Aortic valve sclerosis in mice deficient in endothelial nitric oxide synthase

Abstract

Risk factors for fibrocalcific aortic valve disease (FCAVD) are associated with systemic decreases in bioavailability of endothelium-derived nitric oxide (EDNO). In patients with bicuspid aortic valve (BAV), vascular expression of endothelial nitric oxide synthase (eNOS) is decreased, and eNOS^−/− mice have increased prevalence of BAV. The goal of this study was to test the hypotheses that EDNO attenuates profibrotic actions of valve interstitial cells (VICs) in vitro and that EDNO deficiency accelerates development of FCAVD in vivo. As a result of the study, coculture of VICs with aortic valve endothelial cells (vlvECs) significantly decreased VIC activation, a critical early phase of FCAVD. Inhibition of VIC activation by vlvECs was attenuated by N^G-nitro-l-arginine methyl ester or indomethacin. Coculture with vlvECs attenuated VIC expression of matrix metalloproteinase-9, which depended on stiffness of the culture matrix. Coculture with vlvECs preferentially inhibited collagen-3, compared with collagen-1, gene expression. BAV occurred in 30% of eNOS^−/− mice. At age 6 mo, collagen was increased in both bicuspid and trileaflet eNOS^−/− aortic valves, compared with wild-type valves. At 18 mo, total collagen was similar in eNOS^−/− and wild-type mice, but collagen-3 was preferentially increased in eNOS^−/− mice. Calcification and apoptosis were significantly increased in BAV of eNOS^−/− mice at ages 6 and 18 mo. Remarkably, these histological changes were not accompanied by physiologically significant valve stenosis or regurgitation. In conclusion, coculture with vlvECs inhibits specific profibrotic VIC processes. In vivo, eNOS deficiency produces fibrosis in both trileaflet and BAVs but produces calcification only in BAVs.

endothelial cells play a central role in cardiovascular homeostasis through release of a wide variety of agents including nitric oxide (NO) (16). In addition to its vasodilator and antiatherogenic effects (16), NO inhibits myocardial fibrosis (51). In mice deficient in endothelial NO synthase (eNOS), prevalence of bicuspid aortic valves (BAVs) is increased (22, 34). In cultured porcine aortic valve interstitial cells (VICs), NO inhibits formation of calcific nodules (30). Effects of postnatal endothelium-derived NO (EDNO) deficiency on development of aortic valve disease, however, have not yet been fully established.

The pathogenesis of fibrocalcific aortic valve disease (FCAVD) is not completely understood, and no medical therapy is known to be clinically effective in reversing or slowing its progression (8, 44). Scarce data are available on the effect of risk factors for FCAVD (50) upon endothelium of aortic valves. In vascular tissue, those factors are associated with increased oxidative stress, leading to a decrease in bioavailability of EDNO, and endothelial dysfunction (7, 10). During aging, which is the major risk factor for aortic valve sclerosis (11, 50), there is a systemic increase in levels of superoxide (9), impairment in antioxidant mechanisms (37), and a decrease in expression of eNOS (2). BAV, which is present in 1 to 2% of the population, is the most prevalent congenital abnormality of the heart (20). In patients with BAV, expression of eNOS in aortic endothelial cells is significantly decreased (1).

In this study, we first tested the hypotheses that endothelium attenuates profibrotic actions of VICs in vitro and that stiffness of matrix modulates the interactions between endothelium and VICs. We next tested the hypothesis that eNOS deficiency promotes aortic valve fibrosis and calcification in mice. If stiffness of matrix modulates the interaction of endothelium and VICs, and eNOS deficiency promotes fibrosis and calcification, the findings could provide new insight into mechanisms that lead to FCAVD.

MATERIALS AND METHODS

Studies In Vitro

Primary porcine aortic VICs and aortic valve endothelial cells (vlvECs) were isolated from aortic valve leaflets, excised from pigs of either sex within 24 h of slaughter, using sequential collagenase digestion and cryopreservation as previously described (3, 28). All materials were purchased from Life Technologies (Grand Island, NY) unless otherwise specified.

To study the effect of culture matrix stiffness upon VIC activity, two matrix preparations were used. A stiffer matrix was comprised of standard tissue culture polystyrene (TCPS, E = ∼1 GPa). In separate experiments, a less stiff matrix was comprised of photo-activated hydrogel (E = 28 kPa) and prepared as follows. An eight-armed polyethylene glycol-norbornene was dissolved in phosphate-buffered saline with dithiol linker peptide Cys-Gly-Arg-Gly-Asp-Ser and the photoinitiator I-2959 (Ciba-Geigy Chemical, Tom River, NJ) (19) and polymerized with 365-nm ultraviolet light at 8.5 mW/cm² for 10 min to create hydrogels with a Young's modulus of 28 kPa (31). VICs and vIvECs were used for experiments at the third passage.

