Reducing Endoglin Activity Limits Calcineurin and TRPC‐6 Expression and Improves Survival in a Mouse Model of Right Ventricular Pressure Overload

Reagents

Polyclonal Abs against human calcineurin, α‐SMA, phosphorylated (p)Smad3, and total Smad2/3 were purchased from Cell Signaling Technology (2614S; Danvers, MA), Sigma‐Aldrich (A2547; St. Louis, MO), and Cell Signaling Technology (8769S and 3102S), respectively. Polyclonal Abs against mouse endoglin, type I collagen, and α‐SMA were purchased from R&D Systems (BAF1320; Minneapolis, MN), Santa Cruz Biotechnology (SC‐25974; Santa Cruz, CA), and Sigma‐Aldrich (A2547), respectively. Rabbit polyclonal Abs to mouse calcineurin (2614S) were purchased from Cell Signaling Technology. Polyclonal Abs to mouse pSmad3 (9520), and phosphorylated extracellular signal‐regulated kinases 1 and 2 (pERK1/2; SC‐134900) were purchased from Cell Signaling Technology and Santa Cruz Biotechnology. Polyclonal Abs to mouse total Smad3 (SC‐101154) and total ERK (SC‐135900) were purchased from Santa Cruz Biotechnology. Sugen (SU5416) was purchased from Sigma‐Aldrich. A neutralizing immunoglobulin G (IgG)1 Ab that binds human and mouse endoglin (TRC105) was provided by Tracon Pharmaceuticals (San Diego, CA). An ELISA kit for the detection of active TGF‐β1 levels in mice was purchased from R&D Systems.

Mouse Model of Pharmacologically Induced RVPO

Animals were treated in compliance with the Guide for the Care and Use of Laboratory Animals (National Academy of Science), and protocols were approved by the Tufts Medical Center Institutional Animal Care and Use Committee (Boston, MA). Adult, male, 12‐ to 14‐week‐old C57BL/6 WT and congenic Eng^+/− mice received once‐weekly intraperitoneal injections of Sugen and were exposed to either normoxic conditions (room air) or chronic normobaric hypoxia (10% O₂), as previously described.²⁸ After 5 weeks of exposure to either Sugen+Normoxia (Su‐Norm) or Sugen+Hypoxia (Su‐Hypox), mice underwent hemodynamic analysis with a RV conductance catheter (Millar Instruments Inc., Houston, TX), as described below, and tissue was then obtained for further analysis. Eng^+/− mice were generously provided by Dr Michelle Letarte, University of Toronto (Toronto, Ontario, Canada).

Mouse Model of Surgically Induced RVPO

Adult, male, 12‐ to 14‐week‐old C57BL/6 WT and congenic Eng^+/− mice underwent pulmonary artery constriction (PAC), as previously described.^27,29 Specifically, mice were intubated using a 24G angiocath and mechanically ventilated (Harvard Apparatus, Holliston, MA) at 95 breaths per minute with a tidal volume of 0.3 mL with 2.0% to 2.5% Isoflurane and 100% flow‐through oxygen. Depth of anesthesia was monitored by assessing palpebral reflex, toe pinch, respirations, and general response to touch. Using a sterile technique, a left thoracotomy was performed to isolate and encircle the main pulmonary artery using a 7‐0 nylon suture that was then tied tightly around a presterilized, blunt‐end needle. After deairing, the thorax was closed with layered 6‐0 Dexon sutures to eliminate the risk of pneumothorax. Postoperative analgesia was immediately provided with buprenorphine (0.1 mL), which was continued twice‐daily and as needed for an additional 72 hours. Severe RVPO was induced by PAC with a 25G needle for 7 days in WT and Eng^+/− mice. To investigate the role of endoglin in RVPO, WT mice received 15 mg/kg of either a neutralizing Ab to endoglin (N‐Eng; TRC105; Tracon Pharmaceuticals) or an IgG1 control Ab (IgG Ab; R&D Systems) by single intraperitoneal injection 1 day before and 3 days after induction of severe RVPO. To study the effect of blocking endoglin activity after induction of RVPO, WT mice were randomized to receive biweekly intraperitoneal injections for 3 weeks of 15 mg/kg of N‐Eng antibody or IgG control Ab beginning 3 weeks after induction of moderate RVPO using a 23G needle for PAC. The Ab dose was based on a previous clinical study demonstrating effective saturation of endoglin receptors.³⁰ After 7 days of severe PAC or 3 to 6 weeks of moderate RVPO, mice underwent hemodynamic analysis with an RV conductance catheter (Millar Instruments), as described below, and tissue was then obtained for further analysis.

