Exploring molecular profiles of calcification in aortic vascular smooth muscle cells and aortic valvular interstitial cells

Abstract

Cardiovascular calcification can occur in vascular and valvular structures and is commonly associated with calcium deposition and tissue mineralization leading to stiffness and dysfunction. Patients with chronic kidney disease and associated hyperphosphatemia have an elevated risk for coronary artery calcification (CAC) and calcific aortic valve disease (CAVD). However, there is mounting evidence to suggest that the susceptibility and pathobiology of calcification in these two cardiovascular structures may be different, yet clinically they are similarly treated. To better understand diversity in molecular and cellular processes that underlie hyperphosphatemia-induced calcification in vascular and valvular structures, we exposed aortic vascular smooth muscle cells (AVSMCs) and aortic valve interstitial cells (AVICs) to high (2.5mM) phosphate (Ph) conditions in vitro, and examined cell-specific responses. To further identify hyperphosphatemic-specific responses, parallel studies were performed using osteogenic media (OM) as an alternative calcific stimulus.

Consistent with clinical observations made by others, we show that AVSMCs are more susceptible to calcification than AVICs. In addition, bulk RNA-sequencing reveals that AVSMCs and AVICs activate robust ossification-programs in response to high phosphate or OM treatments, however, the signaling pathways, cellular processes and osteogenic-associated markers involved are cell- and treatment-specific. For example, compared to VSMCs, VIC-mediated calcification involves biological processes related to osteo-chondro differentiation and down regulation of ‘actin cytoskeleton’-related genes, that are not observed in VSMCs. Furthermore, hyperphosphatemic-induced calcification in AVICs and AVSMCs is independent of P13K signaling, which plays a role in OM-treated cells. Together, this study provides a wealth of information suggesting that the pathogenesis of cardiovascular calcifications is significantly more diverse than previously appreciated.

Graphical Abstract

Introduction.

Cardiovascular calcification is a major debilitating process associated with calcium deposition within soft tissues including the vasculature and heart valves. The onset is associated with cardiovascular diseases including atherosclerosis, diabetes, certain hereditary conditions and kidney disease, specifically hyperphosphatemic chronic kidney disease (CKD).[1] Pathological calcification of vascular structures including the aorta and coronary arteries causes vessel stiffening and occlusion, and contributes to coronary artery disease and hypertension.[2] Mineralization of valvular tissue similarly impedes function and causes calcific aortic stenosis. Despite shared risk factors and implicated common molecular pathways, only 25–40% of patients with calcific aortic stenosis present with significant arterial disease, and there are several studies suggesting that valve calcification arises independent of calcium deposition in vascular structures.[2, 3] Therefore, it is considered that these distinct biological phenotypes may be more diverse than previously appreciated.[2]

A major difference between calcification-induced coronary artery disease and calcific aortic stenosis is the tissue site of mineralization. The aortic heart valve is a highly organized trilaminar structure composed of collagens, proteoglycans and elastin fiber layers arranged according to blood flow. The valve extracellular matrix (ECM) is maintained by a heterogenous population of fibroblast-like valve interstitial cells (VICs) interspersed throughout the valve cusp, and an overlying endothelial layer. The connective tissue within the artery walls is also layered, although the structure is distinct from the valve. The outer adventitia layer of the artery is rich in collagen and fibroblasts, while the middle medial layer contains vascular smooth muscle cells (VSMCs) and aligned elastin fibers, and a single cell endothelial layer and sub-endothelial connective tissue from the intimal layer adjacent to laminar blood flow. Calcification of the valve largely occurs in the collagen-rich fibrosa layer situated in regions of oscillatory blood flow and attributed by osteogenic changes in VICs including increased expression of pro-calcification markers and repression of inhibitors, while arterial calcification can take place in both medial and intimal layers, with grossly similar molecular changes in resident VSMCs.[4, 5]

There is a collection of in vitro and in vivo studies to suggest that the molecular and hemodynamic processes that contribute to calcification are conserved between VSMCs and VICs,[6–11] however, there is also a growing body of evidence to suggesting that independent mechanisms also exist.[12–14] Age is a significant risk factor for calcification of both these cardiovascular tissues, but the prevalence of calcific aortic stenosis is significantly lower than that for arterial calcification in those aged over 65 (2–13% versus 31–55%).[15–17] [18, 19]. Furthermore, studies have shown that VSMCs exhibit a more robust calcification response to osteogenic stimuli (greater Alizarin Red and alkaline phosphatase activity) than VICs.[8] Therefore, suggesting that the time required for valve calcification is longer than that needed for vascular mineralization, and/or VSMCs are more susceptible to calcification than VICs.

As clinical manifestations and emerging findings suggest that the underlying processes of calcification in VSMCs and VICs may be different (reviewed [20]).[21, 22], the goal of this study was to utilize ex-vivo, in vitro and mouse model systems to examine common and unique molecular and cellular responses in VSMCs and VICs under calcification conditions induced by hyperphosphatemia. To further determine high phosphate-specific responses, cells were also treated with osteogenic media (OM) as an alternative, phosphate-independent calcific stimulus. Insights from this study will be highly informative for the development of precision medicine approaches to treat the spectrum of cardiovascular calcification phenotypes.

Materials and Methods.

Methods

Cell Culture:

Porcine aortic valve interstitial cells (AVICs) were isolated as previously described,[23, 24] and cultured in control media (DMEM media (containing L-glutamine), 10% fetal bovine serum, 1% penicillin/streptomycin). AVICs were utilized from passage number 2 to 6 for all the experiments reported. Porcine aortic vascular smooth muscle cells (AVSMC) were obtained from Cell Applications (P354–05) and cultured in recommended porcine smooth muscle cell media (P311–500) from Cell Applications.