VlvECs were grown to 90% confluence, trypsinized, pelleted, and resuspended in serum and phenol red free RPMI media with red Cell-Tracker dye. Cells were incubated at 37°C for 15 min to allow discrimination between vlvECs and VICs during image analysis. In parallel, VICs were grown to 80% confluence, trypsinized, pelleted, and resuspended in 1% FBS low-glucose DMEM media. VlvECs were then diluted in 1% FBS low-glucose DMEM media and seeded on TCPS or hydrogel at 50,000 cells/cm². VICs were then seeded on top of vlvECs at 20,000 cells/cm². The concentration of VICs was selected to minimize VIC-VIC contact. In the same wells, the eNOS inhibitor, N^G-nitro-l-arginine methyl ester (l-NAME, 100 μM) or indomethacin (5 μM), a prostaglandin inhibitor, or both, were added to the coculture medium, to characterize paracrine signaling between vlvECs and VICs.

For each series of histology studies, protein level assays, or gene expression quantitation in vitro, each experimental condition, e.g., TCPS versus hydrogel, was performed using cells from the same cell source. Each data set included results from three independent cell isolation procedures. Quantitative procedures were performed in triplicate, yielding 9 distinct determinations for each experimental parameter, under each experimental condition. After 3 days of culture, the cells were fixed in 10 vol/vol% formalin (Sigma) and stained with mouse anti-α-smooth muscle actin (α-SMA) primary antibody (Abcam), goat anti-mouse FITC-488 secondary antibody, and 4′,6-diamidino-2-phenylindole (DAPI). The stained cells were imaged using a Zeiss dual-photon microscope. Three images for VIC αSMA (FITC), vlvEC cell tracker (tetramethylrhodamine isothiocyanate), and nuclei (DAPI) were taken for each of the 3 wells per condition. VICs were considered activated if their cytoskeleton contained polymerized fibers of α-SMA and not simply diffuse staining (17, 23). The number of activated cells was normalized to the total cell count, (i.e., number of nuclei), determined using the DAPI image, yielding a percentage of activated cells in the total population.

Phospho-Smad2/total Smad2 Western blot.

VICs were lysed for 20 min at 4°C before removing cell debris via centrifugation. Total protein concentrations were determined using a micro-bicinchoninic acid assay (Thermo Scientific). Samples were then supplemented with an SDS-β-mercaptoethanol solution and boiled for 5 min to denature proteins. An aliquot of 10 μg of total protein was loaded into a 10 wt% gel per lane and resolved with a Bio-Rad system. The separated proteins in the gel were then transferred to a polyvinylidene difluoride membrane. The membrane was washed in 0.05 wt% Tween-20 TBS (TBST) and blocked in a 5 wt% BSA TBST solution for phospho (p)-SMAD2/total (t)-SMAD2 for 1 h at room temperature. The membrane was then incubated overnight at 4°C with GAPDH (Cell Signaling Technologies), p-Smad2 (Cell Signaling Technologies), or t-Smad2 (Cell Signaling Technologies) primary antibodies at previously determined dilutions in the same blocking solution. Afterward, the membrane was washed in TBST and incubated at room temperature for 1 h with an horseradish peroxidase secondary antibody. The membrane was washed again before addition of the horseradish peroxidase luminescent substrate, film exposure, and development. Afterward, the developed film was scanned and analyzed using ImageJ (National Institutes of Health, Bethesda, MD) software to determine the ratio of p-Smad2 to t-Smad2 for each sample.

Gene regulation RT-PCR.

VICs were seeded at 20,000 cells/cm² with or without vlvECs at 50,000 cells/cm² in 1 vol/vol% FBS low glucose DMEM media to prevent proliferation and cultured for 3 days in an incubator at 37°C and 5% CO₂. Messenger RNA was isolated cells using a Tri Reagent/1-bromo-3-chloropropane extraction followed by precipitation by addition of isopropanol. RNA was purified by 75% ethanol washes. Purity and amount of mRNA were confirmed with a NanoDrop spectrophotometer (Fisher). Reverse transcription was performed using the iScript cDNA Synthesis kit (Bio-Rad) in an Eppendorf Mastercycler personal. Quantitative real-time polymerase chain reaction (qRT-PCR) was then conducted using an iCycler qRT-PCR machine (Bio-Rad), with iQ SYBR Green Supermix (Bio-Rad) and primers specific for L30, α-SMA, collagen-1, collagen-3, and matrix metalloproteinase-9 (MMP-9) (Invitrogen). Threshold cycle and primer efficiency were analyzed according to the standard curve method and normalized to L30.

