Sclerostin ablation prevents aortic valve stenosis in mice

Abstract

The objective of this study was to test the hypothesis that targeting sclerostin would accelerate the progression of aortic valve stenosis. Sclerostin (mouse gene, Sost) is a secreted glycoprotein that acts as a potent regulator of bone remodeling. Antibody therapy targeting sclerostin is approved for osteoporosis but results from a stage III clinical trial showed multiple off-target cardiovascular effects. Wild-type (WT, Sost^+/+) and Sost-gene knockout-expression (Null, Sost^−/−) mice were generated and maintained to 12 mo of age on a high-cholesterol diet to induce aortic valve stenosis. Mice were examined by echocardiography, histology, and RNAseq. Immortalized valve interstitial cells were developed from each genotype for in vitro studies. Null mice developed a bone overgrowth phenotype, similar to patients with sclerosteosis. Surprisingly, however, WT mice developed hemodynamic signs of aortic valve stenosis, whereas Null mice were unchanged. WT mice had thicker aortic valve leaflets and higher amounts of α-smooth muscle actin, a marker myofibroblast activation and dystrophic calcification, with very little evidence of Runx2 expression, a marker of osteogenic calcification. RNAseq analysis of aortic roots indicated the HOX family of transcription factors was significantly upregulated in Null mice, and valve interstitial cells from Null animals were enriched with Hoxa1, Hoxb2, and Hoxd3 subtypes with downregulated Hoxa7. In addition, Null valve interstitial cells were shown to be less contractile than their WT counterparts. Contrary to our hypothesis, sclerostin targeting prevented hallmarks of aortic valve stenosis and indicates that targeted antibody treatments for osteoporosis may be beneficial for these patients regarding aortic stenosis.

NEW & NOTEWORTHY We have found that genetic ablation of the Sost gene (protein: sclerostin) prevents aortic valve stenosis in aged, Western diet mice. This is a new role for sclerostin in the cardiovascular system. To the knowledge of the authors, this is one of the first studies directly manipulating sclerostin in a cardiovascular disease model and the first to specifically study the aortic valve. We also provide a potential new role for Hox genes in cardiovascular disease, noting pan-Hox upregulation in the aortic roots of sclerostin genetic knockouts. The role of Hox genes in postnatal cardiovascular health and disease is another burgeoning field of study to which this article contributes.

INTRODUCTION

Sclerostin (human gene, SOST; mouse gene, Sost) is a secreted glycoprotein that has shown to be a potent regulator of bone remodeling (1). It was first discovered to be the causative root of excessive bone overgrowth in sclerosteosis (full SOST deletion) and van Buchem diseases (deletion of a genetic regulator) (2–5). Further studies clarified the mechanisms whereby sclerostin regulates ossification, resorption, and mechanotransduction in bone by altering the Wnt and receptor activator of nuclear factor-κβ ligand (RANKL) pathways of the skeletal system (6–9).

Because of the protein’s involvement in bone remodeling and its apparent localization to the skeletal system, sclerostin became a target of great interest for the treatment of low bone mass diseases such as osteoporosis. Clinical trials of a monoclonal antibody targeting sclerostin, known as romosozumab, substantially improved bone quality metrics and reduced hospitalization in patients with postmenopausal osteoporosis (10–14). However, the stage III Active-Controlled Fracture Study in Postmenopausal Women with Osteoporosis at High Risk (ARCH) trial noted an imbalance in cardiovascular effects compared with the current standard bisphosponate medication, alendronate. The authors specifically noted a significant increase in cardiac and cerebrovascular ischemic events and a numerical decrease in heart failure (12). This resulted in the drug receiving a warning of cardiovascular side effects which must be weighed when considering prescribing romosozumab to patients who may have, or be at risk for, cardiovascular disease.

The discovery of off-target effects in the cardiovascular system prompted a review of existing literature, as well as new studies directly interrogating the role of sclerostin in various disease models. Although the body of literature on sclerostin’s role in the aortic valve is quite limited, there are some studies of interest. Aortic valves excised from human patients with aortic valve stenosis (AVS) showed that sclerostin protein and mRNA are upregulated in diseased tissue near sites of ectopic calcification (15). A study from the same year found that serum sclerostin levels are associated with valve calcification in individuals undergoing hemodialysis. This was also confirmed at the tissue level using immunohistochemistry and quantitative real-time polymerase chain reaction (qPCR) (16). Although these studies show the correlation between valve disease and sclerostin expression, it is unclear if the increase is a direct contributor to AVS.

The pathways associated with sclerostin signaling in the skeletal system also contribute to the progression of AVS that augment the necessity for direct investigation. Activated Wnt and RANKL signaling have shown to be involved in ectopic calcification of the aortic valve, indicating there may be shared processes sclerostin acts on in the skeletal and cardiovascular systems (17, 18). Furthermore, the risk of a cardiovascular event scales with the severity of postmenopausal osteoporosis measured by vertebral fracture number and severity (19).

