Focal Adhesion Kinase Promotes Calcification of Vascular Smooth Muscle Cells via Regulation of Histone Deacetylase 4 and 5.
Oct 30, 2025DOI: 10.1161/atvbaha.123.319010Journal ArticleAHA
Written by: Wenjie TianKuldeep SinghLova P KajuluriJing LuSujin LeeKangsan RohChristopher J NicholsonWanlin JiangHanna J BarnesTadatoshi SatoSophie BoerboomKatrina OstromRebecca H LiClaire BirchenoughElizabeth MooreHelena TattersfieldHaakon H SigurslidMohammed S MahamdehJose V MosqueraChani J HodonskyAdam L JohnsonElizabeth L ChouSagar NigwekarMark E LindsayDonald B BlochChristian L Lino CardenasFumito IchinoseClint L MillerMarc N WeinRajeev Malhotra
Vascular calcification is highly prevalent among patients with diabetes, hypertension, and chronic kidney disease and serves as a strong and independent predictor of cardiovascular events.1,2 The pathologic accumulation of calcium phosphate is similar to bone formation and is an active process driven by the osteogenic phenotype switch of vascular smooth muscle cells (VSMCs), involving various genetic and molecular signaling regulators such as runt-related transcription factor 2 (RUNX2) and alkaline phosphatase (ALPL).3–5 RUNX2 serves as an important transcriptional regulator of the phenotypic switch and calcification of VSMCs.Epigenetic regulation such as DNA methylation and histone modification has been found to play an important role in the phenotypic transformation of VSMCs and the development of vascular calcification.6,7 Histone deacetylase 9 (HDAC9), a member of the class IIa HDAC family, acts as a regulator to promote osteogenic switch and calcification of vascular tissue. In a murine model of medial arterial calcification (matrix Gla protein deficiency), the lack of HDAC9 resulted in a 40% reduction in aortic calcification and improved survival.7 In previous studies related to osteocyte biology, two other members of the class IIa HDAC family, histone deacetylase 4 (HDAC4) and histone deacetylase 5 (HDAC5), were found to regulate bone formation by inhibiting the expression of sclerostin (SOST), an inhibitor of Wingless-type MMTV integration site family (Wnt) signaling.8 These studies suggest that the class IIa HDAC family is important for calcification regulation. Focal adhesion kinase (FAK), an integrin-associated protein tyrosine kinase, phosphorylates HDAC4 and HDAC5 at specific tyrosine residues, thus favoring the cytosolic localization of these HDACs in osteocytes. Inhibition of FAK results in the translocation of HDAC5 from the cytosol into the nucleus and therefore regulates bone formation.8Several studies have identified that FAK activity is associated with vascular diseases, including restenosis, atherosclerosis, pulmonary arterial hypertension, and aneurysm.9–11 Abnormal FAK activation may promote phenotypic switch of VSMCs, and inhibition of FAK catalytic activity attenuates VSMC proliferation and neointimal hyperplasia in animal models of vessel injury.9 One study demonstrated FAK-mediated osteogenic differentiation via Wnt and mitogen activated protein kinase (MAPK) signaling pathways in synovial mesenchymal stem cells. FAK knockdown effectively inhibited bone formation in vivo.12 Based on these findings, we hypothesized that FAK has an important role in regulating the calcification of VSMCs.
Given the effect of FAK and Class IIa HDACs on vascular function, we aimed to test the hypothesis that FAK regulates the calcification of VSMCs, to determine the effect of HDAC4 and HDAC5 on calcification, and to investigate if the role of FAK on VSMC calcification is HDAC4 and HDAC5-dependent.
METHODS
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Human aortic smooth muscle cells (HASMCs) were obtained from Cell Applications (#355–75a) and cultured in smooth muscle cell medium (Cell Applications, #311–500). To induce calcification in HASMCs, cells were treated with SMC medium (Cell Applications, #311F-500) supplemented with 10% fetal bovine serum, 10 mM β-glycerophosphate disodium, 50 μg/mL L-ascorbic acid, and 10 nM dexamethasone, as described previously7. To assess the effect of FAK inhibitors (VS-6063 #S7654, VS-4718 #S7653, Selleckchem), HDAC4/5 inhibitors (LMK-235 #S7569, Selleckchem), nucleus export inhibitor (Leptomycin B #9626, Cell Signaling Technology) on calcification, cells were treated with either drugs or vehicle control.
Total RNA from cells was extracted with TRIzol (Life Technologies) following the manufacturer’s instructions. For quantitative reverse transcription polymerase chain reaction (qRT-PCR), complementary DNA (cDNA) was prepared with 500ng of RNA using the High-Capacity cDNA Reverse Transcription Kit (#4374966, Applied Biosystems) and analyzed with TaqMan probes and the Fast Advanced Master Mix (#444557, Thermofisher Scientific) in a Real-time PCR System (QuantStudio 3, Applied Biosystems). Expression was calculated using the relative Ct method by normalizing with 18S RNA expression.
