PRMT3-Mediated Arginine Methylation Stabilizes PCSK9 to Promote Aortic Valve Calcification

Abstract

BACKGROUND:

Aortic valve calcification increases leaflet stiffness and contributes to the development of calcific aortic valve disease. The molecular and cellular mechanisms underlying calcification remain unclear. Here, we aimed to investigate the role of PRMT3 (protein arginine methyltransferase 3) in valvular calcification and calcific aortic valve disease progression.

METHODS:

Both aortic valve leaflets and valvular interstitial cells from patients were used to evaluate the expression pattern and investigate the underlying mechanism of PRMT3 in calcific aortic valve disease pathogenesis. High-cholesterol diet–fed Apoe (apolipoprotein E)–deficient (ApoE^−/−) mice with Prmt3 haploinsufficiency were used to examine the role of PRMT3 in aortic valve calcification. The effects of PRMT3 inhibition (SGC707) and degradation (PROTAC compound 11) on aortic valve calcification were investigated in vivo. To elucidate the mechanisms underlying the procalcific effects of PRMT3, we performed immunoprecipitation coupled with liquid chromatography–tandem mass spectrometry and coimmunoprecipitation assays with an enzymatically inactive PRMT3 variant as well as arginine-to-lysine and lysine-to-alanine substitution PCSK9 variants.

RESULTS:

We found that PRMT3 expression was significantly upregulated during aortic valve calcification. RUNX2 (runt-related transcription factor 2) recruited P300 to promote PRMT3 expression through histone H3 lysine 27 acetylation. Moreover, Prmt3 haploinsufficiency markedly ameliorated aortic valve calcification in high-cholesterol diet–fed ApoE^−/− mice, as revealed by reduced thickness and calcium deposition in the aortic valve leaflets, improved echocardiographic measures (decreased peak transvalvular jet velocity, reduced mean transvalvular pressure gradient, and increased aortic valve area), and decreased levels of osteogenic markers OPN (osteopontin) and Osx (Osterix) in the aortic valve leaflets. In addition, anticalcific effects were achieved through pharmacologic PRMT3 inhibition with SGC707 and PRMT3 degradation through PROTAC. Consistent with the in vivo results, we found that PRMT3 promoted the osteogenic differentiation of human valvular interstitial cells through its enzymatic activity. Mechanistically, we revealed that PRMT3 catalyzed the asymmetric dimethylation of PCSK9 at the arginine 582 residue (R582), which was accompanied by a prolonged PCSK9 half-life. This methylation prevented the binding of the E3 ligase CHIP (carboxyl terminus of Hsc70-interacting protein) to PCSK9 at lysine 575 (K575), thereby abrogating ubiquitination-mediated degradation of PCSK9, which is a procalcific factor, and accelerating aortic valve calcification progression.

CONCLUSIONS:

We identify a previously unrecognized posttranslational mechanism regulating PCSK9 stability in valve interstitial cells during calcific aortic valve disease and establish a link between PRMT3-mediated arginine methylation and valve-specific lipid–osteogenic coupling.

Clinical Perspective.

What Is New?

• We identify a previously unrecognized posttranslational mechanism regulating PCSK9 (proprotein convertase subtilisin/kexin type 9) stability in valve interstitial cells in calcific aortic valve disease and establish a link between PRMT3 (protein arginine methyltransferase 3)–mediated arginine methylation and valve-specific lipid–osteogenic coupling.

• We prove PRMT3 as an amplifier of calcific aortic valve disease and demonstrate that PRMT3 expression is significantly elevated in human and murine calcified aortic valves.

• We reveal that PRMT3 catalyzes asymmetric dimethylation of PCSK9 at R582, which prevents CHIP (carboxyl terminus of Hsc70-interacting protein)–mediated ubiquitination at K575, thereby stabilizing PCSK9.

• We demonstrate that pharmacologic PRMT3 inhibition effectively attenuates aortic valve calcification, highlighting PRMT3 as a druggable target.

What Are the Clinical Implications?

• PRMT3 represents a novel therapeutic target for slowing or preventing progression of calcific aortic valve disease, a disease currently lacking effective medical therapy.