AVS is a complicated and poorly understood disease process, and it is unclear whether sclerostin plays a significant role in its progression. To this end, we hypothesized, based on the cardiovascular events in the stage III ARCH trial, long-term loss of sclerostin would worsen AVS in a mouse model. We were unable to secure a sclerostin monoclonal antibody from an industry partner and instead used a genetic model of mice with a complete Sost knockout. Surprisingly, we found Sost ablation had a protective effect in the progression of AVS via reduced myofibroblast activation and aortic valve interstitial cell contractility potentially because of upregulation of protective Hox signaling.

MATERIALS AND METHODS

Animals

All mouse experiments were carried out under appropriate approval and supervision from the Vanderbilt University Institutional Animal Care and Use Committee. Sost knockout mice (Null, Sost^−/−) were provided by Dr. Gabriela Loots and bred as previously described (20, 21). Mice were maintained on the C57BL/6J background strain. Littermates homozygous for the wild-type Sost gene (Sost^+/+) were used as controls (WT). Equal amounts of male and female mice were used. Biological sex variables did not create statistical differences, and all data sets are reported as combined sets, however female data points are delineated as needed in figures.

At 4 mo of age, mice transitioned from standard chow to a 1% cholesterol Western diet (TestDiet 5TJT). Food and water were provided ad libitum. Mice were aged for an additional 8 mo on the Western diet to 12 mo of age, euthanized by carbon dioxide inhalation followed by cervical dislocation, and aortic valve tissue was harvested for processing and analysis (Fig. 1A).

Figure 1.

The effects of genetic ablation of the Sost gene, aging, and high-cholesterol diet were assessed using in vivo and in vitro models. A: mice were given a high-cholesterol diet beginning at 4 mo and aged to 12 mo. The development of AVS was tracked using echocardiography at 4, 9, and 12 mo of age. Black bar = 2 mo. B: in parallel, immortalized mouse lines were generated for WT and Null groups. Aortic valve interstitial cells were isolated and expanded for in vitro analysis. Image partially created with Biorender.com and published with permission. AVS, aortic valve stenosis; WT, wild type.

Cells

Mouse aortic valve interstitial cells (AVICs) were isolated as described previously (22). Briefly, wild-type mice were crossed with the “Immortomouse” line (Charles River, 237 HO, 238 HE) to generate cell lines capable of undergoing prolonged growth under specific culture conditions. Mice were euthanized and hearts were carefully excised under sterile conditions. With a dissection microscope, the aortic root was isolated and the individual leaflets were removed, soaked in 600 U/mL collagenase II (Worthington Biochemical, Lakewood, NJ) for 30 min and centrifuged. Leaflets were then placed on 0.1% gelatin-coated tissue culture dishes in immortalized media (10% fetal bovine serum, 1% penicillin-streptomycin antibiotic, 10 U/mL interferon-γ in Dulbecco’s modified Eagle’s medium), with environmental conditions of 33°C and 5% CO₂. AVICs were allowed to adhere to the culture dish, migrate from the leaflet tissue, and expand for ∼1 wk until confluence was reached (Fig. 1B). To verify the isolation of a myofibroblast cell line, cell morphology was confirmed visually, and the presence of the contractile protein α-smooth muscle actin (αSMA) was confirmed using immunostaining. Before all experiments, cells were maintained for 24 h at 37°C and 5% CO₂ in complete media without interferon-γ (10% FBS, 1% penicillin-streptomycin antibiotic in Dulbecco’s modified Eagle’s medium) to inactivate the immortalization element. Male mice were used to establish all cell lines.

Echocardiography

Echocardiographic assessment was performed at 4, 9, and 12 mo of age to identify changes in cardiac structure and function. All imaging was performed by skilled technicians in the Vanderbilt Cardiovascular Physiology Core using the Vevo 2100 small animal imaging system. Mice were lightly anesthetized (mean, 492 beats/min) using isoflurane and laid supine on a heated platform. Transthoracic aortic pulsed-wave Doppler imaging was used to generate aortic valve velocity profiles. Parasternal short-axis M-mode imaging was performed to measure left ventricular performance and cardiac function as indicated by changes in strain throughout the cardiac cycle.

For aortic valve-specific measurements, a custom MATLAB script was used to automatically trace pulsed-wave Doppler waveforms and determine cardiac-gated hemodynamic metrics such as peak velocity and mean pressure gradient (23). At each time point, three Doppler measurements were gathered for each mouse resulting in 50–100 independent cardiac cycles being averaged to produce representative metrics. Ventricular thickness and motion were manually measured over each cardiac cycle. Three cycles were averaged for each M-Mode image for a total of nine cycles analyzed per mouse per time point.