Studies in Mice

All studies were approved by the Institutional Animal Care and Use Committee at the University of Iowa (Public Health Service Animal Welfare Assurance No. A3021-01). Endothelial NOS^−/− mice, bred on a C57BL/6 background, were purchased from Jackson Laboratory (Bar Harbor, ME). Age-matched wild-type (WT) C57BL/6 mice were used as controls. Genotypes were confirmed by PCR (Qiagen). All mice were maintained on a normal chow diet (Harlan Teklad 7004 rodent diet).

Arterial pressure.

Systolic blood pressure was measured in conscious mice using a computerized noninvasive tail-cuff system (Visitech Systems BP-2000 Blood Pressure Analysis System, Apex, NC). Measurements were repeatedly performed over 5 consecutive days to allow acclimatization to the procedure.

Echocardiography.

Echocardiography was performed on all mice at 6 mo of age and on all surviving mice at 12, 15, and 18 mo of age. Mice were lightly sedated using midazolam (0.15 mg sc). Parasternal long- and short-axis views were obtained using high-frequency echocardiography (Vevo 2100, VisualSonics, Toronto, Canada) to assess left ventricular (LV) volumes and mass, aortic valve cusp number, and aortic root dimensions. M-mode images were then acquired to measure aortic cusp separation distance (ACS) during systole as previously described (52).

Cardiac magnetic resonance imaging.

Cardiac magnetic resonance imaging (MRI) was performed at 6 mo of age. Mice were deeply sedated with midazolam (8 mg/kg sc) and morphine (4 mg/kg sc), which produce only modest depression of heart rate and LV contractility (5). MRI was performed using an Innova 4.7-T instrument (Varian, Palo Alto, CA), as previously described (4). The presence and qualitative severity of aortic regurgitation were assessed by analysis of dephasing of white blood in the LV outflow tract during early diastole. Regurgitant fraction was calculated as the difference between left and right ventricular stroke volumes, divided by LV stroke volume (4).

Invasive hemodynamic measurement.

Hemodynamic measurements were made at 6 or 18 mo of age. A 1.4-Fr microtransducer-tipped catheter (Millar, Houston, TX) was inserted into the right common carotid artery and advanced into the LV. Pressure was continuously recorded as the catheter tip was pulled back into the ascending aorta. Aortic valve gradient was calculated as the difference between peak LV pressure and peak aortic pressure. Maximum first derivative of LV pressure (dP/dt_max) was determined by electronic differentiation of the LV pressure signal. Aortic pulse pressure was calculated as the difference between aortic systolic pressure and aortic diastolic pressure.

Histological studies.

Using a cryostat, we obtained 10-μm-thick serial sections of aortic valves. Sections from proximal, mid, and distal valve were placed on each slide. Slides were stained with Alizarin Red, Masson's Trichrome, or Oil Red-O to quantitate the amount of calcium, collagen or lipid, respectively, as previously described (39). Immunofluorescence was used to quantitate osterix, collagen-1, collagen-3, and activated caspase-3. For negative controls, primary antibodies were replaced with immune-naïve IgG, after which the secondary (fluorescent-tagged) antibody was applied. Areas staining positive for each histological marker were planimetered electronically off-line (Adobe Systems San Jose, CA) and expressed as a percentage of the total area within the valve annulus.

Statistical analysis.

In vitro data are presented as means ± SE for three biological and three sample replicates. Data were compared using Tukey's test. All continuous variables in mice are reported as means ± SE. Findings in bicuspid valves in vivo were compared with findings in trileaflet valves, based on histological assessment of cusp number for each mouse using unpaired t-tests. Significance was established for P < 0.05. Because slides for histology in 6- and 18-mo-old mice were prepared, stained, imaged, and quantified separately, direct statistical comparison of the data from the two time points was not performed.

RESULTS

vlvECs Inhibit Activation of VICs In Vitro

When VICs were cultured alone, >85% of cells were activated to the α-SMA-expressing phenotype (Fig. 1). When vlvECs and VICs were cocultured, VIC activation was significantly reduced, compared with VICs cultured alone. The inhibitory effect of vlvECs upon activation of VICs tended to be reduced by addition of either l-NAME or indomethacin. l-NAME or indomethacin had no effect on VIC activation in the absence of vlvECs. Expression of MMP-9, an effector of matrix remodeling, was strongly inhibited by coculture with vlvECs on gel matrix, but not on TCPS (Fig. 1).

Fig. 1.

Inhibition of valve interstitial cell (VIC) activation and expression of matrix metalloproteinase-9 (MMP-9) by valve endothelial cells (vlvECs) in vitro. A: culture well micrographs demonstrating immunostaining for α-smooth muscle actin (α-SMA, green) and Cell Tracker-tagged endothelial cells (red). B and C: percentage of VICs with positive immunostaining for α-SMA. D and E: normalized expression of MMP-9. TCPS, tissue culture polystyrene, Young's modulus = ∼1 GPa; Gel, photoactivatable hydrogel, Young's modulus = 28 kPa; l-NAME, N^G-nitro-l-arginine methyl ester; Indo, indomethacin. *P < 0.05 for VICs vs. VICs + vlvECs.