• Selective PRMT3 inhibition—including targeted protein degradation strategies—may provide a noninvasive therapeutic alternative for patients with early-stage calcific aortic valve disease, potentially delaying or reducing the need for aortic valve replacement.

Calcific aortic valve disease (CAVD) is a highly prevalent heart valvular disorder characterized by progressive leaflet thickening, fibrosis, calcification, and aortic stenosis (AS).¹ CAVD affects 1% to 2% of adults older than 65 years and ≈12% of adults >75 years of age.² Symptomatic patients with severe AS have a mortality rate of up to 50% after 1 year of diagnosis.² To date, effective pharmacologic therapies are lacking, primarily owing to limited insights into the molecular mechanisms underlying disease progression.³ Surgical or transcatheter aortic valve replacement remains the only treatment option; this invasive approach is often associated with complications and suboptimal long-term outcomes.^1,4 Therefore, alternative pharmacologic strategies are urgently needed. Emerging evidence suggests that the osteogenic differentiation of human aortic valve interstitial cells (hVICs) is a central event in the pathogenesis of aortic valve calcification.⁴ Targeting this phenotypic switch represents a promising therapeutic approach to slow or halt CAVD progression.

PCSK9 (proprotein convertase subtilisin/kexin type 9) is a well-characterized regulator of lipid metabolism that promotes the degradation of low-density lipoprotein (LDL) receptors in hepatocytes and other cell types, thereby reducing LDL clearance.⁵ In addition to its classic role in cholesterol homeostasis, accumulating evidence suggests that PCSK9 may play a procalcific role in CAVD development.^6,7 Consistent with previous reports, our recent study showed that PCSK9 is the downstream effector and interacting partner of CD36 and mediates the procalcific effects of a novel Piwi-interacting RNA (AVCAPIR [aortic valve calcification-associated PIWI-interacting RNA]).⁸ Despite the vital role of PCSK9 in CAVD pathogenesis, the molecular mechanisms governing PCSK9 expression remain largely unexplored. Posttranslational modifications (PTMs) play pivotal roles in various cardiovascular diseases.^9–11 PCSK9 undergoes several PTMs, including N-glycosylation,¹² Tyr-sulfation,¹³ Ser-phosphorylation,¹⁴ and palmitoylation.¹⁵ Elucidating the PTM-mediated regulation of PCSK9 may offer new strategies for treating CAVD.

Protein arginine methylation, which is one of the most prevalent PTMs, affects ≈0.5% of the arginine residues in the human proteome. This modification plays essential roles in various cellular processes, including RNA metabolism, signal transduction, and epigenetic regulation of gene expression.¹⁶ PRMTs (protein arginine methyltransferases) are enzymes that catalyze the transfer of methyl groups from S-adenosyl-L-methionine to the guanidine nitrogen of arginine residues,¹⁷ and can be divided into 3 types on the basis of the methylation they catalyze. Type 1 enzymes (PRMT1–PRMT4, PRMT6, and PRMT8) catalyze monomethylation (MMA) and asymmetric dimethylation (ADMA). Type 2 enzymes (PRMT5 and PRMT9) catalyze the formation of MMA and symmetric dimethylation (SDMA). The type 3 enzyme (PRMT7) catalyzes only the formation of MMA.¹⁷ The role of PRMTs in CAVD remains unexplored. In recent years, a number of studies addressing the key roles of PRMT3 in a diverse array of human cancers have been published.¹⁸ PRMT3 has also been recognized as a key contributor to atherosclerosis,^19,20 which has a pathogenesis similar to that of CAVD. More recently, PRMT3 was shown to methylate HIF-1α (hypoxia-inducible factor 1-α) to increase the vascular calcification induced by chronic kidney disease.²¹ A previous report also revealed a potential role of PRMT3 in the osteoblastic differentiation of human mesenchymal stem cells.²² However, the role of PRMT3 in osteogenic reprogramming during CAVD development remains largely undefined.

In this study, we aimed to identify the functional PRMTs that are implicated in CAVD pathogenesis and delineate their roles in aortic valve calcification. The evidence we provide indicates that PRMT3 promotes aortic valve calcification in vitro and in vivo. Mechanistically, PRMT3 stabilizes PCSK9 by catalyzing asymmetric dimethylation at arginine 582 (R582), which in turn attenuates CHIP (carboxyl terminus of Hsc70-interacting protein)–mediated ubiquitination at lysine 575 (K575) and degradation. Our findings reveal a previously unrecognized posttranslational mechanism regulating PCSK9 stability in valvular interstitial cells during CAVD and establish a link between PRMT3-mediated arginine methylation and valve-specific lipid–osteogenic coupling.