Calcification Stimuli and Drug Treatment:

To stimulate calcification in AVICs and/or AVSMCs, cells were cultured for 3, or 5 days in osteogenic media (Gibco StemPro^™ Osteogenesis Differentiation Kit, A10072–01, containing StemPro Osteocyte/Chondrocyte Differentiation Basal Medium (A10069–01) and StemPro Osteogenesis Supplement (A10066–01)). For phosphate-induced calcification, sodium phosphate dibasic salt (Sigma #S9390) (pH7.4) was prepared at a stock concentration of 300mM and a working concentration of 2.5 mM was added to cells for 3–5 days. Additionally, the media was supplemented with pyrophosphatase (inorganic from baker’s yeast; Sigma Aldrich; I1643–100UN; used at a concentration of 0.4 units/ml) to facilitate phosphate-mediated calcification. For PI3K/AKT inhibition, AVICs were cultured for 4 days in CM or 2.5mM Ph, or 6 days in CM or OM, in the presence of 10mM LY294002 hydrocholoride (Bio Techne, 1130) or 0.1% DMSO vehicle control[25]. Treatments were replenished every 48 hours. Following treatments, mRNA was extracted, or cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature and subject to Alizarin Red staining or immunohistochemistry (see below).

Aortic Rings:

Hearts were removed from 4 month old C57/Bl6 wildtype mice and rinsed in cold 1XPhosphate Buffered Saline (PBS) several times to remove excess blood. Following, the aortic structure was isolated[26] and thin aortic ring slices were prepared that include the aorta and valve cusps. Aortic rings were then cultured in 35mm dishes in control media, osteogenic media, or 2.5mM sodium phosphate dibasic salt plus pyrophosphatase for 7–8 days. Following, aortic rings were fixed overnight in 4% paraformaldehyde and processed for paraffin-embedding, sectioning, and Alizarin Red staining or immunohistochemistry (see below).

Histology, Alizarin Red Staining and Immunohistochemistry.

Histology:

Paraffin-embedded aortic rings and whole hearts from C57/Bl6 mice were sectioned (7μm thick), deparaffinized and rehydrated as described,[27] then subjected to Alizarin Red staining or immunohistochemistry.

Alizarin Red Staining:

Following fixation, aortic ring tissue sections or treated cells were washed for 10 minutes in 1xPBS and stained with 2% Alizarin Red (dissolved in water (pH-4.2)) for 10 minutes, then further rinsed 3 times in ddH₂0 for 5 minutes each, before being mounted with coverslips using xylene-based mounting media. Images were visualized using EVOS M700 (Invitrogen by Thermo Fisher Scientific, Inc), and Alizarin Red reactivity was quantitated by measuring the total number of reactive pixels (red) normalized to area, or by manually outlining positively stained areas normalized to total area using Image J. Data is presented as an arbitrary unit comparing control media controls to respective treatments, and across treatments, and normalized to area. One-way ANOVA with post-hoc Tukey analysis was performed to determine statistical significance. For sample sizes <n=6, adjusted p-values were calculated using Bonferroni correction.

Immunohistochemistry:

Following fixation, aortic ring tissue sections, treated cells or whole heart tissue sections from wild type or 1.8% phosphate diet-treated (see below) C57/Bl6 mice were washed for 10 minutes in 1xPBS. Tissue sections were subject to antigen retrieval by boiling for 10 minutes in Unmasking Solution at pH6.0 (Vector Laboratories). All samples were immersed in blocking solution (1% BSA, 0.1% Cold water fish skin gelatin, 0.1% Tween 20 in PBS with 0.05% NaN3) for 1 hour at room temperature. Following, primary antibodies were added overnight at 4°C. These include Osteopontin (sc-10593, Santa Cruz, 1:100) and Acta2 (A5228, Sigma, 1:200). After overnight incubation, appropriate secondary antibodies were applied (Alexa Fluor 488 or 586; Invitrogen; 1:400) for 1 hour at room temperature. Tissue sections or cells were then mounted with coverslips using Vectashield anti-fade medium with DAPI (Vector Laboratories). Images were visualized using EVOS M700 (Invitrogen by Thermo Fisher Scientific, Inc). For Figures 1I, 2I, 3I and 3J, Alizarin Red reactivity was determined using Image J by setting a constant threshold value across all samples within experimental comparison groups and normalizing to area (reactivity/area). A similar strategy for Spp-1 immunoreactivity was taken for Figures 3K and L. For Figure 6I, total Acta2 and DAPI immunoreactivity were calculated by setting a constant threshold value across all samples within experimental comparison groups. Normalized values were then calculated and plotted (ACTA2/DAPI). For Figure 6J, Acta2 immunoreactivity was determined as above, but within nodule areas as defined by nuclei clustering. Normalization was determined by area (Acta2/area of nodule). Following one-way ANOVA with post-hoc Tukey analysis was performed to determine statistical significance. For sample sizes <n=6, adjusted p-values were calculated using Bonferroni correction.

Figure 1. High phosphate-induced calcification of aortic valve interstitial and aortic vascular smooth muscle cells.

Porcine aortic valve interstitial cells (AVICs) (A, B, E, F) and porcine aortic vascular smooth muscle cells (AVSMCs) were cultured in control media (CM) (A, C, E, G) or CM containing 2.5mM Phosphate (Ph) for 3 (B, D) or 5 (F, H) days. Following, cells were fixed and stained for Alizarin Red to detect calcium deposition indicative of calcification. Arrows indicate regions of positive reactivity. (I) Quantitation of Alizarin Red reactivity normalized to area of 2.5mM Phosphate-treated AVICs and AVSMCs at 3 and 5 days of culture. Individual data points for biological replicates are indicated. Errors bars are based on standard deviation values, and p-values (<0.05) were obtained by one-way ANOVA with post-hoc Tukey analysis. As sample sizes are <n=6, adjusted p-values were calculated using Bonferroni correction. *p, <0.05 in 2.5mM Ph over Cond. Media controls, #p, <0.05 between AVICs and AVSMCs with 2.5mM Ph treatment. (J, K) 4 week old wildtype C57/Bl6 male and female mice were placed on 1.8% phosphate diet without adenine for 8 weeks, in parallel with littermate controls fed regular chow. Following Alizarin Red was performed on prepared tissue sections. Note reactivity in regions of the aortic arch of the aorta (Ao) and not the aortic valve (arrows) with 1.8% phosphate diet. Inset in (K) is a higher magnification of boxed area howing alizarin red reactivity within the aortic vessel wall. (L) Summary of Alizarin Red reactivity detected in the aortic arch, and aortic valve regions from control and 1.8% Phosphate fed mice.

Figure 2. Osteogenic media-induced calcification of aortic valve interstitial and aortic vascular smooth muscle cells.