Cell lysates were prepared using Radio-immunoprecipitation assay buffer (RIPA) containing protease and phosphatase inhibitors (#78430, Thermofisher Scientific). HASMCs were washed with cold phosphate-buffered saline (PBS) twice, then scraped into RIPA buffer, and transferred into 1.5mL Eppendorf tubes. After vortex for 15 s five times, cells were centrifuged at 14,000 g for 10 min at 4°C and the supernatant was saved as total protein. For subcellular fractionation, cytosolic and nuclear protein were extracted as previously described.8 Protein was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked with 5% fat-free milk or 5% bovine serum albumin (BSA) in tris-buffered saline plus 0.05% Tween-20 (TBST) and incubated with primary antibodies overnight at 4°C. Then, membranes were washed and incubated with horse radish peroxidase (HRP)-coupled secondary antibody and subsequently were detected with enhanced chemiluminescence (ECL) Western Blotting Substrate (#32106, Pierce). The primary antibodies in our study were FAK (1:1000, #3285, Cell Signaling Technology), p-FAK (Y397) (1:1000, #8556, Cell Signaling Technology), HDAC4 (1:1000, #17449, Proteintech), HDAC5 (1:1000, #16166, Proteintech), phospho-HDAC4/HDAC5 (1:1000, #3443, Cell Signaling Technology), RUNX2 (1:1000, #12556, Cell Signaling Technology), ALPL (1:1000, #11187, Proteintech), GAPDH (1:1000, #2118S, Cell Signaling Technology). Tyrosine 642 phosphorylated HDAC5 antibody was previously described.8 For phospho/total protein analyses, equal lysate volumes were run on parallel gels under identical conditions. GAPDH served as a loading control, and densitometric values of phospho and total proteins were normalized to GAPDH before calculating phospho/total ratios.
HASMCs cultured in 6-well or 12-well plates were fixed with 4% paraformaldehyde for 15 min and washed with deionized water twice, then treated with a 1% Alizarin Red solution (pH 4.1–4.3) for 20 minutes, followed by multiple washes with distilled water. For tissue Alizarin Red staining, sections of mouse aortas or human carotid artery were deparaffinized to distilled water and stained with 1% Alizarin Red solution for 10 minutes. After shaking off the excess dye, the slides were dehydrated with acetone, cleared in xylene, and mounted in a synthetic mounting medium. For von Kossa staining, cells were incubated in 5% silver nitrate solution and exposed to a 100 Watt bulb for 1 to 2 hours. Unreacted silver was removed with the treatment of 5% sodium thiosulfate followed by distilled water wash.
Calcium ion content was measured using a colorimetric calcium assay kit (Abcam#AB102505) according to the manufacturer’s instructions. Briefly, VSMCs were seeded in 6 well plates and cultured in osteogenic media under different conditions for 21 days. VSMCs were washed with calcium-free PBS and decalcified with 0.6N hydrochloric acid (HCl) for 24 hours at 37°C. The calcium levels in the HCl supernatants were determined by the chromogenic complex formed between calcium ions and o-cresolphthalein, measured at 575 nm and proportional to the concentration of calcium ions present. Total protein from each sample was extracted and quantified using the bicinchoninic acid (BCA) method. Calcium content was normalized to the protein content and expressed as μg calcium per mg protein.
Male wild-type mice aged 18 weeks were euthanized with CO2 asphyxiation. Thoracic aortas were dissected from mice. Adventitial tissue was removed with sterilized forceps and aortas were cut into 2–3 mm wide sections. The human carotid artery was obtained from a patient undergoing carotid endarterectomy surgery as part of an IRB-approved protocol with patient informed consent (IRB #2000P001531). The artery was cut with a surgical blade into 3mm×3mm sections. Aortic rings or human artery sections were cultured in common or osteogenic SMC medium in the presence or absence of FAK inhibitor (VS-6063, 2μM or 4μM). Medium and treatment were replenished every 48 h for 21 days. All sections were fixed with 4% paraformaldehyde for 48 h and saved in 70% ethanol for histological sections.
Small interfering RNA (siRNA) directed against HDAC4 (#L-003497–00-0005), HDAC5 (#L-003498–00-0005), FAK (L-003164–00-0005) and control siRNA (#D-001810–01-50) were obtained from Horizon Discovery Biosciences LTD (SMARTpool). HASMCs were transfected with siRNA using Lipofectamine RNAiMAX reagent (#13778150, Thermo Scientific), as described by the manufacturer (Life Technologies). Recombinant adenoviruses expressing human HDAC4 or HDAC5 were obtained from Vector Biolabs (HDAC4: #1435; HDAC5: #210890). HASMCs were transfected with the adenovirus vectors in common VSMC medium for 24 hours and the expression of HDAC4 and HDAC5 was assessed by Western blotting after 72 hours. For Alizarin Red staining analysis after overexpressing HDAC4/5, HASMCs were treated with adenoviral vector HDAC4/5 for 24 hours, then incubated in osteogenic media with drug (VS-6063, Leptomycin B) or vehicle. Alizarin Red staining was performed after 14 days.