METHODS

Data Availability

The data supporting the findings of this study are available within the article and Supplemental Material. A detailed description of the materials and methods is provided in the Methods in the Supplemental Material. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Human Studies and Ethics

Human calcified aortic valve leaflets were obtained from patients with CAVD during aortic valve replacement. Control noncalcified aortic valve leaflets were collected from the explanted hearts of patients who underwent heart transplantation for dilated cardiomyopathy. The exclusion criteria included rheumatic aortic valvopathy, infective endocarditis, bicuspid aortic valves, congenital valve disease, or valves with moderate to severe aortic valve regurgitation. The protocol for this study complied with the Declaration of Helsinki and was approved by the Tongji Medical College Institutional Review Board, Huazhong University of Science and Technology (XWKYZMY-035). Written informed consent was obtained from each participant.

Animal Studies and Ethics

All the animal experiments were approved by the Animal Care and Use Committee of Tongji Medical College and complied with the European Communities Council Directive 86/609/EEC and 2010/63/EU for the protection of animals used for experimental purposes. Apoe (apolipoprotein E)–deficient mice (ApoE^−/−) were acquired from Beijing Vital River Laboratory Animal Technology Co, Ltd. The animals were randomly assigned to the experimental groups. The animal randomization procedures were performed using a simple and free online randomization tool designed in GraphPad (http://www.graphpad.com/quickcalcs/randomize1.cfm). Allocation concealment with the sequentially numbered method was performed to eliminate selection bias in the intervention assignment. Blinding procedures with respect to the masking of group and treatment assignments were used to ensure that the experimenter was blinded to the treatment allocation. Animal group sizes were determined by referring to previous similar studies and confirmed with an online tool (http://clincalc.com/stats/samplesize.aspx) on the basis of our preliminary data.

Statistical Analysis

Statistical analysis was performed with GraphPad Prism 8 (GraphPad Software, Inc). Unless otherwise noted, all the quantitative studies were performed in at least triplicate. The dots in the bar graph represent independent individual biologic replicates, each of which represents the average of at least 3 technical replicates. The Kolmogorov-Smirnov test and a Q-Q plot were used to determine the normality of the data distribution. For continuous data with normal distributions, the F test was used to check the equality of variances. For 2 groups, normally distributed data were analyzed with an unpaired, 2-tailed Student t test (equal variance) or a Welch t test (unequal variance). For data that were not normally distributed, a 2-tailed Mann-Whitney U test (2 groups) was used. For ≥3 groups, normally distributed data with equal variance were analyzed by 1- or 2-way ANOVA, followed by the Sidak, Tukey, or Dunnett multiple-comparisons test. The Sidak multiple-comparisons test was applied when prespecified pairwise comparisons between selected groups were of primary interest. The Tukey honestly significant difference test was used when all possible pairwise group comparisons were performed. The Dunnett multiple-comparisons test was applied when multiple experimental groups were compared exclusively with a single control group. Linear association between 2 quantitative variables was assessed using the Pearson correlation coefficient, and significance was assessed using a 2-tailed t test. Detailed statistical analysis can be found in Table S1. The data are presented as mean±SD. All reported P values were 2-tailed (unless otherwise indicated). P<0.05 was considered to indicate statistical significance.