Porcine aortic valve interstitial cells (AVICs) (A, B, E, F) and porcine aortic vascular smooth muscle cells (AVSMCs) were cultured in control media (CM) (A, C, E, G) or osteogenic media (osteo media) for 3 (B, D) or 5 (F, H) days. Following, cells were fixed and stained for Alizarin Red to detect calcium deposition indicative of calcification. Arrows indicate regions of positive reactivity. (I) Quantitation of Alizarin Red reactivity normalized to area of osteo media-treated AVICs and AVSMCs at 3 and 5 days of culture. Individual data points for biological replicates are indicated. Errors bars are based on standard deviation values, and p-values (<0.05) were obtained by one-way ANOVA with post-hoc Tukey analysis. As sample sizes are <n=6, adjusted p-values were calculated using Bonferroni correction. *p, <0.05 in Osteo Media over Cond. Media controls, #p, <0.05 between AVICs and AVSMCs with Osteo Media treatment. (J) Percent of phospho-histone H3 (pHH3) positive cells (AVICs, AVSMCs) per imaging area following 48 hour culture in control media, osteo media and 2.5mM phosphate (Ph).

Figure 3. High phosphate- and osteogenic media-induced calcification of ex-vivo aortic rings.

Aortic rings isolated from 4 month old C57/Bl6 wildtype mice were cultured in control media (Cond. Media) (A, E, C, G), 2.5mM Phosphate (Ph) (B, F), or osteogenic media (Osteo Media) (D, H) for 8 (A, B, E, F) or 7 (C, D, G, H) days. Following tissue sections were prepared from aortic rings tissue and stained for Alizarin Red (A-D) or subjected to immunohistochemistry (E-H) to detect expression of the calcification marker, Osteopontin (Opn) (Spp1). Arrows indicate reactivity within regions of the aorta (Ao) while arrowheads indicate reactivity within the aortic valve cusps (AoV). (I, J) Quantitation of Alizarin Red reactivity normalized to area within defined regions of the aorta and aortic valve cusps in aortic rings treated with 2.5mM Ph (I) or osteogenic media (J). (K, L) Quantitation of Spp-1 immunoreactivity normalized to area within defined regions of the aorta and aortic valve cusps in aortic rings treated with 2.5mM Ph (K) or osteogenic media (L). Individual data points for three biological replicates are indicated. Errors bars are based on standard deviation values, and p-values (<0.05) were obtained by one-way ANOVA with post-hoc Tukey analysis. *p, <0.05 in treated aortic rings over Cond. Media controls, #p, >0.05 between AVICs and AVSMCs with treatment as indicated.

Figure 6. Actin stress fiber-related changes in aortic valve interstitial cells under calcific stimulus.

mRNA findings and supporting immunohistochemistry to detect alpha smooth muscle actin (SMA) expression in aortic valve interstitial cells (AVICs) and aortic vascular smooth muscle cells (AVSMCs) treated with osteogenic (osteo) media or 2.5mM Phosphate (Ph) for 5 days. (A, B) RNA-seq findings showing fold change in expression of “actin stress fiber”-related genes in AVICs and AVSMCs treated with Ph (A) and osteo media (B) relative to control media (Cond. Media) controls after 5 days of culture. *p, <0.05 in treated cells over Cond. Media controls. (C-H) Immunohistochemistry to detect SMA expression in AVICs (C-E) and AVSMCS (F-H) cultured in Cond. Media (C, F) or treated with Osteo Media (D, E, G, H) for 5 days. (D, G) SMA immunoreactivity is shown in green, while DAPI in blue indicates cell nuclei. Arrows highlight cell aggregates indicative of calcific nodules. Arrowheads in (H) indicate SMA-positive cells within the calcific nodule formed by Osteo Media-treated AVSMCs. (I) Quantitation of total SMA immunoreactivity in each magnification field normalized to DAPI in AVICs and AVSMCs treated with Osteo Media (OM) over Cond. Media controls (n=3 for each). (J) Quantitation of total SMA immunoreactivity within defined calcific nodules (based on cell clustering) normalized to area in AVICs and AVSMCs treated with Osteo Media (OM) over Cond. Media controls (n=3 for each). *p, <0.05 in OM-treated AVSMCs over OM-treated AVICs.

Animal Studies:

All mice for this study were housed in a USDA-certified, AALAC-accredited facility within the Medical College of Wisconsin and monitored by the investigator and animal care technicians. All animal protocols for the proposed studies have been approved by the Institutional Animal Care and Use Committee (IACUC #AUA0006769), and all individuals involved in this study have received general animal training. 4 month old C57/Bl6 wildtype mice were used for the aortic ring (Figure 3), whole heart tissue section studies, and high phosphate diet studies (Figure 1). For the latter, 4 week old mice were placed on 1.8% phosphate diet without adenine (TD. 170459, Envigo) for 8 weeks, in parallel with littermate controls fed regular chow. Following, all mice were euthanized using CO₂ anoxia, as recommended by the Panel of Euthanasia of the American Veterinary Medical Association (AUA#00006769), and efforts were made to ensure the mice do not suffer any unnecessary pain or discomfort.

RNA-Seq and Analysis:

Porcine AVICs and AVSMCs were cultured as previously described and harvested either 3 or 5 days after treatment in control media, osteogenic media, or 2.5mM phosphate. 200ng total RNA was extracted using the RNeasy Mini Kit from Qiagen (#74104), quantified using NANOdrop, and sent to Psomagen, Rockville, MD for analyses. Amplified cDNA libraries suitable for sequencing were prepared from 200 (ng) of DNA-free total RNA using the Universal Plus mRNA-Seq Library Prep Kit (NuGEN Technologies, Inc. #0508–96). The RNA Integrity Number (RIN) was determined using Ribogreen on a TapeStation to measure concentration and quality, with a value of >7.0 meeting threshold. Libraries were prepared using the TruSeq Stranded mRNA LT Sample Prep kit and sequenced on an Illumina platform. Raw FASTQs were split into files containing ~40,000,000 total reads and checked for quality using the FASTX-Toolkit. The reads were filtered (removing sequences that did not pass Illumina’s quality filter) and trimmed based on the quality results (3 nucleotides at the left end of the R1 reads and 1nt at the left end of the R2 reads). Trimmed reads were mapped to reference genome (Sus Scrofa, GCF_000003025.6_Sscrofa11.1) with HISAT2. After the read mapping, Stringtie was used for transcript assembly. Expression profiles were calculated for each sample and transcript/gene as read count and FPKM (Fragment per Kilobase of transcript per Million mapped reads). All raw data from this bulk RNA-seq analysis is available through the Gene Expression Omnibus (GEO), accession number: GSE227131.