Dual X-Ray Absorptiometry

Femur bone metrics were measured using the Hologic UltraFocus Dual Energy X-Ray Absorptiometry Vision and associated software. Samples were placed on the stage at ×2 magnification and images were acquired in a series of four captures at 40 kV followed by four captures at 80-kV captures. With the use of a custom region of interest analysis in the system software (version 3.1), the bone mineral content, bone mineral density, and femur length were measured.

Histological Staining

Excised aortic roots were embedded in optimal cutting temperature compound and flash frozen. With the use of a −20°C cryostat, aortic root samples were serially sectioned at 10 μm. For all staining analysis, three sections per aortic root were analyzed for a representative metric. Picrosirius red (Fisher Scientific) staining was performed to assess collagen characteristics and morphology of the aortic roots. Alizarin Red S (Sigma-Aldrich) staining was performed to assess the extent of valve calcification. For both stains, standard manufacturer protocols were followed without deviation. After being stained, slides were dehydrated in progressively concentrated alcohol baths (70%, 90%, 100%), cleared in xylene, mounted in organic mounting media, coverslipped, dried overnight, and sealed before imaging. Bright-field images were captured at a ×4 magnification objective using a Nikon Eclipse E800 microscope, equipped with an Olympus DP74 digital polychromatic camera. Representative images were shown to best illustrate the quantitatively determined phenotype.

Immunohistochemical Staining

Fluorescent immunostaining was used to characterize the prevalence and localization of αSMA (contractile) and Runx2 (osteogenic) proteins, both of which are common markers of aortic valve disease. Slides were washed in phosphate-buffered saline (PBS) before fixation/permeabilization for 10 min in a 4% paraformaldehyde-0.1% Triton solution in PBS. Sections were then blocked for nonspecific binding using 10% bovine serum albumin in PBS for 1 h at room temperature. Primary antibodies were added to a 10% diluted blocking solution at the following concentrations: 1:100 rabbit polyclonal anti-αSMA (Abcam ab5694), 1:100 rabbit monoclonal anti-Runx2 (Cell Signaling Technology 12556) and incubated overnight at 4°C. The following day, sections were washed in PBS before incubation in 1:1,000 goat anti-rabbit IgG Alexa Fluor 647 (ThermoFisher A-21245) for 1 h at room temperature and covered from light. After secondary staining, slides were once again washed in PBS before being mounted in ProLong Gold Antifade with DAPI (Cell Signaling Technology 8961), coverslipped, allowed to dry overnight at room temperature, and then sealed. Fluorescent images were captured at ×20 magnification objective using an Olympus BX53 microscope and Qimaging Retiga 3000 digital monochromatic camera.

Quantitative Image Analysis

For the analysis of Picrosirius red S staining, bright-field images of the stained aortic root were acquired and cropped to contain only the aortic valve leaflets. The total leaflet area was calculated from bright-field images, and collagen composition within the leaflet area was determined as previously described (24, 25). Expression of contractile (αSMA) and osteogenic (Runx2) proteins of interest in the aortic valve was determined from analysis of immunofluorescence images, as previously described (26). Briefly, aortic valve sections fluorescently labeled for either αSMA or Runx2 were stained with DAPI to indicate leaflet nuclei and provide an estimate of leaflet area based on a manually defined boundary. Segmentation of individual nuclei was performed using a modified watershed transform and concave object separation algorithm (24, 27). αSMA area fractions were then computed as the ratio of positive pixels to total pixels within the leaflet boundary. Runx2-positive nuclei were defined based on the ratio of positive pixels to total pixels within the identified nuclear boundary. All analysis was performed using custom MATLAB scripts, which are available on request.

RNA Sequencing

Aortic roots from male WT and Null mice were isolated via microdissection and flash frozen in liquid nitrogen before storage at −80°C until RNA isolation. To collect total RNA from each sample, single aortic roots were homogenized in TRIzol before isolation using the Zymo Direct-zol RNA Microprep kit. RNA integrity was measured with an Agilent bioanalyzer before library preparations. All samples had RNA integrity number values greater than 7. Aortic roots were analyzed as isolating adequate quantities of high-quality RNA from individual leaflets has proven technically infeasible.

The Vanderbilt Technologies for Advanced Genomics (VANTAGE) center performed library preparation, sequencing, and read alignment. Briefly, cDNA libraries were generated from total RNA using the NEBNext Ultra II Directional RNA Library Prep Kit and then sequenced on an Illumina NovaSeq 6000 to an average depth of 63.5 M reads per sample using 150-bp paired-end chemistry. Sequencing quality was assessed using FastQC. Reads were aligned to mouse genome mm10, and gene counts were determined using the Illumina DRAGEN pipeline.