RESULTS

PRMT3 Expression Is Elevated in Human Calcific Aortic Valves

Protein arginine methylation has been implicated in various diseases, including chronic myeloid leukemia,²³ asthma,²⁴ and cancer.²⁵ To explore whether this PTM contributes to the pathogenesis of CAVD, we first assessed arginine methylation patterns in human aortic valves. Immunoblotting of aortic valve leaflet homogenates revealed a marked increase in global ADMA levels in human calcified aortic valves (hCAVs) compared with noncalcified controls, whereas the MMA and SDMA levels remained unchanged (Figure 1A and 1B). These findings suggest specific dysregulation of type 1 PRMTs, which catalyze the formation of asymmetric dimethylation.¹⁷ To identify the specific type 1 PRMT (PRMT1–PRMT4, PRMT6, or PRMT8) involved, we analyzed the expression profiles of these enzymes in 8 hCAVs and 8 noncalcified controls obtained from male patients (see Table S2 for demographic characteristics). Among the type 1 PRMTs, PRMT3 and PRMT8 were differentially expressed between hCAVs and noncalcified controls (Figure 1C–1E and Figure S1A–S1E). Notably, only the PRMT3 protein levels were significantly elevated in hCAVs, which may account for the increased global asymmetric dimethylation levels. This result was accompanied by an increase in OPN (osteopontin) and Osx (Osterix) expression (Figure 1F and 1G). Moreover, quantitative polymerase chain reaction analysis of 50 paired human aortic valve leaflets confirmed that PRMT3 mRNA expression was consistently elevated in hCAVs (50 hCAVs and 50 noncalcified controls; see Table S2 for demographic characteristics; Figure 1H). Next, to identify the predominant cellular source of PRMT3 within the aortic valve, hVICs and human valvular endothelial cells (hVECs) were isolated from 10 male donors with CAVD (see Table S3 for demographic characteristics). PRMT3 expression was significantly higher in hVICs than in hVECs (Figure 1I), suggesting that hVICs were the primary source of PRMT3 upregulation in hCAVs.

Figure 1.

PRMT3 expression is elevated in human calcific aortic valves. A, General overview of protein arginine methylation modes. Type 1, 2, 3 PRMTs (protein arginine methyltransferases) responsible for conversion of arginine to monomethyl arginine (MMA). Further generation of symmetrical dimethyl arginine (SDMA) is catalyzed by type 3 PMRTs; formation of asymmetrical dimethyl arginine (ADMA) is catalyzed by type 1 PMRTs. B, Representative immunoblots show the levels of pan-MMA, pan-SDMA, and pan-ADMA in aortic valve leaflets from human calcified aortic valves (hCAVs) and noncalcified controls. C, Protein levels of PRMT3, PRMT8, OPN (osteopontin), and Osx (Osterix) in aortic valve leaflets from hCAVs and noncalcified controls measured by Western blotting. D through G, Bar graphs show the quantification of PRMT3, PRMT8, OPN, and Osx protein levels (n=8; unpaired Student t test). H, mRNA levels of PRMT3 in aortic valve leaflets from hCAVs and noncalcified controls were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of the PRMT3 mRNA levels (n=50; unpaired Welch t test). I, mRNA levels of PRMT3 in human aortic valve interstitial cells (hVICs) and human valvular endothelial cells (hVECs) from hCAVs (n=10; unpaired Welch t test). J, Analysis of the correlation between the PRMT3 mRNA levels and SPP1 mRNA levels in aortic valve leaflets from hCAVs. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). K, Analysis of the correlation between PRMT3 mRNA levels and SP7 mRNA levels in aortic valve leaflets from hCAVs. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). L, Analysis of the correlation between PRMT3 mRNA levels in aortic valve leaflets from hCAVs and the aortic valve area (cm²) of the heart. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). M, Analysis of the correlation between PRMT3 mRNA levels in aortic valve leaflets from hCAVs and the mean transvalvular pressure gradient (mm Hg) of the heart. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). N, Analysis of the correlation between PRMT3 mRNA levels in aortic valve leaflets from hCAVs and the peak transvalvular pressure gradient (mm Hg) of the heart. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). O, Analysis of the correlation between PRMT3 mRNA levels in aortic valve leaflets from hCAVs and the Agatston scores of aortic valves. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test).

To further evaluate the clinical relevance of PRMT3 in CAVD, we examined its association with osteogenic markers and echocardiographic and computed tomographic measures in a validation cohort including 50 paired human aortic valve leaflets (50 hCAVs and 50 noncalcified controls; see Table S2 for demographic characteristics). PRMT3 expression was positively correlated with expression of the osteogenic markers OPN and Osx (Figure 1J and 1K), supporting a potential role in aortic valve calcification pathogenesis. Furthermore, higher PRMT3 expression levels were significantly associated with echocardiographic and computed tomographic indicators of CAVD severity, including a reduced aortic valve area, increased mean and peak transvalvular pressure gradients, and Agatston scores of aortic valves (Figure 1L–1O and Figure S1F and S1G). These results suggested that PRMT3 might serve as a novel stratification criterion for CAVD. Collectively, these findings demonstrated that PRMT3 expression is elevated in hCAVs during CAVD.