Differentially Expressed Genes and Pathway Analysis.

DEG (Differentially Expressed Gene) analysis was performed on comparison pairs as indicated (Test_vs_Control) using DESeq2, and Figures 4A, B show the number of genes which satisfied |fc|>=2 & nbinomWaldTest raw p-value<0.05 conditions in comparison pairs. DEGs were further analyzed with gProfiler (https://biit.cs.ut.ee/gprofiler/orth) for gene set enrichment analysis per Gene Ontology (GO) terms of biological process (BP), cellular component (CC) and molecular function (MF),[28] as well as subject to analysis through the DAVID platform (https://david.ncifcrf.gov).[29, 30] In addition, Venn diagrams based on DEG data sets (Figure 5) were generated using Venny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/).[31]

Figure 4. RNA-seq analysis of high phosphate- and osteogenic media-induced calcification of aortic valve interstitial and aortic vascular smooth muscle cells.

Bulk RNA-seq analysis was performed on porcine aortic valve interstitial (AVICs) and aortic vascular smooth muscle cells (AVSMCs) treated with 2.5mM Phosphate (Ph) or osteogenic media (OM) for 3 or 5 days for three biological replicates. (A, B) The number of up- and down- differentially expressed genes (DEGs) at the 3 (A) and 5 (B) day time points based on pair-wise analysis. (C, D) Hierarchical heat map (C) and principal component analysis (PCA) plot (D) of all 36 RNA-seq samples.

Figure 5. Venn diagrams based on comparisons between two, pairwise analysis groups.

Venn diagrams to indicate the number of common and uniquely expressed differentially expressed genes (DEGs) between two, pairwise analysis groups as indicated. (A) Aortic vascular smooth muscle cells (AVSMCs) treated with osteogenic media (OM) vs. control media (CM) for 3 days, compared to AVSMCs treated with 2.5mM Phosphate (Ph) vs CM for 3 days. (B) AVSMCs treated with osteogenic media (OM) vs. CM for 5 days, compared to AVSMCs treated with Ph vs CM for 5 days. (C) Aortic valve interstitial cells (AVICs) treated with OM vs. CM for 3 days, compared to AVICs treated with Ph vs CM for 3 days. (D) AVICs treated with OM vs. CM for 5 days, compared to AVICs treated with Ph vs CM for 5 days. (E) AVSMCs treated with OM vs. CM for 3 days, compared to AVICs treated with OM vs CM for 3 days. (F) AVSMCs treated with Ph vs. CM for 3 days, compared to AVICs treated with Ph vs CM for 3 days. (G) AVSMCs treated with OM vs. CM for 5 days, compared to AVICs treated with OM vs CM for 5 days. (H) AVSMCs treated with Ph vs. CM for 5 days, compared to AVICs treated with Ph vs CM for 5 days.

PCA Analysis.

Principal Component Analysis (PCA) was performed on the FPKM data for all detectable (>= RQT in at least 25% of the samples) protein-coding Sus scrofa genes (removing those with zero gene expression). The FPKM values were log10-transformed, centered so that each sample had mean 0, and the first three principal components were calculated. To find genes that substantially contribute to each PC value, the correlation and fold-change in expression of each gene with the first three principal components was calculated according to the methods of Sharov et al.[32]

Hierarchical Clustering.

FPKM data for all protein-coding genes with detectable mRNA levels (removing those with zero gene expression) was used for hierarchical clustering analysis by Cluster 3.0 software [33]. Genes were median centered prior to hierarchical clustering. Hierarchical clustering was conducted using centered correlation as the similarity metric and average linkage as the clustering method. Along with clustering gene expressions was shown based on supplied sample grouping in the form of a heatmap for the genes with a Tukey FDR <0.1.

Taqman Gene Expression Array Cards.

AVICs and AVSMCs were cultured in CM, OM, or 2.5mM Ph as for 5 days as described above. RNA was exacted using Trizol reagent (Invitrogen by Thermo Fisher Scientific, Inc) following the manufacturer’s protocol. RNA was converted to cDNA using SuperScript VILO Master Mix (Invitrogen by Thermo Fisher Scientific, Inc). 10ng/μl final concentration of cDNA was mixed with TaqMan Fast Advanced Master Mix (Applied Biosystems by Thermo Fisher Scientific, Inc) and loaded onto a custom TaqMan^™ Gene Expression Assay TaqMan^™ Array Cards (Design ID: RTEPTVD Lot: B7707, Applied Biosystems by Thermo Fisher Scientific, Inc) and ran using the ViiA 7 Real-Time PCR System (Applied Biosystems by Thermo Fisher Scientific, Inc) per the manufacturer’s protocols. Data was analyzed using the Relative Quantification App for realtime qPCR on the Thermo Fisher Cloud Connect Platform. Unpaired, parametric t-test was performed to determine statistical significance. For sample sizes, VSMCs n=4, VICs n=3.

Results.

Aortic Vascular Smooth Muscle Cells Have Increased Susceptibility To Calcification.

High phosphate conditions are known to promote calcification of VSMCs and VICs within vascular and valvular tissue respectively. However, similarities or differences in the pathobiology of calcification between these two cell types has not been extensively examined. To address this, porcine aortic VICs (AVICs) (Figures 1A, B, E, F) and aortic VSMCs (AVSMCs) (Figures 1C, D, G, H) were cultured for 3 (Figures 1A–D) or 5 (Figures 1E–H) days in control media (CM) (Figures 1A, C, E, G) or CM supplemented with 2.5mM inorganic phosphate (2.5mM Ph) (sodium phosphate dibasic salt) to create hyperphosphatemic conditions.[34] Following, Alizarin Red staining was performed to detect calcium deposits (Figures 1A–H) as an indicator of calcification. As quantitated in Figure 1I, 2.5mM Ph conditions significantly increased Alizarin Red reactivity in both AVICs and AVSMCs at 3 and 5 days, however, reactivity was higher in AVSMCs at the earlier time point, suggesting that calcium deposition occurs sooner and is more severe in this cell type under these in vitro conditions.