HASMCs were seeded in 8-well chamber slides and cultured at 37°C for 2 days. After treatment with FAK inhibitor VS-6063 or vehicle (Dimethyl sulfoxide (DMSO)) for 2 hours, or HDAC4/5 nuclear export inhibitor Leptomycin B (5 nM, 3 hours for HDAC4 and 30 nM, 24 hours for HDAC5) or vehicle, cells were fixed with 4% paraformaldehyde for 10 minutes. After washing three times with Dulbecco’s phosphate buffered saline (DPBS), cells were treated with 0.1% Triton X-100 for 10min and blocked with 1.5% BSA in DPBS for 1 hour. Then, cells were incubated with primary antibodies (1:50) in 1.5% BSA at 4°C overnight. After washing three times with DPBS, secondary antibodies conjugated with Alexa 488/568 (1:200, #A21202, #A10042, Thermofisher Scientific) were incubated at room temperature for 1 hour. 4’,6-diamidino-2-phenylindole (DAPI) staining (#H1200, Vectashied) was then performed before mounting the coverslip (#1.5) with Fluoromount-GR (SouthernBiotech). The primary antibodies were mouse anti-HDAC5 (C-11) (#sc-133225, Santa Cruz Biotechnology) and rabbit anti-HDAC4 (#ab12172, Abcam). High-resolution images of VSMCs were acquired on a Leica SP8 point scanner confocal and a Nikon W1-Yokogawa spinning disc confocal using a 60x/1.4 objective. Next, the images (8-bit from Leica, 16-bit from Nikon) were segmented using the Cellpose deep learning cellular segmentation algorithm implemented in the open source image processing software Fiji-ImageJ using the TrackMate plugin. Cytoplasmic and nuclear segmentation was performed using the built-in Cytoplasm.2 and Nucleus models, respectively. Mean intensities of the cytoplasm and nuclei of respective groups were plotted, and statistical analyses were carried out using GraphPad Prism software.
3-well silicone inserts were used for migration assays (Ibidi, # 80369). Inserts were fixed on the surface of a 12-well plate. 70 μL of cell suspension with a density of 5×105/mL was put into each well of the insert and the cells were incubated at 37°C and 5% CO2 for 24 hours to form a confluent cell layer. Then, the inserts were removed with tweezers, thus establishing blank gaps between cell layers. After aspirating the medium, cells were incubated in reduced serum medium (Opti-MEM, Gibco, #31985–070) for 12 hours. Cells were then grown in common medium, or osteogenic medium with different concentrations of VS-6063 or LMK-235 (0, 50nM, 200nM, 500nM,1μM, 2μM, n=6 for each group) for 12 hours. The cells were fixed with 4% paraformaldehyde for 10 minutes and stained with Giemsa solution (Gibco, #10092–013) for 15 minutes followed by multiple washes with distilled water. Cell migration was assessed by comparing the area of migrated cells in the gap among different treatments using Image J.
GraphPad Prism (Version 10.2.2, GraphPad Software Inc.) was used for data analysis of in vitro and ex vivo experiments. Normality was tested using the Shapiro-Wilk test. For variables that were non-normal, the Mann Whitney test or Kruskal Wallis test were used, as appropriate. Non-normal data are presented as median (IQR). For normally distributed variables with two groups, the F test for equal variances was performed. If the variances were equal, then the unpaired t test was used while Welch’s t test was used when variances were unequal. For normally distributed variables with more than two groups, the Bartlett’s test for equal variances was performed and, if the variances were equal, then the ordinary 1-way ANOVA with Sidak’s post-hoc test was used. If the variances were unequal, then the Brown Forsythe test was used with Dunnett’s post-hoc testing. Normally distributed data are presented as mean ± SEM. All statistical tests were two-tailed. An adjusted p-value less than 0.05 was considered statistically significant.
Phenome-wide association study (PheWAS) results were obtained from summary statistics available in the genome-wide association study (GWAS) Atlas (https://atlas.ctglab.nl/) by querying protein tyrosine kinase-2 (PTK2), HDAC4 and HDAC5. All results with nominal p-value<0.05 were included, and top associations were labelled by specific study phenotype. PheWAS was also performed using gene-based tests (SKAT-O using SKAT R package v2.2.5)13 of predicted loss-of-function and missense variants associated with human phenotypes in the UK Biobank whole exome sequencing dataset,14 also available through the GeneBass web resource (https://app.genebass.org).14 Gene prioritization of top associations for PTK2, HDAC4, and HDAC5 loci was performed using the OpenTargets locus-to-gene pipeline (https://opentargets.org).15 Locus-to-gene (L2G) prioritization scores for the associated single-nucleotide polymorphisms (SNPs) were derived from variant pathogenicity scores, distance to gene transcription start site (TSS), evidence of quantitative trait locus (QTL) colocalization with GWAS signal, and evidence of chromatin interaction from high-throughput chromosome confirmation capture (Hi-C) datasets.