Nuclear Factor of Activated T‐Cell Activity In Vivo

Nuclear factor of activated T‐cell (NFAT)‐luciferase (NFAT‐Luc) mice with nine copies of an NFAT‐binding site from the interleukin (IL)‐4 promoter (5′‐TGGAAAATT‐3′) inserted upstream of the luciferase reporter gene, driven by the α‐myosin heavy‐chain promoter,¹⁷ were purchased from The Jackson Laboratory (Bar Harbor, ME). Eng^+/−‐NFAT luciferase reporter mice were generated by crossing Eng^+/− mice with the NFAT‐luciferase mice. Severe RVPO was induced by PAC in 10‐ to 12‐week‐old Eng^+/+‐NFAT‐Luc and Eng^+/−‐NFAT‐Luc. After 7 days of severe PAC, RV tissue was then obtained for quantification of luciferase activity using firefly luciferase assays that were carried out as follows: 20 μL of whole RV tissue lysate was added to 100 μL of firefly luciferase assay buffer (Promega, Madison, WI). Samples were placed in a luminometer (Luminoskan Ascent; Labsystems Oy, Helsinki, Finland), and luminescence was determined in triplicate per sample over a 10‐second interval.

Hemodynamic Assessment of RV Function

All animals underwent terminal hemodynamic evaluation. Right heart catheterization was performed at the time of sacrifice in all animals. Mice were anesthetized with 2.0% isoflurane administered by a noninvasive nose cone. Body temperature was monitored by a rectal thermistor probe and maintained at 37.5°C with heating pads and a cycling heat lamp. In the supine position, the right common carotid and right external jugular vein were surgically isolated. Silk ties were placed at the distal ends of both vessels while overhand loops were placed at the proximal ends with 7‐0 nylon. A Millar PVR‐1035 (Millar Instruments) mouse conductance catheter was used for RV recordings. Before insertion, conductance catheter calibration was performed using the cuvette method with freshly heparinized warm blood, then zeroed in warm saline as previously described.^31–32 A transverse venotomy was performed using iris scissors at the proximal end of the external jugular vein. The PVR‐1035 catheter was advanced through the superior vena cava and right atrium into the RV, leaving the chest wall intact. Once hemodynamic stability was achieved, steady‐state baseline conditions were recorded from the RV. Stroke volume was calculated as end‐diastolic minus end‐systolic volume. Arterial elastance was calculated under steady‐state conditions as end‐systolic pressure/stroke volume. Tau, a measure of instantaneous isovolumic relaxation, was calculated using the Glantz method as P(t)=P₀e^−t/τ_E+ Pα, where P is pressure at time t, P₀ is the amplitude constant, τ_E is the Glantz relaxation constant, and Pα is the nonzero asymptote resulting from pleural and pericardial pressure. RV compliance was calculated as stroke volume divided by peak RV pressure. Pressure‐volume loop acquisition and analysis was performed using IOX software (emka TECHNOLOGIES, Paris, France). After completion of the hemodynamic study, with the animal still under isoflurane anesthesia, the chest was rapidly opened, and the mouse was euthanized by arresting the heart in diastole with 0.3 mL of 1 N of KCL injected directly into the LV. The heart was then removed and processed for either biochemical or histologic analyses.

Histologic Quantification of Cardiac Hypertrophy and Fibrosis

RV collagen abundance was quantified by picrosirius red staining, as previously described.³³ Cardiomyocyte cross‐sectional area was quantified as previously described.³⁴