For analysis of RNAseq data sets, differential gene expression and normalization to library size was performed using the R package DEseq2 using Cook’s outliers to filter low gene counts and α = 0.05. KEGG and GO overrepresentation analysis was performed using the R package clusterProfiler using the respective enrich function with default parameters. Gene sets were considered overrepresented if P_adj < 0.05. Visualizations were generated using a combination of enrichplot and ggplot2 packages in R.

Three-Dimensional Cell Contractility Assay

The contractile phenotype of AVICs was assessed using the 3-D Gel Contraction Assay as described previously (28). A collagen gel solution was mixed (80% PureCol, 10% 10× PBS, 10% 0.1 M NaOH) and stored on ice. AVICs were suspended in complete media supplemented in either 0 or 1 ng/mL of recombinant transforming growth factor β1 (TGFβ1, R&D Systems 7666MB005) or 0 or 100 ng/mL of recombinant sclerostin (R&D Systems 1406-ST). Cells were added to the collagen gel solution at a 1:1 ratio to achieve a concentration of 250 cells/μL collagen:media solution. The combined solution was mixed thoroughly to ensure a uniform distribution of cells across the gels. Sterilized Teflon rings (Cat. No. 5612-303-62, Seastrom Manufacturing) were placed in nontissue culture-treated dishes, and 250 μL of the mixture was evenly pipetted in the rings. The mixture of collagen gel, cells, and media was allowed to solidify for 1 h in a 37°C incubator. Solidified gels were covered with complete media supplemented with 0 or 1 ng/mL TGFβ1, and the Teflon rings were carefully removed using sterilized forceps. The gels were allowed to float freely in the media solution. The circular gels were imaged after 24 h of treatment for each condition using a Leica M165 FC stereo microscope, and the contraction ratio was calculated as the change in gel area relative to the initial area (at time = 0 h).

Real-Time Quantitative Polymerase Chain Reaction

WT and Null immortalized VICs were lysed using TRIzol Reagent (ThermoFisher Scientific 15596026). The Direct-zol RNA Miniprep Kit (Zymo Research R2050) was used to isolate high-quality mRNA directly from lysed samples. Isolation was performed by following manufacturer protocols without deviation. Samples were analyzed for concentration and purity using a NanoDrop UV Spectrophotometer (ND-ONE-W). A 260/280 wavelength ratio of ∼2.0 was considered pure RNA and used for downstream analysis. cDNA was synthesized using SuperScript IV First-Strand Synthesis System (ThermoFisher 18091150). SYBR Green PCR Master Mix (ThermoFisher 4309155) was used to allow fluorescent detection during amplification. Manufacturer protocols were followed without deviation. CFX96 Real-Time PCR Detection System (Bio-Rad) was used for thermocycling amplification and measurement of fluorescent signal. cDNA was denatured at 95°C for 4 min, and the following cycle was then carried out 40 times for amplification: 10 s 95°C, 10 s 55°C annealing, and then a plate read to determine fluorescent intensity. After amplification, melt curves were generated for each sample using the following protocol: initialize 65°C, plate read, 0.5°C, plate read, until a maximum of 95°C was reached. Only samples with strong single-peak melt curves around ∼80°C without evidence of primer-dimer formation were considered in the analysis. Bio-Rad CFX Manager 3.1 software was used to quantify amplification by defining detection thresholds and assigning each sample a C_t value. Sample C_t values were normalized to that of a stable housekeeper, Gapdh, and statistics were performed on nontransformed ΔC_t values. Primer sequences are defined in Supplemental Table S1 (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.21280110).

Validated primers were selected from the Harvard PrimerBank database. Sequences were independently verified using the NCBI Primer-BLAST resource to ensure target specification and no secondary binding potential.

Statistical Analysis

All data are presented as means ± SE. The following statistical tests were used: Student’s t test (normal) and Mann–Whitney U test (nonnormal) were used for two-group comparison, one-way analysis of variance (ANOVA) for three or more groups. For paired ANOVA tests, mixed-effects analysis was performed with the Geisser-Greenhouse correction and Tukey’s multiple comparisons test. Multiple U tests were corrected using Holm–Šidák's correction. For all tests, a P value of 0.05 was considered statistically significant. Data storage and statistical analysis were performed using Microsoft Excel, MATLAB r2019a, and GraphPad Prism 9.1.1. Analysis was performed by the authors, but sample identification was blinded and quantitative analysis was performed using custom programs instead of qualitative or manual measurement to reduce bias.