Ischemic and non-ischemic explanted human coronary artery tissue biospecimens were obtained at Stanford University from diseased heart transplant donors consenting for research studies as described.16 Hearts were arrested in cardioplegic solution and transported on ice prior to dissecting proximal coronary artery segments from main branches of left anterior descending, circumflex or right coronary arteries. Epicardial and perivascular adipose was trimmed on ice, rinsed in cold phosphate buffered saline, and rapidly frozen in liquid nitrogen, and stored at −80C. Normal human coronary artery tissue biospecimens were also obtained at Stanford University from non-diseased donor hearts rejected for orthotopic heart transplantation processed following the same protocol as hearts for transplant. Tissues were de-identified and clinical and histopathology information was used to classify ischemic, non-ischemic hearts and lesion and non-lesion containing arteries. All normal arteries originated from hearts with left ventricular ejection fraction (LVEF) greater than 50%. Frozen tissues were transferred to the University of Virginia through a material transfer agreement and Institutional Review Board approved protocols.
Total RNA was extracted from frozen coronary artery segments using the Qiagen miRNeasy Mini RNA Extraction kit (catalog #217004). Approximately 50 mg of frozen tissue was pulverized using a mortar and pestle under liquid nitrogen. Tissue powder was then further homogenized in Qiazol lysis buffer using stainless steel beads in a Bullet Blender (Next Advance) homogenizer, followed by column-based purification. RNA concentration was determined using Qubit 3.0 and RNA quality was determined using Agilent 4200 TapeStation. Samples with RNA Integrity Number (RIN) greater than 5.5 and Illumina DV200 values greater than 75 were including for library construction. Total RNA libraries were constructed using the Illumina TruSeq Stranded Total RNA Gold kit (catalog #20020599) and barcoded using Illumina TruSeq RNA unique dual indexes (catalog # 20022371). After re-evaluating library quality using TapeStation, individually barcoded libraries were sent to Novogene for next generation sequencing. After passing additional QC, libraries were multiplexed and subjected to paired end 150 bp read sequencing on an Illumina NovaSeq S4 Flowcell to a median depth of 100 million total reads (>30 G) per library.
The raw passed filter sequencing reads obtained from Novogene were demultiplexed using the bcl2fastq script. The quality of the reads was assessed using FASTQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and the adapter sequences were trimmed using trimgalore. Trimmed reads were aligned to the hg38 human reference genome using STAR v2.7.3a17 according to the GATK Best Practices for RNA-seq. To increase mapping efficiency and sensitivity, novel splice junctions discovered in a first alignment pass with high stringency, were used as annotation in a second pass to permit lower stringency alignment and therefore increase sensitivity. PCR duplicates were marked using Picard (http://broadinstitute.github.io/picard/) and WASP18 was used to filter reads prone to mapping bias. Total read counts and RPKM were calculated with RNA-SeQC v1.1.819 using default parameters and additional flags “-n 1000 -noDoC -strictMode” and GENCODE v30 reference annotation. The transcript and isoform expression levels were estimated using the RNA-seq by expectation maximization (RSEM) package20.
Differential gene expression of coronary artery RNA-seq data was performed using DESeq2, either comparing ischemic vs normal disease classification or lesion vs non-lesion sample histopathology classification. All normal samples were from donor hearts with no evidence of atherosclerosis. Filtered raw read counts and metadata were provided and age, sex and RNA RIN values were used as covariates. Significantly differentially expressed genes were determined at Benjamin-Hochberg FDR p-value <0.05 and fold change greater than ±50%.
Raw count matrices from each library across the 4 studies were downloaded from GEO and Zenodo. The 22 sequencing libraries were processed in standardized manner using Seurat v4 and first pass clustering with SCTransform normalization without removing low-quality cells. Doublets were removed using scDblFinder and ambient RNA removed using DecontX. The decontaminated raw count matrices were loaded into each Seurat object and the following quality filters were used to retain cells that had 1) >= 200 and <= 4000 uniquely expressed genes 2) >= 200 and <= 20000 UMIs 3) <= 10% of reads mapped to the mitochondrial genome. To avoid clustering results confounded by cell cycle state, cell cycle variance was regressed out during SCTransform normalization. We then carried out dimensionality reduction of the normalized count matrix using PCA. The first 30 principal components (PCs) were used as input for clustering in Seurat, which relies on a Shared-Nearest-Neighbors (SNN) and Louvain community detection approach. We then applied Uniform Manifold Approximation and Projection (UMAP) non-linear dimensionality reduction using the first 30 PCs. UMAP embeddings were used for visualization of Louvain clustering results. Processed matrices were then stored as seurat objects for batch-correction.