Loss‐of‐Function Studies in Cardiac Fibroblasts

Human RV (RVFB) and LV (LVFB) fibroblasts were isolated from myocardial tissue harvested during cardiac surgery at Tufts Medical Center, and mouse RVFB and LVFB were isolated from WT and Eng^+/− mice. Fibroblasts were stimulated with TGF‐β1 for analysis, as previously described.^6,35–36 For calcineurin inhibition studies, human RVFB were pretreated with 5 nM of cyclosporine A (CsA) or vehicle control for 24 hours in fibroblast basal medium (FBM) without supplementation, followed by stimulation with TGF‐β1 (10 ng/mL) for 24 hours.³⁷ For TRPC‐6 silencing experiments, 50 μmol/L of siRNA stock was diluted to 5 nmol/L in Optimem (Invitrogen, Carlsbad, CA) and combined with 2 μL of Lipofectamine (Invitrogen) diluted in 98 μL of Optimem. After 20 minutes of incubation, cells were exposed to human TRPC‐6 siRNA (Catalog No.: 439420; Ambion, Austin, TX) or scrambled siRNA (negative control; Catalog No.: 4390844; Ambion). After 48 hours after transfection, cells were treated with TGF‐β1 (10 ng/mL) for 16 to 24 hours, then harvested for analysis. For neutralizing Ab studies in vitro, human RVFB and LVFB were pretreated with 10, 50, or 100 μg/mL of either an N‐Eng Ab or control IgG Ab for 24 hours in FBM before stimulation with TGF‐β1 (10 ng/mL). After 24 hours, cells were harvested for analysis. The Ab dose was based on previous studies demonstrating effective neutralization of endoglin activity in endothelium.³⁰ All RVFB and LVFB stimulation studies were conducted in triplicate with cells cultured to within three lineage passes only.

Real‐Time Quantitative Polymerase Chain Reaction

For all cell‐based reverse‐transcription polymerase chain reaction (RT‐PCR) experiments, total RNA was extracted directly using Trizol (Invitrogen), then converted to cDNA using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). For all RT‐PCR experiments, samples were quantified in triplicate using 40 cycles performed at 94°C for 30 seconds, 60°C for 45 seconds, and 72°C for 45 seconds using an ABI Prism^® 7900 (Applied Biosystems Inc) Sequence Detection System with appropriate primers (Table 1), as previously described.^6,34

Immunoblot Analysis (Western)

Total protein was extracted and quantified from tissue homogenates or cultured cells, as previously described.^6,35 Immunoblot analysis was then performed, as previously described, using Abs for mouse targeted proteins.^6,35

Statistical Analysis

Results are presented as mean±SD. Intergroup comparisons were made for each variable with a two‐way ANOVA, followed by a post‐hoc Bonferonni's correction for multiple comparisons (Tables 2 through 5). An alpha level of P<0.01 was considered to indicate a significant effect or between‐groups difference. Multiple comparisons versus a control group were performed using one‐way ANOVA, followed by post‐hoc analysis with Dunnett's method (Figures 1 through 7). Kaplan‐Meier's analysis with log‐rank testing was employed for survival analysis (Figures 4C and 7A). All statistical analyses were performed using SigmaStat software (Version 3.1; Systat Software Inc., Richmond, CA). An alpha level of P<0.05 was considered to indicate a significant effect or between‐groups difference.

Endoglin Promotes TGF‐β1‐Mediated Calcineurin Expression and Myofibroblast Conversion in RV Fibroblasts

We first explored a role for calcineurin as a mediator of myofibroblast transformation in the RV. Human RVFB were stimulated with TGF‐β1 in the presence or absence of the calcineurin inhibitor, CsA. Pretreatment with cyclosporine attenuated TGF‐β1‐mediated increases in protein and mRNA levels of calcineurin and α‐SMA (Figure 1A through 1C). TGF‐β1 stimulation also increased TRPC‐6 mRNA expression in human RVFB, which was prevented by cyclosporine treatment (Figure 1D). To examine the role of TRPC‐6 in RV myofibroblast transformation, we employed a siRNA against TRPC‐6 (siTRPC‐6), which achieved a greater than 75% knockdown of TRPC‐6 protein expression in RVFB (Figure 1E). Silencing TRPC‐6 attenuated TGF‐β1‐mediated up‐regulation of calcineurin and α‐SMA in human RVFB (Figure 1F). These data indicate that TGF‐β1 increases expression of TRPC‐6 and α‐SMA in a calcineurin‐dependent manner in human RV fibroblasts.