In parallel with the in vitro system, differences in phosphate-induced calcification susceptibility were further explored in vivo by feeding 4 month old wild type C57/Bl6 mice a high phosphate (1.8%) diet (Figure 1K), or normal chow (Figure 1J) for 16 weeks. As shown in Figures 1J–L, histological examination of tissue mineralization by Alizarin Red staining demonstrates that 67% of mice fed a high phosphate diet developed calcification of the aortic arch in the absence of aortic valve mineralization. Isolated valve calcification was not detected in any mice, although Alizarin Red reactivity within the valve cusps was observed in 11% of mice that also exhibited aortic arch calcification. 22% did not develop detectable cardiovascular calcification. Like previous reports[35] serum phosphate levels were not significantly increased in 1.8% Ph diet-fed mice (16.34mg/dL±0.67mg/dL (1.8% Ph diet), versus 13.37mg/dL±3.08mg/dL (control)).

To determine if differences in calcification susceptibilities between AVSMCs and AVICs are limited to high Ph conditions, in vitro experiments were repeated substituting 2.5mM Ph for an alternative calcification stimulus, osteogenic media (OM) (Figures 2B, D, F, H) versus control media (Figures 2A, C, E, G). As expected, OM increases Alizarin Red reactivity over CM controls (Figure 2I), and similar to 2.5mM Ph, reactivity is significantly higher in AVSMCs treated with OM compared to AVICs after 3 days, and this robust increase in reactivity seen in AVSMCs continues at the later 5 day time point.

To confirm that changes in cell proliferation caused by differences in culture conditions were not influencing calcification, treated cells were subjected to immunohistochemistry to detect pHH3 expression as a marker of mitosis after 48 hours. As shown in Figure 2J, significant differences were not observed.

To further validate these findings, we introduced an ex-vivo model. In brief, aortic rings containing both aortic valve cusps and aorta were isolated from 12–16 week wild type C57/Bl6 mice and cultured in either 2.5mM Ph with pyrophosphatase for 8 days (Figures 3B, F), or OM (Figures 3D, H) for 7 days, with respective timed CM controls (Figures 3A, C, E, G). These time points were based on optimization of Alizarin Red reactivity. Following treatment, aortic rings were subjected to Alizarin Red staining (Figures 3A–D), in addition to immunohistochemistry to detect expression of Osteopontin (Opn) (Spp1) (Figures 3E–H) as a molecular indicator of calcific changes.[36, 37] Quantitative analyses of Alizarin Red and Opn reactivity within the aorta and aortic valve cusps of treated aortic rings indicates higher levels of calcification with 2.5mM Ph treatment compared to CM controls (Figures 3I, K), and there is a strong association between Alizarin Red reactivity and Spp-1 localization (Figures 3B, F). Like cultured cells, Alizarin Red is significantly greater in the aorta than the aortic valve cusps (Figure 3I). Following 7 days of OM treatment, Alizarin Red (Figure 3D) and Spp1 (Figure 3H) reactivity are detected in the aorta (Figure 3J), and while detectable levels of Alizarin Red were not consistently observed in the aortic valve cusps (Figure 3J), Figure 3H shows an example of some Opn reactivity within the commissure region (arrow). Together, these data using in vitro, ex-vivo and in vivo systems suggest that under calcification conditions, vascular structures (AVSMCs) are more susceptible to calcification than aortic valve cusps (AVICs).

Molecular Profiling Captures The Diversity Of Calcification Processes Between Aortic Vascular Smooth Muscle Cells And Aortic Valve Interstitial Cells.

To determine differential, or common molecular profiles underlying calcification between AVSMCs and AVICs in response to 2.5mM Ph, or OM as a phosphate-independent stimuli, bulk RNA-sequencing was performed. A total of 6 conditions were collected for each cell type (a total of 12 experimental groups for 3 biological replicates). This included CM and 2.5mM Ph for 3 and 5 days, in addition to CM and OM for 3 and 5 days for both AVICs and AVSMCs. Following sequencing data analysis, trimmed reads were mapped to the sus scrofa reference genome using HISAT2, followed by transcript assembly by Stringtie. Figures 4A and B indicate the number of differentially expressed genes (DEGs) following DEQeq2 analysis for each experimental treatment group (2.5mM Ph, OM) versus respective CM controls, in each cell type (AVICs, AVSMCs), at 3 (Figure 4A) and 5 (Figure 4B) days. The number of DEGs for each comparison pair are based on a stringency threshold of a fold change>2, and corrected p-value<0.05.

Overall, the number of DEGs upregulated for each comparison pair is similar to the number downregulated, and the total number of DEGs across each experimental group at day 3 (Figure 4A), versus day 5 (Figure 4B) is comparable. Supplementary Table 1 indicates the top 10 DEGs for each pairwise comparison group. It is noted that the total number of DEGs is significantly higher in both cell types treated with OM (vs. CM) compared to 2.5mM Ph treatment (vs. CM), particularly at the 3 day time point. For example, a total of 1196 mRNAs are differentially expressed in response to OM treatment in AVSMCs, and 1377 for AVICs. While in contrast, only 144 (AVSMCs) and 573 (AVICs) DEGs are identified under 2.5mM Ph conditions.

Hierarchical heat map analysis (Figure 4C) based on DEGs, reveals positive clustering amongst biological replicates of experimental groups indicative of low sample variation, with notable molecular diversity between AVICs and AVSMCs regardless of treatment (see clustering of AVSMCs groups on the left, and AVICs on the right). Additional principal component analysis (PCA) further supports cell-specific molecular diversity with AVIC experimental groups clustering below the line indicating 0% of the primary principal component (Dim2, x-axis), and AVSMCs above the line. In addition, the PCA highlights molecular similarities amongst the 3 day treated samples (above 0% line of Dim2), that show variance from 5 day treated samples (below 0% line of Dim 2). The segregation of OM, versus 2.5mM Ph-treated groups is less well defined. These visual analyses suggest that in response to calcific stimuli, the molecular responses of AVICs and AVSMCs at early (day 3) and late (day 5) time points are diverse.

Pathway Analysis Reveals Cell- and Calcific Stimulus-Dependent Responses Following 2.5mM Phosphate Treatment, compared to Osteogenic Media.