We used reciprocal PCA (rPCA) method in Seurat to harmonize the processing sequencing libraries based on several benchmarking metrics, including running time, clustering purity and biological conservation scores. Upon integration we performed dimensionality reduction of the data using PCA and a shared nearest neighbors (SNN) graph was constructed using 50 nearest neighbors and the first PCs as input. Gene markers for each cluster were identified in Seurat v4 using PrepSCTMarkers() and FindAllMarkers() functions. Differentially expressed genes by cluster were considered for genes expressed in >25% of the clusters being compared (one versus all) at log Fold change =0.25 and B-H adjusted p-value < 0.05. Clusters were annotated using a combination of manual curation and comparison with lineage training datasets, as well as automated label transfer from the SCTransform processed Tabula Sapiens vasculature reference. Subclustering analysis was performed by subsetting the integrated object SMCs, and then identifying the variable features, repeating the dimensionality reduction and embedding, SCTransform normalization and then repeating the PCA and UMAP embedding. Feature plots were generated for candidate genes HDAC4, HDAC5, and PTK2. A link to fully documented scripts used for processing of single cell datasets can be found here: https://github.com/MillerLab-CPHG/scRNAutils
RESULTS
We evaluated the effect of pharmacologic inhibition or silencing of FAK on the calcification of HASMCs. HASMCs were treated with common medium (regular human VSMC growth medium) or osteogenic medium in the presence or absence of the FAK inhibitor VS-6063. Biologically, FAK undergoes autophosphorylation at tyrosine 397, which is used as a marker of FAK cellular activity.21 As anticipated, treatment with VS-6063 for 3 days decreased Y397 autophosphorylation in HASMCs without changing total FAK levels (Figure 1A). Compared to HASMCs incubated in common medium, cells cultured in osteogenic medium for 3 days exhibited a > 5-fold increase in mRNA levels of RUNX2 and ALPL. However, treatment with VS-6063 abrogated the increase in RUNX2 and ALPL mRNA levels (Figure 1B and 1C). Similarly, protein expression of RUNX2 and ALPL was significantly upregulated with culture in osteogenic medium for 6 days, which was blunted by VS-6063 treatment (Figure 1D and Figure S1A and S1B). As shown by Alizarin Red and Von Kossa staining, calcium deposits formed with HASMCs grown in osteogenic medium for 14 days, and VS-6063 treatment prevented this calcification (Figure 1E). To determine whether the effect of FAK inhibition on calcification is dependent on the dose of the drug, we cultured HASMCs in osteogenic medium with two FAK inhibitors (VS-6063, VS-4718) under different concentrations. As shown by Alizarin Red staining and calcium quantification, both VS-6063 and VS-4718 inhibited the calcification of HASMCs in a dose-dependent manner (Figure 1F–1G and Figure S1C and S1D).
Calcification by VSMCs involves a switch from a contractile to an osteogenic and migratory cellular phenotype. Therefore, we performed migration assays to assess the effect of FAK inhibition (VS-6063) on the migration of HASMCs. HASMC were cultured in chambers with intervening cell inserts and cell migration was assessed by measuring the surface area of the migrated cells into a cell-free gap 12 hours after removing the inserts. Compared with cells cultured in common medium, cells grown in osteogenic medium exhibited a significantly increased migration area (44.5% vs. 22.6%, P<0.0001). Across a range of doses, VS-6063 inhibited the migration of cells in osteogenic medium (Figure 1H). These results provide evidence that FAK activity is a critical regulator in the process of vascular smooth muscle cell phenotype switch and calcification.
Silencing of FAK expression with siRNA (siFAK) was performed by targeting PTK2, the gene encoding FAK. FAK gene expression was reduced by 63% and 67%, respectively, in common medium and osteogenic medium treated with siFAK (Figure 2A), and the protein level of total FAK was reduced accordingly (Figure 2B). As evidenced by mRNA and protein levels for RUNX2 and ALPL, as well as Alizarin Red staining and calcium quantification, silencing of FAK exhibited a similar result as pharmacologic FAK inhibition in preventing calcification of HASMCs (Figure 2C–2G and Figure S2A and S2B).
Given the effectiveness of FAK inhibition on preventing the osteogenic phenotype switch and calcification of VSMCs, we next performed RNA-seq analysis to assess the gene expression changes associated with FAK inhibition. RNA of HASMCs was isolated after cells were treated with common or osteogenic medium in the presence or absence of VS-6063 (1 μM) for 3 days. There were 7,211 differentially expressed genes (Q value ≤ 0.05) when comparing osteogenic medium versus common medium conditions, and 3,217 differentially expressed genes when comparing treatment with or without FAK inhibitor under osteogenic conditions (Figure 3A). Pathway analysis revealed that several pathways critical for VSMC phenotype, function, and calcification including receptor for advanced glycation end products (RAGE) signaling, MAPK signaling, protein processing, and autophagy are affected by FAK, as well as genes known to be involved in bone formation (Figure 3B, Table S1a-b). Among the genes dramatically upregulated by osteogenic media and downregulated by FAK inhibitor (differences were determined using a Q value threshold of <1.0×10−10, Table S2), 28 genes have been reported to be associated with calcification and/or atherosclerosis (Figure 3C), including Runx2 and ALPL. These RNA-seq findings underscore the importance of FAK in regulating global gene expression critical for VSMC phenotype, function, and calcification.