Collagen-1 expression by VICs was modestly inhibited by coculture with vlvECs on TCPS, but not on gel matrix (Fig. 2). In contrast, collagen-3 expression by VICs was very strongly inhibited by coculture with vlvECs, on both TCPS and gel matrixes, indicating a selective effect of endothelium upon expression of genes encoding specific collagen species.

Fig. 2.

Profibrotic signaling in vitro. A–D: quantitative PCR for expression of collagen-1 or collagen-3. E and F: Western blotting for phospho (p)-Smad2 and total (t)-Smad2. TCPS Young's modulus = ∼1 GPa gel photo-activatable hydrogel, Young's modulus = 28 kPa. *P < 0.05 vs. VICs alone.

Levels of p-Smad2, an activated effector of TGF-β signaling, were reduced when VICs were cocultured with vlvECs, on both TCPS and gel matrices (Fig. 2). Inhibition of p-SMAD2 levels was not significantly reversed by addition of l-NAME and indomethacin.

Studies in Mice

Body mass was lower in eNOS^−/− mice than in WT mice at 18 mo of age (Table 1). Systolic blood pressure was 133 ± 2 mmHg in eNOS^−/− mice and 120 ± 1 in WT mice, at age 12 mo (P < 0.05). At age 18 mo, systolic blood pressure was 119 ± 1 mmHg in eNOS^−/− mice and 119 ± 1 in WT mice [P = not significant (NS)].

Table 1.

Characteristics of eNOS^−/− and WT mice

	6 mo old		18 mo old
	WT	eNOS−/−	WT	eNOS−/−
Total number	9	17	5	10
Male:female	4:5	9:8	3:2	3:7
Heart rate, beats/min	640 ± 25	612 ± 16	514 ± 78	573 ± 22
Body mass, g	26.5 ± 1.6	25.8 ± 1.0	41.9 ± 5.6	29.3 ± 1.6*
LVEF, %	70 ± 3	77 ± 1*	72 ± 5	76 ± 4
LVSV, μl	25 ± 3	23 ± 2	27 ± 4	33 ± 4
LV mass, mg	94 ± 6	90 ± 3	139 ± 16	140 ± 9
LV mass/body mass, mg/g	3.6 ± 2	3.5 ± 0.1	3.6 ± 0.6	4.9 ± 0.4
Aortic pulse pressure, mmHg	32 ± 2	38 ± 2	35 ± 2	35 ± 3
LV dP/dtmax, mmHg/s	5,246 ± 1,257	4,473 ± 430	5,119 ± 1,051	6,148 ± 395

Values are means ± SE.

eNOS^−/−, endothelial nitirc oxide synthase knockout mice; LVEF, left ventricular (LV) ejection fraction; LVSV, LV stroke volume; LV dP/dt_max, maximum first derivative of LV pressure.

^*P < 0.05 vs. age-matched wild-type (WT).

During conscious sedation, echocardiography revealed no physiologically important differences in LV volumes, mass, or systolic function in eNOS^−/− mice. During general anesthesia for invasive hemodynamic study, LV dP/dt_max and aortic pulse pressure were normal in eNOS^−/− mice (Table 1).

Rate of survival to 18 mo of age was 81% for WT mice and 65% for eNOS^−/− mice (P = NS). Analysis of interim echocardiograms revealed no mice with severe aortic stenosis (ACS < 0.66 mm) before death. LV ejection fraction obtained before demise was not different from mice that survived to 18 mo of age (72 ± 2 vs. 76 ± 4%, P = NS).

Prevalence of BAV in eNOS^−/− Mice

BAV, confirmed by postmortem studies, was present in 8 of 27 eNOS^−/− mice (30%), whereas all 14 aortic valves from WT mice that were examined histologically were trileaflet. Overall accuracy of two-dimensional echocardiography in determining aortic valve cusp number in eNOS^−/− mice was only 75%, when postmortem histology was used as the reference standard. M-mode echocardiography produced images sufficient to measure ACS in all mice at both ages.

eNOS Deficiency Accelerates Fibrosis in the Aortic Valve

When compared with that in age-matched WT mice, a two- to threefold increase in total collagen content of the aortic valve was found in 6-mo-old eNOS^−/− mice with either bicuspid or trileaflet valves (Fig. 3). At age 18 mo, there was a trend toward greater fibrosis in BAV compared with WT, but the difference did not reach statistical significance. Immunostaining revealed normal levels of collagen-1 in the aortic valves of eNOS^−/− mice with either trileaflet valves or BAV at both ages (Fig. 4). Levels of collagen-3, however, were increased in eNOS^−/− mice, regardless of cusp number, at 18 mo of age.