Next, to explore the dependence of calcineurin expression on endoglin in human RVFB and LVFB, cells were treated with TGF‐β1 in the presence of increasing concentrations of an N‐Eng Ab (Figure 2A through 2B). Neutralizing endoglin activity blocked TGF‐β1‐induced calcineurin and α‐SMA protein expression in human RVFB, not LVFB. Neutralizing endoglin attenuated pSmad3 expression in both human RVFB and LVFB. To further examine the role of endoglin as a positive regulator of calcineurin in RVFB and LVFB, cells were isolated from WT and Eng^+/− mice (Figure 2C through 2D). In WT RVFB and LVFB, TGF‐β1 stimulated calcineurin and α‐SMA mRNA expression, which was prevented in Eng^+/− RVFB, not LVFB (Figure 2C). TGF‐β1 also up‐regulated protein levels of calcineurin and α‐SMA in WT RVFB and LVFB, which was attenuated in Eng^+/− RVFB, but not in Eng^+/− LVFB (Figure 2D). These findings support that endoglin is required for TGF‐β1‐induced calcineurin expression and myofibroblast transformation in RVFB, a critical component of cardiac fibrosis.

Reduced Endoglin Expression Preserves RV Function and Limits RV Fibrosis in a Model of Angio‐Obliterative Pulmonary Hypertension

To begin exploring a functional role for endoglin in RV remodeling, we studied the well‐established model of angio‐obliterative pulmonary hypertension induced by exposure to hypoxia and the anti‐vascular endothelial growth factor compound, Sugen, in WT, compared to Eng^+/−, mice. All mice survived treatment with Sugen+Hypoxia for 5 weeks, and no significant change in total body weight, RV or LV weights, RV stroke volume, or cardiac output was observed between groups (Table 2). We observed increased RV systolic pressure (RVSP) in both WT and Eng^+/− mice after 5 weeks of exposure to Sugen+Hypoxia (Figure 3A). No difference in RV dP/dtmax was observed between WT and Eng^+/− groups treated with Sugen+Hypoxia, demonstrating a similar response to RVPO in both types of mice. WT mice exposed to Sugen+hypoxia developed evidence of abnormal diastolic RV function, including increased Tau (a measure of instantaneous isovolumic relaxation) and decreased RV compliance (Figure 3B and 3C), whereas Eng^+/− mice demonstrated no change in Tau and relatively preserved RV compliance. To explore the mechanism for the differences in RV diastolic function, we examined RV fibrosis and calcineurin signaling. Exposure to Sugen+Hypoxia increased type I collagen mRNA expression and histologic levels of collagen abundance in WT, not Eng^+/−, mice (Figure 3D through 3F). Calcineurin, TRPC‐6, and α‐SMA mRNA levels were increased by Sugen+Hypoxia in WT, not Eng^+/−, mice (Figure 3G through 3I). These findings suggest that, despite identical degrees of RVPO, reduced endoglin expression in Eng^+/− mice preserved indices of RV diastolic function, limited RV collagen accumulation, attenuated up‐regulation of calcineurin and TRPC‐6, and limited myofibroblast transformation in the RV.

Reduced Endoglin Expression Preserves RV Function and Improves Survival in RVPO

Given the limited degree of RVF observed in the Sugen+Hypoxia model, we next explored the functional role of endoglin in a surgical model of severe RVPO induced by surgical constriction of the pulmonary artery (PAC) for 7 days in WT and Eng^+/− mice. Compared to WT, baseline RV endoglin expression was lower in Eng^+/− mice (Figure 4A and 4B). Compared to sham controls, PAC increased endoglin levels in the RV in WT mice, suggesting a direct effect of RVPO on endoglin expression. RVPO also increased endoglin expression in Eng^+/− mice, but levels were significantly lower, compared to WT mice (Figure 4A and 4B). We next examined the functional impact of reduced endoglin levels in RVPO. Despite equally increased RVSP in both WT and Eng^+/− mice after PAC, RV stroke volume was decreased in WT, but not Eng^+/−, mice (Figure 4C through 4D). WT mice also manifested reduced total body weight after RVPO, whereas Eng^+/− mice did not (Figure 4E). Eng^+/− mice demonstrated substantially improved survival (100% vs. 58%, respectively; P=0.01), compared to WT mice after PAC (Figure 4F). These findings suggest that, despite identical degrees of RVPO, reduced endoglin expression in Eng^+/− mice preserved RV function and improved survival.

Reduced Endoglin Expression Limits RV Fibrosis and Signaling by TGF‐β1 and Calcineurin in RVPO

To study the mechanism underlying improved survival in Eng^+/− mice, we first examined changes in RV structure. RVPO increased RV mass in WT, but not Eng^+/−, mice (Table 3). RVPO increased RV fibrosis in WT, but not Eng^+/−, mice (Figure 5A and 5B). RV cardiomyocyte cross‐sectional area was also increased in both WT and Eng^+/− mice after RVPO, but the degree of hypertrophy was lower in Eng^+/− mice (Figure 5C and 5D). These findings suggest that endoglin regulates changes in RV structure in response to RVPO.