Both HDAC4 (Ser246) and HDAC5 (Ser259) undergo phosphorylation at conserved serine residues via salt-inducible kinases leading to their cytoplasmic retention.22 FAK has recently been shown to phosphorylate HDAC5 (at tyrosine residue 642) and HDAC4 (at unclear tyrosine residue) in osteocytes thereby altering their cellular localization.8 HDAC5 tyrosine and serine phosphorylation are highly correlated through unclear mechanisms. We therefore sought to explore whether FAK regulates VSMC calcification, at least in part, via regulation of HDAC4/5. Although a Y642-HDAC5 specific antibody is available, no similar antibody for tyrosine-phosphorylated HDAC4 exists. Since tyrosine phosphorylation correlates with serine phosphorylation for HDAC5 and HDAC4 and HDAC5 are similar in sequence and function, here we monitored HDAC4 phosphorylation status using an antibody that recognizes HDAC4 phospho-S246. First, we treated HASMCs with VS-6063 (10μM, 2h) and examined the expression of phosphorylated HDAC4 and HDAC5 relative to total levels. As we hypothesized, inhibition of FAK activity with VS-6063 reduced phosphorylation of both HDAC4 and HDAC5 without altering total HDAC4/5 levels (Figure 4A). Similar results were found with the knockdown of FAK with siRNA (Figure 4B). In response to FAK inhibition, HDAC4 and HDAC5 showed altered cellular localization with a preference for the nucleus, as evidenced by immunofluorescence (Figure 4C) and by Western blot (Figure S3).
We further explored whether inhibition of HDAC4/5 nuclear export can attenuate calcification. HASMCs were cultured in common medium or osteogenic medium in the presence or absence of Leptomycin B, a nuclear export inhibitor (10 nM)23 for 3 hours for immunofluorescence staining, and 6 days (5 nM) for assessment by Western blot. Immunofluorescence staining (Figure 5A-B) showed that leptomycin B effectively increased the nuclear localization of HDAC4/5. Furthermore, leptomycin B reduced the phosphorylation of HDAC4 and HDAC5 in both normal media and osteogenic media (Figure 5C). The increased expression of RUNX2 and ALPL induced by osteogenic medium was attenuated by leptomycin B (Figure 5C and Figure S4).
To test the effect of overexpression of HDAC4 and HDAC5 on VSMC calcification, HASMCs were treated with adenoviral vectors expressing either HDAC4 or HDAC5 in osteogenic media. Western blot analysis showed increased expression of HDAC4 and HDAC5 along with their phosphorylated states in adenovirus-treated cells (Figure 6A and Figure S5A-B). Compared to control, adenoviral expression of HDAC4 or HDAC5 resulted in increased calcification of HASMCs which was prevented by treatment with leptomycin B (5 nM for 12 hours every two days) as evidenced by both Alizarin Red staining (Figure 6B) and calcium quantification (Figure 6C-D). Further, on immunofluorescence analysis, both cytosolic and nuclear levels of HDAC4 and HDAC5 were increased after adenoviral overexpression. As anticipated, leptomycin B treatment significantly decreased HDAC4/5 cytosolic localization and increased their nuclear localization (Figure 6E-F). These results suggest that, while cytoplasmic HDAC4 and HDAC5 promote VSMC calcification, retention of HDAC4 or HDAC5 in the nucleus does not promote calcification.
Thus far, we have demonstrated an essential role for FAK in VSMC calcification. FAK inhibition reduces osteogenic phenotype switch and calcification and is associated with reduced phosphorylation and increased nuclear retention of HDAC4 and HDAC5. We next investigated the role of HDAC4 and HDAC5 on VSMC calcification. HASMCs were transfected with siRNA specific for either HDAC4 (siHDAC4) or HDAC5 (siHDAC5) for 24 hours and subsequently incubated with common medium or osteogenic medium for 3 days for qPCR, 6 days for Western blot, or with osteogenic medium for 14 days for calcification quantification with Alizarin Red staining. During the 14-day experiments, HASMCs were transfected with siHDAC4 or siHDAC5 once more on the 6th day to ensure consistent low expression of HDAC4/5. Compared to HASMCs grown with control siRNA (siC), HASMCs treated with siHDAC4 or siHDAC5 exhibited >65% knockdown of HDAC4 and >50% knockdown of HDAC5 mRNA level, respectively (Figure 7A and 7C), with concordant reductions in protein level (Figure 7B–7C and Figure S6A, S7A). Both siHDAC4 and siHDAC5 reduced the protein expression of RUNX2 and ALPL that was induced by osteogenic media (Figure 7B, 7D and Figure S6B-S6C and S7B-S7C). As shown by Alizarin Red staining, treatment with either siHDAC4, siHDAC5, or the combination of siHDAC4/siHDAC5 ameliorated calcium deposition induced by osteogenic medium (Figure 7E). To determine whether HDAC4 and HDAC5 inhibitors can prevent calcification of VSMCs, we treated HASMCs with LMK-235, a specific HDAC4/5 inhibitor.24,25 After growing HASMCs in osteogenic medium for 14 days, Alizarin Red staining showed that LMK235 prevented calcium deposition of HASMCs induced by osteogenic medium in a dose-dependent manner (Figure 7F). Similar to the FAK inhibitor, LMK-235 reduced the migration of VSMCs induced by osteogenic medium (Figure 7G). These results, including the adenoviral HDAC4 or HDAC5 data, suggest that cytosolic HDAC4 and HDAC5 promote VSMC calcification, whereas silencing, inhibiting or driving nuclear retention of HDAC4 and HDAC5 prevents the calcification process.