Fig. 3.

Fibrosis in the aortic valve. Masson's trichrome stain demonstrates total collagen content (blue) in bicuspid (B) and trileaflet (T) valves from endothelial nitric oxide synthase knockout (eNOS^−/−) and wild-type (WT) mice at 6 and 18 mo of age. *P < 0.05 vs. WT; N = 9 and 5 for 6 and 18 mo, respectively; T eNOS^−/−, N = 13, 6; B eNOS^−/−, N = 4, 4. Scale bar = 300 μm.

Fig. 4.

Collagen-1 and collagen-3 in the aortic valve. 3,3′-Diaminobenzidine (DAB) immunostains backstained with hematoxylin. Negative control denotes DAB staining without primary antibody. WT, N = 4 and 4 for 6 and 18 mo, respectively; T eNOS^−/−, N = 4, 4; B eNOS^−/−, N = 4, 4. *P < 0.05 vs. WT. Scale bar = 300 μm.

Bicuspid eNOS^−/− Valves Are Prone to Calcification

At 6 and 18 mo of age, calcification was much greater in BAV from eNOS^−/− mice compared with WT and trileaflet eNOS^−/− valves (Fig. 5).

Fig. 5.

Calcification in the aortic valve. Small areas of calcification (Alizarin Red, arrows) in a bicuspid aortic eNOS^−/− valve at age 6 mo and more calcification in all groups at age 18 mo. Planimetry of Alizarin Red-positive staining indicates significant increase in calcification in bicuspid valves from 6- and 18-mo-old eNOS^−/− mice, compared with trileaflet valves from WT or eNOS^−/− mice. WT, N = 9 and 5 for 6 and 18 mo, respectively; T eNOS^−/−, N = 13, 6; B eNOS^−/−, N = 4, 4. *P < 0.05 vs. WT; †P < 0.05 vs. T eNOS^−/−. Scale bar = 300 μm.

At 6 mo of age, the increase in aortic valve calcification in eNOS^−/− mice with BAV was associated with a fourfold increase in expression of osterix, a transcription factor that is unique to mineralization-competent osteoblast-like cells (32, 49). By age 18 mo, osterix levels were comparable in all three groups (Fig. 6).

Fig. 6.

Osterix in the aortic valve. Osterix was increased in bicuspid valves from 6-mo-old, but not 18-mo-old, eNOS^−/− mice (N = 4–6). AF autofluorescence; Osx osterix staining. *P < 0.05 vs. WT. Scale bar = 300 μm.

At age 6 mo, activated caspase-3, a marker for apoptosis, was increased eightfold in BAVs and twofold in trileaflet eNOS^−/− valves, compared with WT valves (Fig. 7). Activated caspase-3 levels remained elevated in bicuspid eNOS^−/− valves at 18 mo of age.

Fig. 7.

Proapoptotic signaling in the aortic valve. A: immunostaining for activated caspase-3 in valves from mice at 6 mo of age. Positive staining (arrows) was observed near sites of cusp attachment. B: group data; N = 4–6. *P < 0.05 vs. WT; †P < 0.05 vs. T eNOS^−/−. Scale bar = 300 μm.

Lipid Deposition in the Aortic Valve

At age 6 mo, the amount of lipid in was higher in eNOS^−/− valves compared with WT valves, but nevertheless was small. With aging, the amount of lipid deposited in the aortic valve was comparable among the groups (Fig. 8).

Fig. 8.

Lipid deposition in the aortic valve. A: there was minimal staining with Oil Red-O in all groups at age 6 mo and increased lipids in all groups at age 18 mo. B: planimetry of Oil Red-O positive staining indicates that even though lipid deposition in the aortic valve of these normocholesterolemic mice was small, it was significantly higher in trileaflet and bicuspid eNOS^−/− valves compared with WT valves at age 6 mo. There was no significant difference at age 18 mo. *P < 0.05 vs. WT. WT, N = 9 and 5 for 6 and 18 mo, respectively; B eNOS^−/−, N = 4, 4; T eNOS^−/−, N = 13, 6. Note differences in scale in 6- and 18-mo-old mice. Scale bar = 300 μm.

eNOS Deficiency Is Not Sufficient to Produce Physiologically Important Valve Dysfunction

At 6 and 18 mo of age, eNOS^−/− mice with trileaflet aortic valves demonstrated no impairment of ACS (Fig. 9). At 6 mo of age, but not at 18 mo, eNOS^−/− mice with BAV had a small but statistically significant decrease in ACS compared with WT mice.

Fig. 9.