RUNX2 Recruits P300 to Promote PRMT3 Expression Through H3K27ac

To examine the molecular mechanisms underlying PRMT3 upregulation in hCAVs, we used the JASPAR database (https://jaspar.genereg.net) to predict transcription factors and corresponding binding sites within the PRMT3 promoter. RUNX2 (runt-related transcription factor 2) was predicted to bind the PRMT3 promoter at −619 to −611 bp upstream of the transcription start site (Figure 2A and 2B). As an osteogenesis-related transcription factor implicated in aortic valve calcification,²⁶ RUNX2 is markedly upregulated in aortic valve leaflets and in hVICs from hCAVs compared with noncalcified controls (Figure 2C and 2D). Chromatin immunoprecipitation–polymerase chain reaction analysis of human aortic valve leaflets correspondingly revealed increased RUNX2 enrichment in the PRMT3 promoter in hCAVs compared with controls (Figure 2E). In hVICs, both RUNX2 overexpression and osteogenic induction increased the binding of RUNX2 to the PRMT3 promoter (Figure 2F and 2G), confirming its enrichment in this promoter region during osteogenic activation. To investigate the relationship between RUNX2 and PRMT3, we conducted luciferase assays with the PRMT3 promoter. RUNX2 overexpression induced a dose-dependent increase in the activity of the wild-type (WT) PRMT3 promoter but not in the activity of the deletion-variant PRMT3 promoter, indicating that RUNX2 directly regulates the transcription of PRMT3 (Figure 2H). In addition, RUNX2 expression was positively correlated with PRMT3 expression in clinical hCAVs (50 hCAVs and 50 noncalcified controls; see Table S2 for demographic characteristics; Figure 2I). These findings suggested that the transcription factor RUNX2 binds to the PRMT3 promoter and promotes PRMT3 transcription.

Figure 2.