Gene Ontology (GO) pathway enrichment analysis (“biological processes”, “molecular function” and “cellular compartments”) was performed on all DEGs that passed the threshold criteria (fold change >2, and corrected p-value<0.05) for each pairwise comparison group, and bubble plots representing the top 20 GO terms are shown in Supplementary Figure 1 (day 3) and Supplementary Figure 2 (day 5).

To better identify cell-specific (AVSMCs, AVICs) mRNA profiles that were unique to 2.5mM Ph treatment or shared with OM media, we identified DEGs that were commonly upregulated or downregulated between 2 pairwise comparison groups, and DEGs that were differentially expressed in only one pairwise comparison group (unique). The number of common and unique DEGs from each comparison analysis are indicated in Venn diagrams in Figure 5, and corresponding DEGs within each section of the Venn are included in Supplementary Table 2. For example, Figure 5A indicates that 557 DEGs are upregulated in the AVSMC OM vs. CM group, compared to AVSMC 2.5mM Ph vs. CM group, while 432 are downregulated. Furthermore, a total of 10 DEGs (7 up, 3 down) are common to both these pairwise groups. Similar large differences in the number of DEGs between the pairwise comparison groups are found throughout the analyses, as evident by few ‘common’ DEGs. Only 1–5% of all DEGs are commonly up- or down-regulated between comparison groups which highlights the molecular diversity between treatments, cell types and temporal responses. The exception is Figure 5E, that indicates 19% DEG commonality between AVSMCs and AVICs treated with OM, which suggests a higher degree of shared molecular responses at least early, as by day 5, overlap is reduced to 1%. Overall, these data are consistent with the heat map (Figure 4C) and PCA analysis (Figure 4D) and further highlight the differential responses by AVSMCs and AVICs to calcific stimuli. These data suggest that calcification induced by OM leads to a greater molecular response than 2.5mM Ph, irrespective of cell type. To validate findings identifying unique mRNAs within comparison groups, high throughput quantitative real-time PCR was performed to detect the mRNA abundance of 68 uniquely expressed genes in independent samples of AVICs and AVSMCs cultured in CM, OM or 2.5mMPh for 5 days. As shown in Supplementary Figure 3, 37 of the 68 genes showed similar significant increases or decreases in expression. 28/34 genes validated in treated AVSMCs, and 9/34 in treated AVICs.

Next, we performed pathway analysis (KEGG and REACTOME) on the list of DEGs within each section of the Venn diagrams from Figure 5 to determine GO-terms that might reflect differences and similarities in the response of AVSMCs and AVICs to 2.5mM Ph, versus OM. Corresponding GO terms and DEG lists that reached statistical significance are included in Supplementary Table 3. GO analysis on the DEG datasets shown in Figure 5 reveal enrichment of biological processes related to ‘ossification/bone mineralization’ in seven (Figures 5A, 5B, 5C, 5D, 5F, 5G, 5H) of the 8 comparison groups. Although, the associated DEGs within the GO terms are cell-, treatment-, and time-specific.

As shown in Supplementary Table 3 “Ossification Table”, we did not observe DEGs associated with ‘ossification/bone mineralization’ GO terms that were common to all seven pairwise analyses. Although Wnt ligands (Wnt11, Wnt5, Wnt10B) were upregulated in six out of the seven comparisons (absent in AVIC CM vs. AVIC OM, d3), and increases in Runx3 and Srgn were common to five out of the seven (absent from AVIC CM vs. AVIC OM at days 3 and 5). Spp1 (osteopontin) is highly expressed in human calcified valves and frequently used as a marker of calcification pathology,[36, 38–40] however in this study, increases were only observed in three groups (AVSMC CM vs. AVSMC OM, d3; and AVIC CM vs. AVIC OM, d3 and d5). OM-specific responses also include upregulation of Col13a, Tgfb3 and Zbtb16 in both AVSMCs and AVICs. Further, AVICs upregulate their own set of DEGs with OM treatment including, Atp6vba1, Clec3b, Fgr, Smoc1, Id3, Ibsp, Mmp9 and Irflnb, while Adrb2, Axin2, Bmp4, Bmp6, Itgb6, Mgp, Notum, Phopho1 and Pth1r are unique to OM-treated AVSMCs. Interestingly, upregulated mRNAs associated with ‘ossification/bone mineralization’ following 2.5mM Ph, were only observed in treated AVICs, and not AVSMCs. Notably, high phosphate-specific increases in ‘ossification/bone mineralization’-related DEGs were not noted, although 2.5mM Ph-treated AVICs for 5 days leads to unique downregulation of Gdf10, Omd, Crim1, Fgf18, Npnt, Sost, Snai1. We also note cell-specific responses to OM and 2.5mM Ph treatments, although p-values of ‘ossification/bone mineralization’-associated DEGs are generally higher in AVICs. In comparison to AVSMCs, AVICs upregulate a unique set of ossification-related DEGs including Alox2, Comp, Cypb21, Ddr2, Ecm1, Oml, Spp1, Wny10B, Ptn and Sox9, and despite lower significance of GO-terms, AVSMCs uniquely upregulate Axin2, Gpc3 and Mgp. A summary of these findings is included in Supplementary Table 3 “Ossification Table”.

In addition to enrichment of ‘ossification/bone mineralization’ GO terms, additional analyses also reveals overrepresentation of downregulated DEGs associated with biological processes related to “actin” (see Supplementary Table 3 “Actin-related GO terms”). Notably, we observe cell-specific downregulation of actin-related genes (Acta2, Cnn1, Tagln, Mhy11, etc.) in AVICs following treatment with 2.5mM Ph and OM treatment (Figures 6A, B). To validate this, pAVSMCs and pAVICs were cultured in OM for 5 days, then subjected to Acta2 immunohistochemistry. Despite significance differences at the transcript level, overall immunoreactivity of total Acta2 protein did not significantly change between treatments (Figure 6I), although immunoreactivity levels were overall very high (Figures 6C–H, I) likely due to the known influence that tissue culture plastic has on inducing Acta2 expression.[13, 41, 42] Upon closer examination, it is noted that within individual calcific nodules (as denoted by clusters of DAPI), Acta2 immunoreactivity is reduced in OM-treated VICs (Figure 6D, E, J), while expression remains high in nodules formed by AVSMCs (Figure 6G, H, J). This may suggest that in response to calcific stimuli, actin fiber organization is altered in calcific nodules formed by VICs, but not AVSMCs.