We next investigated whether FAK inhibition could prevent calcification induced by overexpressing HDAC4/5. HASMCs were treated with adenoviral vectors expressing either HDAC4 or HDAC5 for 24 hours, then incubated with VS-6063 (1 μM) or vehicle (DMSO) for 14 days. Alizarin Red showed enhanced calcification of HASMCs with Ad-HDAC4 or Ad-HDAC5 treatment as compared to control adenovirus treatment. VS-6063 inhibited the calcification induced by overexpression of HDAC4 or HDAC5 (Figure 8A and 8B). We next determined whether knockdown of FAK with siRNA attenuated the calcification induced by HDAC4 or HDAC5. Similar to the pharmacologic inhibition of FAK, siFAK also inhibited the calcification induced by HDAC4 or HDAC5 overexpression (Figure 8C and 8D). Altogether, these data suggest that targeting FAK can inhibit VSMC calcification, at least in part, via the effects of FAK on the phosphorylation and cellular localization of HDAC4 and HDAC5.
In order to determine the effects of FAK inhibition on aortas ex vivo, we induced calcification of mouse aortic rings with osteogenic media and evaluated the effect of VS-6063 on calcification. Mouse aortas were isolated as described previously,26 and the adventitia of the aorta was removed. Aortic rings were obtained by cutting aortas into 2–3 mm segments and the rings were cultured in common medium or osteogenic medium in the presence or absence of VS-6063. After 21 days, extensive calcification was observed on the walls of aortas in the osteogenic medium but not those in common medium. Treatment with VS-6063 (2μM or 4μM) inhibited the calcification of mouse aortas (Figure 8E). Additionally, we tested the effectiveness of VS-6063 on calcification of the human carotid artery. The carotid artery was obtained from a patient undergoing carotid endarterectomy surgery and the vessel was cut into 3mm×3mm pieces, followed by culturing in common medium or osteogenic medium with VS-6063 or vehicle (DMSO) for 21 days. Alizarin Red staining showed that osteogenic medium induced calcification on the surface of the tissue, and treatment with VS-6063 attenuated the calcification of the human carotid artery (Figure 8F), providing indirect evidence for FAK promoting calcification in human vascular calcification.
To further investigate the human disease relevance of the gene encoding FAK (PTK2) and its downstream effectors, HDAC4 and HDAC5, we performed several queries in large-scale human genotype-phenotype databases such as the UK Biobank.14 Phenome-wide association studies using either GWAS common variants or whole exome sequencing variants identified several significant (Bonferroni corrected P<1E-05 or SKAT-O P-value <0.05) cardiovascular phenotype associations for HDAC4 such as mean arterial pressure and systolic blood pressure and for PTK2 including high blood pressure, acute myocardial infarction and stroke (Table S3 and S4). By leveraging the OpenTargets locus-to-gene prioritization pipeline, which calculates a likelihood score by combining multiple functional annotations,15,16 HDAC4 and PTK2 were prioritized as the causal gene for many of these traits (Table S5). Similarly, we observed associations for lipids and bone mineral density phenotypes for HDAC5, consistent with its role in bone mineralization and calcification (Table S4 and S5).
To investigate the relative expression changes in PTK2, HDAC4, and HDAC5 during vascular disease we queried our cohort of healthy and atherosclerotic coronary arteries with bulk transcriptomic data (see Methods). PTK2 gene expression was downregulated in artery segments containing lesions (Padj = 1.0E-03) (Table S6). No difference in HDAC4 or HDAC5 gene expression was noted, perhaps because these HDACs are regulated post-translationally as our in vitro data showed. To evaluate the expression changes of these genes at the single-cell level we queried an integrated meta-analyzed reference of 4 human coronary and carotid artery single-cell datasets relevant to atherosclerosis (see Methods, unpublished). After subsetting the integrated reference to investigate SMC subtypes, we identified PTK2 and HDAC5 expression that was mostly localized in cells annotated as pericytes, contractile SMCs, as well as transitional-ECM producing SMCs, whereas HDAC4 expression was more modest in the SMC populations (Figure 9). Notably we observed higher expression of PTK2 and HDAC5 in a small fraction of cells annotated as fibromyocytes, suggesting that they could be involved in the ECM remodeling process preceding calcification. Together, these results provide additional human genetic and transcriptomic-based evidence that these genes may play critical roles in regulating vascular disease phenotypes, particularly in phenotypically modulated SMCs.
DISCUSSION
Vascular calcification is characterized by the transdifferentiation of VSMCs from a contractile to an osteogenic phenotype, which is mediated by the expression of bone-related transcription factors including RUNX2, SRY-box transcription factor 9 (Sox9), and MSH homeobox 2 (Msx2),5 and epigenetic regulation including DNA methylation and histone modification.6,7,27,28 In this study, we demonstrated that FAK as well as two histone deacetylases, HDAC4 and HDAC5, positively regulate calcification. We show that in human VSMCs, FAK inhibition results in increased nuclear localization of HDAC4 and HDAC5. HDAC4 and HDAC5 promote calcification when phosphorylated in the cytoplasm whereas inhibition of FAK prevents calcification of VSMCs, at least in part, via regulation of HDAC4 and HDAC5.