RESULTS

Deletion of the Sost Gene Recapitulates the High-Bone Mass Phenotype Observed Clinically

Human patients lacking the SOST gene display a bone overgrowth phenotype that can be recapitulated in a mouse model (2, 29). Dual-energy X-ray absorptiometry (DXA) scans were used to confirm this phenotype in experimental groups (Fig. 2A). Consistent with prior reports, there was a clear high-bone mass phenotype based on a doubling of bone mineral density in Null mice, relative to WT (51.81 ± 1.12 vs. 106.9 ± 6.25 mg/cm³; P < 0.0001) without a significant change in femur length (13.21 ± 0.13 vs. 13.30 ± 0.22 mm; P = 0.71; Fig. 2A) (30).

Figure 2.

Genetic ablation of Sost results in bone overgrowth and prevention of AVS. A: dual X-ray absorptiometry was used to assess femur bone mineral density (middle) and length (right). Null mice display a sclerotic phenotype with no change in limb length. B: aortic valve hemodynamics were assessed using pulsed-Doppler imaging tracked over time. Standard markers of AVS (peak velocity and mean gradient) were significantly increased in WT groups while remaining unchanged in Null. C: left ventricle measurements were taken by M-mode echocardiography. Both groups had similar left ventricular hypertrophy (left) to compensate with weight increase. Ejection fraction (right) did not display a clear phenotype with no significant differences. A: means ± SE, Students t test. Gray dots, female; white dots, male. B: box and whisker showing interquartile range and maximum to minimum, multiple Mann–Whitney U test with Holm-Šidák's correction. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Scale bar = 5 mm. AVS, aortic valve stenosis; WT, wild type.

Deletion of the Sost Gene Prevents the Hemodynamic Hallmarks of Aortic Valve Stenosis

Hemodynamic parameters across the aortic valve and structure and functional parameters in the left ventricle were assessed using cardiac echocardiography to track the development of aortic valve stenosis between groups. WT mice developed mild-to-moderate increases in aortic valve peak velocity (140.1 ± 7.15 vs. 125.9 ± 9.32, 160.1 ± 10.05 vs. 122.7 ± 7.31, 176.2 ± 13.82 vs. 130.7 ± 6.81; P = 0.72 cm/s, P < 0.05, P < 0.001; 4, 9, 12 mo, respectively) and mean pressure gradient (2.09 ± 0.2 vs. 1.66 ± 0.95, 2.79 ± 0.48 vs. 1.46 ± 0.19, 4.10 ± 0.85 vs. 1.72 ± 0.23 mmHg; P = 0.92, P = 0.22, P < 0.001; 4, 9, 12 mo, respectively), whereas Null mice remained largely unchanged between 4 and 12 mo of age (Fig. 2B). Left ventricular outflow tract diameter was measured and used to normalize peak velocity, and the trend remained the same (Supplemental Fig. S1). Both groups experienced similar levels of hypertrophy in the left ventricle as measured by left ventricle mass (94.49 ± 2.85 vs. 96.48 ± 3.9, 112 ± 3.81 vs. 116.5 ± 4.91, 118.3 ± 5.73 vs. 113.5 ± 5.71 mg; P = 0.99, P = 0.88, P = 0.86; 4, 9, 12 mo, respectively; Fig. 2C). In addition, left ventricle function remained similar over time with little difference between WT and Null groups left ventricle ejection fraction during the heart cycle (46.26 ± 1.1 vs. 44.14 ± 1.19, 50.65 ± 1.65 vs. 45 ± 2.31, 48.13 ± 1.39 vs. 50.66 ± 2.85; P = 0.78, P = 0.07, P = 0.68; 4, 9, 12 mo, respectively; Fig. 2C). Raw echo data separated by sex are found in Supplemental Table S2.

Sost Knockout Mice Have Thinner Leaflets

Leaflet morphology, collagen conformation, and calcification were assessed in aortic valves from WT and Null mice using Picrosirius red (PSR) and Alizarin Red S (ARS) staining (Fig. 3A). Measurement of the full thickness of the aortic valve leaflets showed WT leaflets were generally 30% thicker than Null leaflets (16.52 ± 1.00 vs. 11.48 ± 1.01 µm; P < 0.01), on average (Fig. 3A). Both groups were highly packed with collagen at similar densities (99.59 ± 0.09 vs. 99.53 ± 0.13 µm; P = 0.71) and did not show any clear patterns of calcification on the leaflets (Fig. 3A).

Figure 3.

Sost knockout prevents early myofibroblast-related AVS phenotype. A: Picrosirius red (top) and Alizarin Red S (bottom) were used to assess valve collagen and calcification, respectively. Although there were no trends regarding calcification or collagen, median leaflet was thicker in WT mice indicating mild hypertrophy. B: immunofluorescent staining was used to determine if myofibroblast (αSMA, top) or osteogenic (Runx2, bottom) was driving phenotypic change. αSMA was significantly more prevalent in WT mice, whereas Runx2 was unchanged, indicating a potential myofibroblast AVS phenotype. A: means ± SE, Student’s t test. Gray dots, female; white dots, male. B: means ± SE, Student’s t test. Gray dots, female; white dots, male. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Scale bar magnitudes: 4× = 500 µm; 20× = 100 µm. AVS, aortic valve stenosis; WT, wild type.