KEGG pathway analysis further revealed calcific stimulus-dependent differences between 2.5mM Ph and OM. As shown in Supplementary Table 3 “KEGG Reactome”, DEGs associated with the ‘PI3K/AKT signaling pathway’ are significantly increased in AVICs and AVSMCs treated with OM (over CM), compared to 2.5mM Ph (over CM) at day 3. Several upregulated DEGs in OM-treated cells related to this pathway are common to both cell types (Gng4, Angpt2, Col4a6, Itga1, Itga7, Spp1, gkK2 and Tlr2) and through KEGG pathway mapping, largely associate with PI3K/AKT. To validate these findings and further determine if PI3K/AKT signaling plays a role in OM-induced calcification, AVICs were cultured in 2.5mM Ph, OM or CM for 4, or 6 days respectively with either the PI3K/AKT inhibitor, LY290045, or 0.1% DMSO vehicle control. Figures 1 and 2 show that 2.5mM Ph leads to more rapid and severe calcification than OM, and therefore days 4 and 6 were selected to. obtain comparable levels of alizarin red reactivity. Alizarin Red reactivity was not detected in AVICs cultured in CM alone at 4 and 6 days (data not shown). As expected, both OM and 2.5mM Ph treatments led to Alizarin Red reactivity (Figures 7A, B), but notably reactivity is significantly reduced in AVICs cultured in OM with LY2940002 (Figures 7C, E). This attenuation of reactivity by LY294002 was not seen in AVICs cultured in 2.5mM Ph (Figures 7D, E), suggesting a PI3K/AKT independent mechanism. Together, these data identify cell- and treatment-specific signaling pathways during the in vitro calcification process.

Figure 7. Inhibition of PI3K/AKT signaling reduces osteogenic media-, but not high phosphate-induced calcification in aortic valve interstitial cells (AVICs).

(A-D) Alizarin Red staining of AVICs cultured in osteogenic media (Osteo Media) (A, C) or 2.5mM Phosphate (Ph) (B, D) in the presence of 0.1% DMSO (vehicle) (A-B), or 10mM PI3K/AKT inhibitor, LY2940002 (C, D), for 4–6 days as indicated. (E) Quantitation of Alizarin Red reactivity normalized to area of treated AVICs. Individual data points for biological replicates are indicated. Errors bars are based on standard deviation values, and p-values (<0.05) were obtained by one-way ANOVA with post-hoc Tukey analysis. As sample sizes are <n=6, adjusted p-values were calculated using Bonferroni correction. *p, <0.05 in LY294002 treated AVICs over DMSO vehicle controls for each calcific stimulus.

Discussion.

Cardiovascular calcification is an increasing worldwide health and economic burden, and patients with hyperphosphatemic chronic kidney disease are at greater risk of developing coronary artery calcification (CAC) and calcific aortic valve disease (CAVD). VSMCs and VICs are key drivers of these pathological processes respectfully, and while they are largely interchangeable when describing cardiovascular calcification, there are clear phenotypic, mechanistic and clinical differences. This is further highlighted in this study, where we show that AVSMCs undergo calcification sooner than AVICs, and this is most severe under high phosphate (versus OM) conditions. These cell-, and calcific stimuli-specific differences are further reflected in findings from transcriptomic analyses. This includes diversity of DEGs associated with ‘ossification/bone mineralization’ biological processes, in addition to VIC-specific downregulation of ‘actin’-associated mRNAs and involvement of P13K/AKT signaling. Interestingly, while OM treatment elicited specific molecular responses in both cell types, high phosphate-specific molecular changes were not observed. Overall, this descriptive, but highly informative study provides a previously unappreciated dataset that can be used by the community to gain better insights into the mechanistic underpinnings of calcific disease in vascular and valvular structures that can be used for more precision medicine approaches in the future.

Cardiovascular calcifications affecting heart valves and arteries share several risk factors and on occasion have been considered as one disease. However, there are several lines of evidence to suggest that CAVD and CAC represent multifaceted disorders (reviewed [2]). In the majority of cases calcific aortic valve disease (CAVD) is isolated in the absence of arterial calcification, and only 25–40% exhibit both CAVD with coronary artery calcification (CAC). Furthermore, studies have reported CAVD in ~2–5% of those aged 65 or over,[43] while others highlight an incidence of CAC in ~28.4% at the age of 50.[44] These clinical observations are further supported by this study demonstrating a degree of diversity in the response of AVSMCs and AVICs to phosphate-dependent and phosphate-independent calcific stimuli. The underlying reason for such discrepancies is unclear, however, in vivo, there are obvious anatomical and structural differences between vascular and valvular structures, in addition to diverse in vivo environments (biomechanics, ECM, neighboring cell types), and some of the phenotypic variation may be attributed to developmental origins and heterogeneity of the major cell types involved. [2][45–47]

Previous studies by us and others have suggested that vascular and valvular calcification involves the differentiation of VSMCs and VICs towards an ‘osteoblast-like lineage’, and collectively the field has identified a signature of molecular markers that are commonly expressed during bone development and increased by VSMCs and VICs under calcific stimulation.[20, 48–50] These include growth factors (e.g., Bmp2), transcription factors (e.g., Runx2), ECM markers (e.g., Spp1) and well as inhibitor regulators of ossification (e.g., Matrix Gla Protein (Mgp)). In association with Alizarin Red reactivity, RNA-seq analysis from this study revealed significant enrichment of mRNAs associated with ‘ossification/bone mineralization’ GO terms (see Supplementary Table 3) that includes previously unappreciated indicators of calcification pathogenesis that are cell-, stimulus-, and time-dependent. Worthy of mention, we did not observe increased expression of Runx2; an established marker of osteoblasts, and commonly used by the field to determine differentiation of VSMCs and VICs as they transition towards a bone-like lineage during calcification. The lack of change in Runx2 expression in this current study raises questions over the robustness of this marker as a definitive indicator of indicator of the disease process.