HDAC4 and HDAC5 belong to the class IIa HDAC family. Members of the class IIa HDAC family contain an N-terminal domain that binds to nuclear myocyte enhancer factor-2 (MEF2) and contributes to transcriptional repression of MEF2 target genes.29 The class IIa HDACs also recruit class I HDACs and other repressors such as C-terminal binding protein (CTBP), which accounts for a portion of their transcriptional repressive activity in the nucleus.30 However, a mechanism for restricting the repressive activity of the class IIa HDACs is by regulation of their subcellular localization. The class IIa HDACs have binding sites for the cytoplasmic 14–3-3 protein, calmodulin, and salt-inducible kinase (SIK).31 Phosphorylation or dephosphorylation of class IIa HDACs regulates the shuttling of HDAC protein between the cytoplasm and nucleus and plays an important role in regulating cellular behavior and function.32 Abend et al. reported that the activity of SIK facilitates serine phosphorylation and cytoplasmic retention of HDAC4 and promotes calcification of mouse VSMCs.33 In our experiments, we used human VSMCs to demonstrate the important role of HDAC4 in the regulation of calcification. Furthermore, our results showed that HDAC5 also plays an important role similar to HDAC4 in the calcification of human VSMCs, verified by gene silencing, adenoviral overexpression, and pharmacological inhibition. Moreover, we identified that FAK increases cytosolic localization of HDAC4/5 through tyrosine phosphorylation in VSMCs thereby promoting calcification. Increasing the relative nuclear localization of HDAC4/5, either with FAK inhibition or the nuclear export inhibitor leptomycin B, prevented the calcification of VSMCs. These findings suggest that the subcellular localization of HDAC4 and HDAC5 plays a critical regulatory role in vascular calcification. Namely, cytosolic HDAC4 and HDAC5, which are favored in the phosphorylated form, promote the calcification of VSMCs. One potential explanation for our observations is that the class IIa HDACs when phosphorylated and in the cytosol exert reduced inhibition of RUNX2, thus enabling the master transcriptional regulator to promote osteogenic phenotype switch of VSMCs.34 In support of this hypothesis, a recent study demonstrated that bone morphogenetic protein (BMP-2) (a well-established pro-osteogenic stimulator of VSMCs)35 induces serine phosphorylation of HDAC4, thereby reducing its binding with RUNX2.36 However, the explanation appears to be more complex as we found that adenoviral overexpression of HDAC4 and HDAC5 promotes vascular calcification. Although much of this HDAC4/5 was found to be phosphorylated, it is important to note that beyond just preventing their repressive function in the nucleus, the shuttling of HDAC4 and HDAC5 to the cytoplasm exerts other cellular effects. For instance, HDAC4 deacetylates cytosolic proteins including myosin heavy chain, peroxisome proliferator-activated receptor gamma co-activator 1alpha (PGC-1α), and heat shock cognate 71 kDa protein (Hsc70) thereby regulating skeletal muscle cell homeostasis.37 The specific cytosolic effects of HDAC4 and HDAC5 in VSMCs thus warrants further investigation. The proclivity for VSMCs to calcify thus is likely dependent on the balance between the nuclear and cytosolic effects of HDAC4 and HDAC5.
As a protein tyrosine kinase, FAK has important physiological and pathological significance by regulating cellular proliferation, migration, and function.38 FAK can alter the phenotype of vascular smooth muscle cells and promote intimal hyperplasia and atherosclerosis.6,9,39 Multiple studies have shown that FAK is involved in the regulation of bone formation and RUNX2 expression.40,41 Based on these studies on the role of FAK in the regulation of phenotypic transformation of VSMCs and osteogenesis of osteocytes, it can be speculated that FAK may also play an important role in vascular calcification. However, to our knowledge, no prior studies have reported a direct relationship between FAK and the calcification of VSMCs. In this study, we demonstrated that inhibition of FAK activity or expression inhibits the expression of pro-calcification factors such as RUNX2 and ALPL, and as a result, effectively prevents calcification of VSMCs. It is important to note that FAK protein levels are not increased under osteogenic conditions; rather, FAK is permissive of VSMC calcification under osteogenic conditions. This would suggest that RUNX2 expression is increased in osteogenic conditions through other mechanisms (ie, increased phosphate-induced BMP signaling),42 but that a basal level of FAK is necessary to maintain these higher RUNX2 levels and for VSMC calcification to occur.
Although vascular calcification has proven to be a strong predictor of cardiovascular and cerebrovascular events and death, there is currently no effective treatment targeting calcification. Our study showed that HDAC4 and HDAC5 are important regulators of vascular calcification and that FAK promotes vascular calcification by regulating phosphorylation of HDAC4 and HDAC5. FAK and HDAC4/5 may therefore represent novel targets of therapy for vascular calcification.
RM
Dr. Rohit Malhotra
Cardiology · Cardiac ElectrophysiologyUniversity of Virginia Health Systemrohit