Next, we studied TGF‐β1 signaling in RVPO. Despite equally increased active TGF‐β1 protein levels in WT and Eng^+/− mice (Figure 6A), levels of type I collagen, pSmad3, and pERK1/2 were increased in WT mice, but not Eng^+/− mice (Figure 6B through 6D). We observed reduced levels of calcineurin protein expression in the RV from Eng^+/− mice, compared to WT, after RVPO (Figure 6E). Levels of downstream targets of calcineurin activity, including MYH7 and TRPC‐6, were also reduced in Eng^+/− mice, compared to WT, after RVPO (Figure 6F and 6G). Levels of α‐SMA mRNA were also increased in WT, but not Eng^+/−, mice after RVPO, indicating reduced fibroblast‐to‐myofibroblast conversion in Eng^+/− mice (Figure 6H). To further explore whether endoglin regulates calcineurin activity, RVPO was induced in Eng^+/+‐NFAT‐Luc and Eng^+/−‐NFAT‐Luc mice. RVPO increased luciferase activity in total RV lysates from Eng^+/+‐NFAT‐Luc, not Eng^+/−‐NFAT‐Luc, mice (Figure 7). These observations suggest that, in addition to regulating canonical and noncanonical TGF‐β1 pathways that promote cardiac fibrosis, reduced endoglin levels in the RV limit calcineurin expression and activity, including myofibroblast transformation. These findings support an important role for endoglin‐mediated regulation of TGF‐β1 and calcineurin activity in RV remodeling.

Neutralizing Endoglin Activity Prevents RV Fibrosis and Improves Survival in RVPO

To further explore the role of endoglin in mediating calcineurin expression in RVPO, WT mice were pretreated with a N‐Eng Ab or control IgG before induction of severe RVPO by PAC. Treatment with the N‐Eng Ab improved survival after 7 days of severe RVPO, compared to treatment with the IgG control (Figure 8A). Despite equally increased RVSPs in both groups (Table 4), RV fibrosis was increased in WT mice treated with the IgG control, but not in mice treated with N‐Eng Ab after severe RVPO (Figure 8B and 8C). RV cardiomyocyte cross‐sectional area was also increased in both groups after PAC, but less cardiomyocyte hypertrophy was observed in WT mice receiving the N‐Eng Ab (Figure 8D). RV mass was also increased in both groups, but the degree of hypertrophy was attenuated in N‐Eng Ab‐treated mice after RVPO (Table 4). Consistent with observations in Eng^+/− mice, WT mice treated with the N‐Eng Ab showed reduced protein levels of type I collagen, pSmad3, pERK1/2, and calcineurin levels in the RV, compared to the IgG, group after RVPO (Figure 8E through 8H). Downstream targets of calcineurin, including mRNA levels of MYH7, TRPC‐6, and α‐SMA, were also reduced in the N‐Eng Ab group, compared to the IgG group, after RVPO (Figure 8I through 8K).

Neutralizing Endoglin Activity Reverses RV Fibrosis in Established RVPO

To study the potential clinical utility of blocking endoglin activity as an approach to reduce RV fibrosis after established RVPO, WT mice subjected to moderate RVPO for 3 weeks were randomized to receive either the N‐Eng Ab or IgG control for an additional 3 weeks (Figure 9A). After 3 weeks of moderate RVPO, total body weight was reduced, whereas RV mass and systolic pressure were increased and RV stroke volume decreased, compared to sham controls (Table 5). RV fibrosis, type I collagen, and calcineurin expression were also increased, compared to sham controls (Figure 9B through 9F). After an additional 3 weeks (6 weeks total) of moderate RVPO, both IgG‐ and N‐Eng Ab‐treated groups had persistently increased RV mass and RVSP with reduced cardiac output (Table 5). No mortality was observed after moderate RVPO in either group at any time point. RV fibrosis progressively worsened in mice treated within the IgG group, but was significantly reduced in the N‐Eng Ab‐treated group (Figure 9B and 9C). Type I collagen and calcineurin protein expression also increased progressively in the IgG group, but were reduced in the N‐Eng Ab group (Figure 9D through 9F). These findings support that blocking endoglin activity reverses established RV fibrosis in chronic RVPO in mice.