Sost Knockout Mice Have Lower Prevalence of Myofibroblast Marker of Aortic Valve Disease

The mechanisms of AVS progression (i.e., dystrophic vs. osteogenic) were assessed using quantitative analysis of αSMA and Runx2 immunostaining (Fig. 3B). Null leaflets had significantly lower expression of αSMA than WT (69 ± 5 vs. 38 ± 6; P < 0.01), a marker of myofibroblast activation and dystrophic calcification associated with early AVS (Fig. 3B). There was no significant difference in Runx2 nuclear expression, a marker of later stage osteogenic calcification (3.6 ± 1.2 vs. 2.0 ± 1.1; P = 0.42; Fig. 3B).

RNA Sequencing of Aortic Roots Reveals Significant Change in Hox Transcription Factors

To determine potential mechanistic explanations for the divergent phenotypes, RNA sequencing was performed on three aortic roots from the WT and Null groups (Fig. 4A). Overrepresentation analysis was performed to identify which genes were interacting with the most GO Biological processes, and the top genes are isolated. Biological processes associated with development, regionalization, connective tissue developments are highlighted (Fig. 4B). Several of the genes in the Hox family of transcription factors were shown to interact with these processes. On further investigation, nearly half of all Hox genes were upregulated in Null groups compared with WT (Supplemental Fig. S2). Six of the most highly regulated are highlighted in the volcano plot (Fig. 4C).

Figure 4.

RNA sequencing of aortic roots shows over 1,000 differentially regulated genes and an increase in pan-Hox gene expression in Null groups. A: schematic of RNAseq read analysis by differential gene analysis, overrepresentation analysis, and Hox identification. B: top overrepresented genes paired with all of the overrepresented GO Bioprocess terms associated. C: volcano plot of differentially expressed genes with the notation of top expressed Hox genes. Blue genes are significantly downregulated in Null mice (n = 388), and red genes are significantly upregulated in Null mice (n = 706).

Several Hox Transcription Factors Are Upregulated in Isolated Aortic Valve Interstitial Cells

AVIC lines were established to assess potential mechanisms contributing to the observed valvular phenotype and Hox genes of interest were assessed using qPCR. Of the 13 Hox genes identified from the RNAseq data, 3 were significantly more highly expressed in Null AVICs (Hoxa1, Hoxb2, Hoxd3). Hoxa7 was significantly downregulated in Null AVICs compared with WT controls (Fig. 5A).

Figure 5.

Null AVICs have altered Hox genes and decreased contractility. A: qPCR was used to assess Hox gene expression in isolated WT and Null AVICs. Hoxa1, Hoxb2, and Hoxd3 were significantly higher in Null mice with Hoxa3 nonsignificantly increased (P = 0.17). Hoxa7 was decreased in Null cells. B: WT and Null AVICs were embedded in collagen gels and treated with 0 or 1 ng/mL TGFβ1 for 24 h. Both groups did not contract with no treatment, but WT AVICs exhibited much greater contractility under treatment when compared with Null. C: WT AVICs were embedded in collagen gels and treated with 0 or 100 ng/mL sclerostin for 24 h. Treatment induced a slight increase in contractility compared with vehicle. A: means ± SE, Student’s t test. B: means ± SE, 2-way analysis of variance with Šidák's multiple comparisons test. White dots, 0 ng/mL TGFβ1; red dots, 1 ng/mL TGFβ1. C: means ± SE, Student’s t test. White dots, 0 ng/mL sclerostin; green dots, 100 ng/mL sclerostin. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Scale bar = 5 mm. AVICs, aortic valve interstitial cells; WT, wild type.

Isolated WT AVICs Are More Contractile than Null

To verify the in vivo data indicating WT valves had a greater level of myofibroblast activation, isolated AVICs from the WT and Null groups were embedded in collagen gels and treated with 0 or 1 ng/mL of recombinant TGFβ1 (Fig. 5B). WT cells were also embedded and treated with 0 or 100 ng/mL of recombinant sclerostin (Fig. 5C). Neither cell type exhibited contractility without stimuli (−2.40 ± 0.95 vs. −1.74 ± 1.14). Addition of TGFβ1 in media caused significant contraction in both cell lines (−2.40 ± 0.95 vs. −76.33 ± 6.54, −1.74 ± 1.14 vs. −28.16 ± 2.8; P < 0.0001, P < 0.001; WT and Null, respectively). Furthermore, WT cells displayed nearly three times as much contractile capacity when compared with Null group (−76.33 ± 6.54 vs. −28.16 ± 2.8; P < 0.0001; Fig. 5B). WT cells treated with recombinant sclerostin protein also exhibited slightly higher contractility than untreated controls (−6.91 ± 1.24 vs. −19.43 ± 2.25; P < 0.001; Fig. 5C).