Aortic valve function. A and B: aortic cusp separation (ACS). C: transaortic valve gradient in an eNOS^−/− mouse, determined by invasive hemodynamic assessment. D: group data for transvalvular gradient; N = 2 and 6 for WT at 6 and 18 mo, respectively; N = 4 and 6 for eNOS^−/−. E: absence of significant aortic regurgitation, as assessed by MRI (N = 15). *P < 0.05 vs. WT.

Systolic pressure gradients across the aortic valve were minimal in eNOS^−/− mice and WT mice at both ages and did not differ between groups (Fig. 9).

Aortic valve regurgitation, which was assessed by MRI, was mild and not significantly different between WT and eNOS^−/− mice (Fig. 9). BAV in eNOS- deficient mice was not associated with aortic root dilation (Fig. 10).

Fig. 10.

Proximal aorta. Aortic root diameters, at the sinuses of Valsalva, and at the sinotubular junction, were similar in 18-mo-old WT mice and eNOS^−/− mice with either trileaflet or bicuspid aortic valve. WT, N = 9 and 5 at 6 and 18 mo, respectively; T eNOS^−/−, N = 13, 6; B eNOS^−/−, N = 4, 4.

DISCUSSION

There are several new findings in this study. First, aortic valve endothelium protects against valve fibrosis. Accelerated fibrosis occurs in both bicuspid and trileaflet valves in eNOS^−/− mice, which implicates eNOS deficiency (not cusp number) as the mechanism for fibrosis. Second, aortic valve endothelium inhibits profibrotic actions of VICs. Third, eNOS^−/− BAVs are prone to calcification, even in the absence of other FCAVD risk factors, such as hypercholesterolemia. Thus a novel finding is that fibrotic and calcific processes in eNOS-deficient aortic valves exhibit distinctively different temporal patterns and rates of progression: fibrosis starts at a young age, whereas calcification predominates in bicuspid valves at an older age.

Valve Endothelium Modulates VIC Profibrotic Activity In Vitro

VICs play a central role in pathogenesis of FCAVD, as they can be activated, to a profibrotic phenotype. We found, using coculture, that valve endothelium exerts a paracrine suppressive effect upon activation of VICs.

Our studies in vitro, using indomethacin to inhibit endothelial cyclooxygenase activity, implicate at least one additional endothelial paracrine inhibitor of VIC activation. Cross talk between EDNO and cyclooxygenase is complex. NO modulates prostaglandin production in vascular endothelium by activation of cyclooxygenase-1 and inhibition of cyclooxygenase-2 (13). Therefore, in addition to a direct paracrine inhibition of VIC activation, EDNO probably plays a modulatory role in other proinflammatory pathways involving VICs.

We report several lines of evidence that indicate that endothelial inhibition of profibrotic processes in VICs is not limited to suppression of α-SMA levels. VIC expression of MMP-9, a protease that participates in remodeling of extracellular matrix, is inhibited by coculture with vlvECs grown on an elastic matrix (gel), but not on stiff matrix (TCPS). Inhibition of VIC collagen deposition by vlvECs is preferential for collagen-3, compared with collagen-1. That finding in vitro was recapitulated in mice, where eNOS deficiency resulted in preferential increase in collagen-3 in the aortic valve. The findings strongly suggest specific regulation of collagen synthesis by NO, in addition to a global protective effect of vlvECs. The findings have implications for future studies of FCAVD, because collagen-1 putatively modulates matrix stiffness, whereas collagen-3 is more associated with matrix elasticity (33).

Levels of p-Smad2, a downstream transcriptional mediator of TGF-β signaling, are attenuated by coculture with vlvECs. Thus the mechanisms responsible for selective inhibition of expression of collagen-3, but not collagen-1, by vlvECs likely occur downstream from p-Smad2. Thus, while we are not able to identify and characterize all of the inter- and intracellular signaling processes leading to fibrosis, our new finding links VIC-vlvEC cross talk to expression of p-SMad2, an established mediator of FCAVD, in vitro (39). Taken together, the findings support the conclusion that vlvECs regulate multiple profibrotic processes in VICs.

Previous studies have demonstrated that profibrotic processes in cultured VICs are profoundly influenced by the mechanical properties of the culture medium. The present study adds an important new concept in the relationship between VICs and matrix: VIC-vlvEC cross talk is also strongly influenced by the culture matrix. Coculture on elastic gel matrix inhibited expression of MMP-9, but MMP-9 inhibition was not seen in cocultures grown on stiffer TCPS medium. Conversely, VIC expression of collagen-1 was inhibited by coculture with vlvECs grown on stiff, but not elastic, matrix. By demonstrating the effect of matrix stiffness over the disease-relevant range up to 1 GPa, the new findings extend the conclusions of a prior study that used gels with a much lower range of matrix stiffness (3–27 kPa) (25). We speculate that mechanical properties of valve extracellular matrix change during aging and disease processes and that those changes regulate cellular processes leading to FCAVD.