Interestingly, our data identified a novel, molecular profile of ‘ossification/bone mineralization’ that was specific to AVICs, and not detected in AVSMCs, irrespective of calcific stimuli. Pathway analysis of AVIC-specific upregulated DEGs, identified enrichment of biological processes and molecular functions association with ‘chondrocyte differentiation’ ‘positive regulators of cartilage development’ and as expected, ‘bone mineralization’. Therefore suggesting that calcification of AVICs involves an osteo/chondrogenic differentiation program mediated by chondrogenic mRNAs including Bmp2, Sox9, Runx3 and Pth1r, and positive bone regulators (Comp, Ecm1, Ibsp) in tandem with bone inhibitors (Sost). In contrast, AVSMCs do not appear to undergo this intermediate osteo-chondrogenic differentiation step, but express high levels of definitive ossification (bone) markers (Enpp1, Mgp, Spp1 etc.). As a result of data generated from this study, we have identified cell-specific markers of calcification that should be considered when determining osteogenic differentiation in models of cardiovascular calcification.

The diverse molecular profiles of ‘ossification/bone mineralization’-related mRNAs between AVSMCs and AVICs suggest that different mechanisms could be in involved, potentially due to factors discussed above (developmental origin, phenotypes, in vivo milieu). In parallel with AVIC-specific osteo/chondrogenic differentiation, we also note enrichment of DEGs associated with GO terms related to ‘actin’ (Supplementary Table 3) in this cell type, independent of calcific stimulus. This includes downregulation of mRNAs (Acta2, Cnn1, Mhy11, Tagln) associated with ‘actin cytoskeleton’, ‘actin binding’ and ‘myofibril’ GO term-genes (Figures 6A, B). Based on these observations, it is speculated that AVICs under remodeling of the actin cytoskeleton during the osteo/chondrogenic differentiation program and formation of calcific nodules. Interestingly, this was not seen in AVSMCs which is somewhat surprising given previous reports demonstrating decreased expression of contractility proteins during osteogenic differentiation.[51] Furthermore, other studies have shown that Acta2 increases in in vitro models of AVICs calcification, although in this context, Acta2 is used as a marker of VIC activation (fibroblast-to-myofibroblast) (reviewed [52]) following culture on stiff substrate surfaces.[42] A point of consideration to explain the observed decrease in Acta2 mRNA in stimulated VICs in Figures 6A, B, is that upon closer examination, decreased levels of the Acta2 protein are not global to all VICs but limited to VICs within nodules (Figures 6E, H) undergoing calcification (Figures 1, 2), which may reflect differences in mRNA and protein expression levels, and highlight a role for localized actin remodeling in these cells.

The molecular mechanisms underlying cell-specific trajectories towards calcification are not clear, and KEGG and Reactome analysis of DEGs did not identify significant signaling pathways enriched in either cell types. However, consistent with the literature, MAPK-related pathways were commonly represented, which likely reflect its established role in cell proliferation and osteogenic differentiation of these cell types.[53–62] However, pathway analysis and did reveal an enrichment for DEGs associated with PI3K/AKT signaling in both AVSMCs and AVICs that was specific to OM treatment. This pathway has previously been reported in vascular and valve calcification, but implicated to play both positive [63–65] and negative [25, 66–68] roles. Validation studies in Figure 7 suggest that active P13K/AKT signaling is required for OM-, or phosphate-independent calcification. Although it remains undetermined if this signaling pathways regulates OM-specific ossification markers including Lrp4, Gpm6b, Col13a1 and Spp1, although interactions have been linked to Lrp4 and Spp1 in cancer cell types.[69, 70]

This study has generated descriptive, but highly informative transcriptomic information that highlights diversities between these VSMCs and VICs in terms of their calcification molecular profiles in response to high phosphate, versus osteogenic media conditions and identifies possible mechanistic underpinnings to further support the notion that cardiovascular calcifications are multifaceted disorders, occurring in different milieu with different pathologies. The discrepancies raised from this study align with clinical observations suggesting that despite common risk factors, the pathological mechanisms of abnormal calcium deposition in vascular and valvular structures may be in part, be independent. This becomes very important when considering personalized treatment plans for patients that present with CAVD in the absence of CAD, or those that display unparalleled disease progression at both anatomical sites. Limitations of our work include the simplistic nature of the in vitro models used (absence of biomechanics, endothelial cells, etc.), and unfortunately the field currently lacks suitable experimental models that mimic human cardiovascular calcification pathologies to test the hypotheses generated. Nonetheless, findings touch upon the underlying clinical conditions of each of the biological calcification phenotypes and identifies opportunities for the development of cardiovascular calcification-specific therapeutic targets.

Supplementary Material

1

Supplementary Figure 1. Gene Ontology (GO) pathway enrichment analysis of AVSMCs and AVICs treated with 2.5mM or OM, vs. CM for 3 days. Analysis shows enriched biological processes (A-D), molecular function (E-H) and cellular compartments (I-L) on all DEGs that passed the threshold criteria (fold change >2, and corrected p-value<0.05) for each pairwise comparison groups. Bubble plots represent the top 20 GO terms.

Supplementary Figure 2. Gene Ontology (GO) pathway enrichment analysis of AVSMCs and AVICs treated with 2.5mM or OM, vs. CM for 5 days. Analysis shows enriched biological processes (A-D), molecular function (E-H) and cellular compartments (I-L) on all DEGs that passed the threshold criteria (fold change >2, and corrected p-value<0.05) for each pairwise comparison groups. Bubble plots represent the top 20 GO terms.

Supplementary Figure 3. Taqman high-throughput qPCR validation of 68 mRNAs uniquely expressed in treated VSMCs and VICs for 5 days based on bilk RNA-seq findings, as indicated. n=4 treated AVSMCs, n=3 treated VICs. p<0.05 based on unpaired parametric t-test. mRNAs highlighted by a red box indicate those that supported bulk RNA-seq findings.

Highlights:

• Aortic vascular smooth muscle cells (AVSMCs) are more susceptible to calcific stimuli (osteogenic media, high (2.5mM) phosphate) than aortic valve interstitial cells (AVICs)

• Transcriptomic analysis demonstrates temporal molecular diversity between AVSMCs and AVICs, and calcific stimuli

• In response to calcific stimuli, AVICs, unlike AVSMCs, downregulate actin stress fiber makers

• In response to osteogenic media but not high phosphate conditions, PI3K/AKT signaling is increased in both cell types.

Abbreviations.

AVIC

aortic valve interstitial cell

AVSMC

aortic vascular smooth muscle cell

CKD

chronic kidney disease

CM

control media

OM

osteogenic media

Ph

phosphate

VIC

valve interstitial cell

VSMC

vascular smooth muscle cell