DISCUSSION

In the current study, we investigated the role of sclerostin in AVS by assessing the effects of Sost genetic ablation on the progression of AVS in an aged Western-diet mouse model. To this end, mice lacking Sost (replaced with a LacZ-neo cassette) were generated and compared with their wild-type littermates. Mice on a C57BL6/J background have been shown to develop hallmarks of AVS (in as few as 4 mo) when receiving a high-cholesterol Western diet (31). Herein, mice were aged for 8 mo on high-cholesterol diet to ensure adequate development of AVS for subsequent analysis (Fig. 1). We have had success with this disease model in previous studies, and based on the results from the current study, we have shown that eliminating sclerostin signaling is sufficient to slow the progression of AVS (26, 32).

In vivo analysis of cardiovascular health using echocardiography indicated the development of hemodynamic hallmarks of AVS in WT mice independent of functional changes in the left ventricle (Fig. 2). Increases in peak velocity and mean pressure gradient across the aortic valve are expected as the mice begin to develop AVS. Sost ablation caused no significant hemodynamic change from baseline over time, indicating attenuated development of AVS in these mice. Therefore, it is possible the Null mice may have been in the earliest stages of AVS development that did not yet manifest at the 12-mo time point in the mouse as this age represents ∼50 yr of human age; valve disease in humans does not typically manifest until advanced old age (70–80 yr and over) (33).

Both WT and Null groups showed consistent increases in left ventricle mass (Fig. 2). This is likely a consequence of the excess cardiac output required because of the weight gain experienced by these mice. Although weight was not tracked over time, the average weight of ∼50 g at 12 mo of age is far greater than average age-matched mouse fed standard chow (Jackson Laboratory). Mouse 12-mo body mass is shown in Supplemental Fig. S3. It is possible this increase in ventricular mass may have contributed to the increase in velocity and pressure across the aortic valve in the WT group; however, the Sost knockout mice did not have altered aortic valve hemodynamics despite having experienced similar levels of cardiac hypertrophy. Taken together, this may indicate a sclerostin-dependent effect that is specific to the aortic valve as opposed to changes in other areas of the heart.

The early aortic sclerosis observed in the data is interesting but presents challenges when considering clinical translation. One in four adults above 60 yr has aortic sclerosis with only 1 in 50 with clinically relevant stenosis (34). Due to the limitations of the animal models previously discussed, it is unknown whether Sost ablation would have any effect on the progression to clinically relevant calcific aortic valve stenosis. Due to the potent effects on bone remodeling, it is unlikely that romosozumab could be applied to the pharmacological treatment of diseases of the aortic valve. Conclusions drawn from the data gathered here should be at most applied to the potential off-target effects of sclerostin targeting on the cardiovascular system of patients with osteoporosis.

Overrepresentation found genes that interacted with a large number of bioprocesses. Some of the most highly involved genes belonged to the Hox family of transcription factors. Approximately half of the Hox genes were expressed in the analyzed tissues, and all of them were higher in the WT group than Null. This family of transcription factors contains four subcategories in vertebrates (A, B, C, D) and is involved in body axis patterning (35). Although the family has long been known to be involved in the development, more recent work has focused on their reactivation in postnatal tissue during cardiovascular disease (36). Of particular interest are the Hoxa1 and Hoxa3 subtypes. Hoxa1 function loss in humans leads to cardiovascular malformations in humans, and knockout mice have been shown to have a myriad of cardiac abnormalities, including outflow tract defects due to the necessity of expression precursor cardiac neural crest cells (37, 38). Hoxa3 knockout has been shown to cause stenosis of the aortic valves, as well as several other cardiovascular abnormalities (39). These protective effects of Hox genes and concurrent upregulation in Null animals may lead to the protective effects in AVS found in this study.