Deficiency of eNOS Accelerates Valve Fibrosis In Vivo

EDNO deficiency is sufficient to cause increased collagen deposition in aortic valve, which is a novel finding. This process starts at a young age in eNOS^−/− mice and occurs in the absence of other FCAVD risk factors.

Aging, which is the main risk factor for aortic stenosis in humans, is associated with variable degrees of degenerative changes in the valve (50). Macroscopically, there is an increase in leaflet calcification (36) and thickness of the cusps (47). Microscopically, degenerative lesions exhibit infiltrates of chronic inflammatory cells, mineralization, and deposition of lipids, collagen, and elastin (42). Aging is also associated with vascular endothelial dysfunction, increased superoxide (9) with impairment of antioxidant mechanisms (37), decreased expression of eNOS in aortic endothelial cells (2), accompanied by decreased bioavailability of EDNO, which is inactivated by superoxide (10). Age-dependent depletion of EDNO and degenerative changes may explain why a similar degree of aortic valve fibrosis was observed in old WT and eNOS^−/− mice. By accelerating fibrosis and other age-related processes in the valve, including lipid accumulation, eNOS deficiency may have mechanistic links to aging in the valve.

Valve Calcification in eNOS^−/− Mice

We found that trileaflet eNOS^−/− valves did not develop abnormal calcification. The finding suggests that an additional factor is necessary to invoke valve calcification, in this case, a congenital anomaly of valve structure. Cholesterol-fed eNOS^−/− mice with BAVs, but not trileaflet aortic valves, develop aortic stenosis (43). With the use of micro-commuted tomography, mild valve calcification was anecdotally observed in BAV eNOS^−/− mice. Neither stenosis nor calcification was detected in aortic valves of normocholesterolemic eNOS^−/− mice, regardless of cusp number, in that study. Our findings demonstrate, however, that these congenitally abnormal eNOS-deficient valves are significantly more prone to calcification at a young age, even in the absence of other identifiable risk factors for FCAVD. It is likely that histological examination is more sensitive for detection of tissue calcification than micro-computed tomography, facilitating detection of increased valve calcification in normocholesterolemic eNOS^−/− mice with BAV.

Mechanisms of Valve Calcification in eNOS^−/− Mice

These studies address mechanisms of calcification of the aortic valve. We identify activation of the osteogenic pathway as a contributing mechanism for development of valvular calcification in young mice with BAV. Previous studies in vitro demonstrated that EDNO suppresses osteoblastic differentiation of VICs (46). In eNOS^−/− mice, however, we found that trileaflet aortic valves did not exhibit increased calcification. This is in agreement with recent findings in cholesterol-fed eNOS^−/− mice (43). We postulate that the procalcific effect of EDNO deficiency is offset by a protective mechanism that is downregulated in the presence of altered flow patterns (27) and unevenly distributed stress (15) associated with BAV. Potential targets for future studies that aim at understanding the effect of blood flow patterns on development of valve dysfunction include plasminogen activators and inhibitors, endothelin-1, TGF-β, collagen, and ephrin-A1, as their expression in endothelial cells is altered by turbulent shear stress (41).

With aging, eNOS^−/− mice with BAV exhibited progressive increase in calcification of the aortic valve without evidence of a sustained increase in osteoblast differentiation. We have identified a second procalcific mechanism by which eNOS deficiency may promote valve calcification: increased apoptosis. Apoptosis contributes to calcification of valve tissue in vitro (14) and in vivo (45, 48).

Lipid Deposition in eNOS^−/− Valves

We found that lipid content in the aortic valve was higher in eNOS^−/− than in WT mice at 6 mo of age. Although the amount of lipid deposition was small, we believe this is a novel finding in normocholesterolemic mice. Lipid deposition is an early step in the initiation of FCAVD, a finding that is not known to be confined to hypercholesterolemic individuals (24). Disruption of endothelium produces lipid deposition in blood vessels, putatively by exposing circulating lipoprotein particles to electrostatic attraction by subendothelial proteoglycans (26). Lipid oxidation produces increased inflammation (6) and osteogenic transformation (18) in blood vessels. We speculate that lipid deposition may contribute to the FCAVD phenotype.

Valve Function in eNOS^−/− Mice

We present three lines of evidence which indicate that eNOS deficiency alone is not sufficient to cause hemodynamically significant aortic valve stenosis. ACS, assessed using M-mode echocardiography, was normal in eNOS^−/− mice with trileaflet aortic valves. Although eNOS^−/− mice with BAV had a statistically significant decrease in ACS compared with WT mice, the reduction was not physiologically significant, based on invasive hemodynamic determinations performed for this study and for a previously published study (52). Indeed, we found that transvalvular gradients, assessed by catheterization, were normal in eNOS^−/− mice. Absence of significant aortic regurgitation was confirmed using MRI.

