs. Two clear bands appeared on TM^OE lane (Figure 5B) in western blot assay, the upper thicker band represented overexpressed mouse TMEM16A protein induced by lentivirus, the down thinner band represented endogenous human TMEM16A protein expressed in HUVECs.

Knockdown of TMEM16A by siRNA^Mix transfection significantly decreased the wound closure percentage (Figure 6A, left), overexpression of TMEM16A by TM^OE transfection significantly increased the wound closure percentage (Figure 6B, left), suggesting that TMEM16A is a positive regulator of migration in HUVECs. Compared with vehicle (PBS), 30 ng/mL VEGF strongly promoted the wound healing progression in HUVECs (Figures 6A,B, left and middle). TMEM16A knockdown significantly decreased the migration promoting effect of VEGF (Figure 6A, middle), TMEM16A overexpression significantly increased the migration promoting effect of VEGF (Figure 6B, middle), suggesting that TMEM16A is a positive regulator of the migration promoting effect of VEGF in HUVECs. 1 μM AngII significantly inhibited the wound healing progression in HUVECs compared with vehicle (PBS) (Figures 6A,B, left and right). TMEM16A is a negative regulator of migration inhibiting effect of AngII in HUVECs since TMEM16A knockdown enhanced (Figure 6A, right) and TMEM16A overexpression attenuated (Figure 6B, right) the migration inhibiting effect of AngII.

As shown in Figures 7A–D, TMEM16A knockdown significantly decreased the total branches length in tube formation assay and the total sprouts length in cell spheroids sprouting assay, TMEM16A overexpression exhibited opposite effects, suggesting that TMEM16A is a positive regulator of angiogenesis in HUVECs. 30 ng/mL VEGF significantly promoted angiogenesis, whereas 1 μM AngII significantly inhibited angiogenesis according to the results of both tube formation assay and cell spheroids sprouting assay. TMEM16A knockdown significantly decreased and TMEM16A overexpression significantly increased the pro-angiogenesis effect of VEGF, suggesting that TMEM16A is a positive regulator of pro-angiogenesis function of VEGF in HUVECs. TMEM16A knockdown significantly enhanced and TMEM16A overexpression significantly attenuated the anti-angiogenesis effect of AngII, suggesting that TMEM16A is a negative regulator of anti-angiogenesis function of AngII in HUVECs.