In vitro analysis of mAVICs from WT and Null groups was performed (Fig. 5). The Hox genes that were expressed in mouse tissue were analyzed. qPCR analysis indicated that Null AVICs had upregulated Hoxa1 mRNA, which lends further evidence toward these factors being involved in the protection Null mice from AVS. In addition to Hoxa1, Hoxb2 and Hoxd3 were also increased, although these factors have less of a literature basis for having a protective effect in the valve. Interestingly, Hoxa7 was significantly decreased in Null mice, although there is little evidence that may point to its role in the aortic valve. Hoxa1 and Hoxa3 were considered as they are upregulated in both groups and have been implicated in cardiovascular disease (37–39). Early experiments were carried out to assess the effect of modulating these genes in the contractility assay. WT cells were transfected with GFP, GFP-Hoxa1, and GFP-Hoxa3 overexpression plasmids in the hope that abundance of the transcription factor would reduce contractility. There was no difference in either group; however, this assay only provides a limited assessment of cell contractility (Supplemental Fig. S4). Future studies should focus on defining the role of Hox genes in altering AVIC behavior.

WT and Null cell lines were also assessed for contractility to determine if the myofibroblast phenotype transition may partially explain the in vivo phenotype. Myofibroblast activation is a hallmark of AVS fibrosis and hypercontractility is associated with dystrophic calcific nodule formation (40). Results indicate that WT cells are far more prone to hypercontractility compared with Null, which further supports that Null mice are protected from the early development of AVS by a lower propensity toward myofibroblast activation and contraction. In addition, exogenous supplementation of recombinant sclerostin protein to culture media caused an increase in WT cell contractility, further providing evidence the protein modulates AVIC contractility.

KEGG analysis of RNAseq data indicated a number of differentially regulated signaling pathways that would further indicate a contractile phenotype because of Sost ablation including PI3K-Akt signaling (mmu04151, P = 4.87e-03), regulation of actin cytoskeleton (mmu04810, P = 1.87e-02), and vascular smooth muscle contraction (mmu04270, P = 2.10e-04).

In summary, the results of this study indicate that after 1 year of high-fat diet, WT mice underwent early dystrophic, myofibroblast-dominated progression of AVS with little evidence of osteogenic calcification. Sost ablation seems to have prevented this disease progression in Null animals. Furthermore, RNAseq analysis of aortic roots indicates a promising avenue for the study of the HOX family of transcription factors in modulating sclerostin’s direct role in the progression of AVS.

Limitations

Mice were only aged to 1 yr, which roughly correlates to middle age in humans, whereas the most striking effects of aortic valve disease generally occur in older individuals at 70–80 yr of age. Ideally, mice could be aged to 18 mo or even 2 yr of age but this is often not feasible because mouse health rapidly deteriorates past 1 yr of age when animals are placed on a high-cholesterol Western diet that can lead to premature deaths at unexpected time points. C57BL6/J mice generally do not develop severe aortic valve disease and more significant phenotypes could be generated by using alternative mouse models. Male mice were used to generate RNAseq and in vitro data sets, and future studies should investigate both sexes equally to ensure representative data.

Conclusions

This study presents a novel role for sclerostin in the cardiovascular system, specifically in the aortic valve. Although sclerostin is an excellent target to diminish the effects of low bone mass diseases such as osteoporosis, the existence of off-target cardiovascular effects should better be understood as sclerostin inhibition becomes a more commonly adopted therapy. This study indicates that genetic ablation of the Sost gene in mice slows the development of AVS in a high-cholesterol diet model. This is divergent from the effect in aortic diseases such as aortic aneurysm and atherosclerosis. Our findings further indicate that sclerostin has an effect on early markers of aortic valve disease involved in myofibroblast activation of resident cells through Hox signaling. Future studies should focus on better understanding the mechanisms through which sclerostin acts in diseases of the aortic valve as well as if this genetic effect can be recapitulated in a pharmacological inhibition study. More importantly, in contrast to the cardiovascular events observed in clinical trials, this study indicates a potentially protective role for sclerostin inhibition, at least in the development of AVS, and can further inform clinical decision making as the drug becomes more commonly adopted as a treatment for osteoporosis.

SUPPLEMENTAL DATA

Supplemental Tables S1 and S2 and Supplemental Figs. S1–S4: https://doi.org/10.6084/m9.figshare.21280110.

GRANTS

This study was funded by National Institutes of Health Grants EB021937 and HL151115 (to J.E.J.), HL154596 (to L.A.R.), HL146951 (to M.R.B.), and HL135790 (to W.D.M.); Center for Integrated Healthcare; U.S. Department of Veterans Affairs Grant I01BX004297 (to J.S.N.); and the Leducq Foundation (to W.D.M.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.E.J. and W.D.M. conceived and designed research; J.E.J. and L.A.R. performed experiments; J.E.J., L.A.R., M.R.B., and J.S.N. analyzed data; J.E.J., L.A.R., J.S.N., and W.D.M. interpreted results of experiments; J.E.J. and L.A.R. prepared figures; J.E.J. drafted manuscript; J.E.J., L.A.R., M.R.B., J.S.N. and W.D.M. edited and revised manuscript; J.E.J., L.A.R., M.R.B., J.S.N., and W.D.M. approved final version of manuscript.