RUNX2 recruits P300 to promote PRMT3 expression through H3K27ac. A, The binding sites of RUNX2 (runt-related transcription factor 2) in the PRMT3 promoter were predicted using the JASPAR database (https://jaspar.genereg.net). The diagram shows the RUNX2-binding peaks in the PRMT3 gene locus. B, Schematic diagram shows that the base sequence represents the consensus RUNX2-binding motif (JASPAR). C, The mRNA levels of RUNX2 in human calcific aortic valves (hCAVs) and noncalcified controls were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of RUNX2 mRNA levels (n=10; unpaired Welch t test). D, mRNA levels of RUNX2 in human aortic valve interstitial cells (hVICs) from hCAVs and noncalcified controls. The bar graph shows the quantification of RUNX2 mRNA levels (n=10; unpaired Welch t test). E, Chromatin immunoprecipitation–polymerase chain reaction (ChIP-PCR) analysis of the enrichment of RUNX2 in the PRMT3 promoter in hCAVs and noncalcified controls. The bar graph shows the quantification of the enrichment of RUNX2 in the PRMT3 promoter (n=10; unpaired Student t test). F, ChIP-PCR analysis of the enrichment of RUNX2 in the PRMT3 promoter in hVICs transfected with or without RUNX2. The bar graph shows the quantification of the enrichment of RUNX2 in the PRMT3 promoter (n=6; 2-way ANOVA followed by the Sidak multiple-comparisons test). G, ChIP-PCR analysis of the enrichment of RUNX2 in the PRMT3 promoter in hVICs cultured with or without osteogenic medium (OM). The bar graph shows the quantification of the enrichment of RUNX2 in the PRMT3 promoter (n=6; 2-way ANOVA followed by the Sidak multiple-comparisons test). H, Luciferase reporter assay shows that RUNX2 transfection increased reporter gene expression in hVICs. Luciferase reporter vectors with promoters containing the indicated wild-type (WT) PRMT3 promoter or the deletion-variant (Del) PRMT3 promoter. The bar graph shows the quantification of the relative luciferase activity of hVICs (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). I, Analysis of the correlation between PRMT3 mRNA levels and RUNX2 mRNA levels in hCAVs. Least squares regression lines shown with 95% confidence bands (n=50; linear association between 2 quantitative variables was assessed using Pearson correlation coefficient; significance was assessed using a 2-tailed t test). J, Coimmunoprecipitation was used to detect the interaction between endogenous RUNX2 and P300. Protein lysates from aortic valve leaflets (left panels) and hVICs (right panels) were immunoprecipitated with an anti-RUNX2 antibody and immunoblotted with an anti-P300 antibody (upper panels) or immunoprecipitated with an anti-P300 antibody and immunoblotted with an anti-RUNX2 antibody (lower panels). K, ChIP-PCR analysis of the enrichment of H3K27ac in the PRMT3 promoter in hCAVs and noncalcified controls. The bar graph shows the quantification of the enrichment of H3K27ac in the PRMT3 promoter (n=10; unpaired Welch t test). L, ChIP-PCR analysis of the enrichment of H3K27ac in the PRMT3 promoter in hVICs transfected with RUNX2 or P300. The bar graph shows the quantification of the enrichment of H3K27ac in the PRMT3 promoter (n=6; 2-way ANOVA followed by the Sidak multiple-comparisons test). M, ChIP-PCR analysis of the enrichment of H3K27ac in the PRMT3 promoter in OM-cultured hVICs treated with ShNC or ShRUNX2. The bar graph shows the quantification of the enrichment of H3K27ac in the PRMT3 promoter (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). N, The mRNA levels of PRMT3 in OM-cultured hVICs treated with CADD522 or ShRUNX2 were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of the PRMT3 mRNA levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). O, ChIP-PCR analysis of the enrichment of RUNX2 in the PRMT3 promoter in OM-cultured hVICs treated with CADD522 or ShRUNX2. The bar graph shows the quantification of the enrichment of RUNX2 in the PRMT3 promoter (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). P, The mRNA levels of PRMT3 in OM-cultured hVICs treated with curcumin or ShP300 were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of the PRMT3 mRNA levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). Q, ChIP-PCR analysis of the enrichment of RUNX2 in the PRMT3 promoter in OM-cultured hVICs treated with curcumin or ShP300. The bar graph shows the quantification of the enrichment of RUNX2 binding to the PRMT3 promoter (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). EV indicates empty vector; and IgG, immunoglobulin G.

Because RUNX2 recruits the histone acetyltransferase P300 to regulate gene transcription,²⁷ we performed coimmunoprecipitation assays in aortic valve leaflets and hVICs from hCAVs; our results confirmed a physical interaction between RUNX2 and P300 (Figure 2J). Because P300 catalyzes H3K27 acetylation,²⁸ we explored whether the RUNX2–P300 interaction promotes histone H3 lysine 27 acetylation (H3K27ac) enrichment in the PRMT3 promoter. Chromatin immunoprecipitation–quantitative polymerase chain reaction analysis confirmed significantly increased H3K27ac levels in the PRMT3 promoter in hCAVs compared with noncalcified controls (Figure 2K). In vitro, we found that RUNX2 or P300 overexpression in hVICs increased H3K27ac enrichment in the PRMT3 promoter (Figure 2L), whereas RUNX2 knockdown reduced osteogenic medium (OM)–induced H3K27ac levels (Figure 2M). To further confirm the role of the RUNX2–P300 interaction in regulating H3K27ac and PRMT3 gene expression, this interaction was disrupted in hVICs. Inhibition of the RUNX2–DNA interaction by CADD522, a RUNX2–DNA binding inhibitor,²⁹ or by silencing RUNX2, markedly decreased both PRMT3 expression and RUNX2 occupancy in the PRMT3 promoter (Figure 2N and 2O). In parallel, treatment with curcumin, which is a selective inhibitor of P300/CBP HAT activity,³⁰ or knockdown of P300, suppressed PRMT3 expression and reduced RUNX2 recruitment to the PRMT3 promoter (Figure 2P and 2Q). These data suggested that RUNX2 recruits P300 to transactivate PRMT3 genes through H3K27ac.