Li et al. (35) reported normal LV systolic function in eNOS^−/− mice at 5.5 mo of age but found significant systolic dysfunction at 21 mo age in male eNOS^−/− mice only. We found that LV systolic function was normal in eNOS^−/− mice at both 6 and 18 mo of age. Critically, in the context of effects of LV systolic function upon assessments of aortic valve function, we found that dP/dt_max and aortic pulse pressure were normal during invasive hemodynamic study. Furthermore, the effect of depressed LV systolic function, if any, upon ACS would be to decrease it. We found, however, that ACS was normal in eNOS^−/− mice with trileaflet aortic valves. The findings do not support the possibility that impaired aortic valve function was masked by alterations in LV function in eNOS^−/− mice. Taken together, the findings which indicate that EDNO deficiency produces accelerated fibrosis regardless of cusp number and calcification in BAV are most consistent with aortic valve sclerosis, not stenosis.

A priori, the finding of near-normal valve function in eNOS^−/− mice with BAV may appear at odds with clinical observations. However, in patients with BAV who reach the fourth decade of life without overt valve dysfunction, aortic valve surgery is required in only about 25% of the patients over the ensuing 20 years (38). Survival into the ninth decade of life, without need for valve surgery, has been observed in patients with BAV (38). Our findings in eNOS^−/− mice are thus consistent with findings in humans. It appears that as yet unidentified genetic and environmental factors determine importantly the course of disease processes in BAV.

Clinically, BAV can be associated with aortic root dilatation that is out of proportion to the degree of valve dysfunction (29). Aortic dilatation is attributed to increased wall stress (15) due to altered flow patterns (27), and to intrinsic abnormalities in the aorta that resemble histologically those seen in patients with Marfan syndrome (40). In eNOS^−/− mice, we found that BAV, which results from fusion in utero of the right and noncoronary cusps (22), was not associated with aortic root dilatation. This finding is in accordance with the observation that pediatric patients with fusion of right and noncoronary cusps are at increased risk for aortic stenosis and regurgitation, whereas fusion of right and left cusps is more strongly associated with aortopathies (12, 21).

Limitations

We studied mice up to 18 mo of age. It is possible that significant valve dysfunction could develop later in eNOS^−/− mice.

Histological samples from 6-mo and 18-mo-old mice, respectively, were obtained and processed at different times, possibly subjecting each age group to “batch effects” related to incubation times and other variables. Samples from WT and eNOS^−/− mice within each age group, however, were processed at the same time, avoiding the possibility that a batch effect could influence comparisons between mouse strains.

Analysis of osterix and activated caspase-3 immunofluorescence was confounded by the appearance of tissue autofluorescence, which was present to a variable degree in the valve annulus. Our convention was to ascribe fluorescence in the valve annulus, which can include cusp attachment sites, to autofluorescence. Thus our results may underestimate the presence of osterix or activated caspase-3 in valve annulus, and also may underestimate the magnitude of differences between groups.

Conclusion

By promoting fibrosis, EDNO deficiency produces aortic valve sclerosis during early and midlife in mice. EDNO deficiency also results in increased prevalence of congenital BAV, which is more prone to calcification throughout life. The findings thus may facilitate the search for effective therapies for patients with BAV and those with endothelial dysfunction who are at an increased risk for rapidly progressive FCAVD.

GRANTS

These studies were supported by National Institutes of Health Grants HL-062984 (to D. D. Heistad and R. M. Weiss) and HL-089260 (to K. S. Anseth) and by the Howard Hughes Medical Institute (to K. S. Anseth).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

R.N.E.A., K.S.A., D.D.H., and R.M.W. conception and design of research; R.N.E.A., S.T.G., G.P.H., Y.C., M.K.D., D.C.K., D.D.L., R.M.B., H.D., K.A.Z., W.K., and R.M.W. performed experiments; R.N.E.A., S.T.G., G.P.H., Y.C., M.K.D., D.D.L., and R.M.W. analyzed data; R.N.E.A., S.T.G., G.P.H., Y.C., M.K.D., D.C.K., D.D.L., K.S.A., D.D.H., and R.M.W. interpreted results of experiments; R.N.E.A., S.T.G., G.P.H., and R.M.W. prepared figures; R.N.E.A. drafted manuscript; R.N.E.A., S.T.G., Y.C., M.K.D., D.D.H., and R.M.W. edited and revised manuscript; R.N.E.A., S.T.G., G.P.H., Y.C., M.K.D., D.C.K., D.D.L., R.M.B., H.D., K.A.Z., W.K., K.S.A., D.D.H., and R.M.W. approved final version of manuscript.