Prmt3 Haploinsufficiency Ameliorates Aortic Valve Calcification in Mice

We next investigated levels of asymmetric dimethylation and Prmt3 in a mouse model of aortic valve calcification. Consistent with clinical observations, global asymmetric dimethylation and Prmt3 expression were elevated in mouse calcified aortic valves from high-cholesterol diet (HCD)–fed ApoE^−/− mice (Figure 3A–3C).⁸ Then, Prmt3-deficient mice were generated through CRISPR-Cas9 gene editing (Figure S2A). However, homozygous Prmt3-knockout (Prmt3^−/−) mice exhibited significant reproductive impairment and reduced body sizes, consistent with a previous report.³¹ Owing to the limited viability of Prmt3^−/− mice, Prmt3 haploinsufficiency (Prmt3^±) mice were used for subsequent experiments. Prmt3^± mice were crossed with ApoE^−/− mice to generate Prmt3^±ApoE^−/− mice. Efficient Prmt3 knockdown in the aortic valves was confirmed by quantitative polymerase chain reaction and immunoblotting (Figure S2B and S2C). The mice were then fed an HCD for 24 weeks to induce CAVD, as previously described.⁸ This model successfully recapitulated key clinical features of CAVD, including increased peak transvalvular jet velocity, elevated mean pressure gradient, and reduced aortic valve area (Figure 3D–3F). Notably, Prmt3 haploinsufficiency attenuated these changes in HCD-fed ApoE^−/− mice (Figure 3D–3F), suggesting that Prmt3 deficiency protects against diet-induced aortic valve dysfunction.

Figure 3.

Prmt3 haploinsufficiency ameliorates aortic valve calcification in mice. A, Representative immunoblots show the levels of pan–monomethylation (MMA), pan–symmetric demethylation (SDMA), and pan–asymmetric dimethylation (ADMA) in aortic valve leaflets from mouse calcified aortic valves (mCAVs) and noncalcified controls. B, The mRNA levels of PRMT3 in mCAVs and noncalcified controls were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of the PRMT3 mRNA levels (n=10; unpaired Welch t test). C, The protein levels of PRMT3 (protein arginine methyltransferase 3) in mCAVs and noncalcified controls were measured by Western blotting. Bar graphs show the quantification of PRMT3 protein levels (n=10; unpaired Student t test). D through F, Echocardiographic assessment of the severity of aortic valve stenosis in ApoE^−/− and PRMT3^±ApoE^−/− mice fed a high-cholesterol diet (HCD) or normal diet (ND). Pulsed-wave Doppler across the aortic valve was performed, and the peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area were computed. Bar graphs show the quantification of peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area (n=10; 2-way ANOVA followed by the Sidak multiple-comparisons test). G and H, Hematoxylin & eosin staining–based analysis of the thickness of aortic valve leaflets. The bar graph shows the quantification of leaflet thickness (n=10; 2-way ANOVA followed by the Sidak multiple-comparisons test). I and J, Alizarin Red staining of aortic valves; the total calcified area was analyzed accordingly. The bar graph shows the quantification of the total calcified area (n=10; 2-way ANOVA followed by the Sidak multiple-comparisons test). K, The bar graph shows the quantification of calcium content in aortic valve leaflets (n=10; 2-way ANOVA followed by the Sidak multiple-comparisons test). L, The protein levels of Opn (osteopontin) and Osx (Osterix) in mouse aortic valves were measured by Western blotting. The bar graph shows the quantification of Opn and Osx protein levels (n=10; 2-way ANOVA followed by the Sidak multiple-comparisons test).

Next, we evaluated aortic valve morphology and calcification in ApoE^−/− mice. As expected, HCD feeding induced pronounced aortic valve calcification, which was characterized by increased leaflet thickness (Figure 3G and 3H) and calcium deposition, as evidenced by hematoxylin & eosin and Alizarin Red staining (Figure 3I–3K). Prmt3 depletion significantly attenuated these pathologic effects (Figure 3I–3K). Consistent with these findings, the expression levels of the osteogenic markers Opn and Osx were markedly reduced in the aortic valves of Prmt3^±ApoE^−/− mice compared with those of control mice (Figure 3L). Notably, Prmt3 silencing did not significantly affect glucose, total cholesterol, LDL, or triglyceride levels (Figure S2D). Taken together, these results indicated that the loss of 1 copy of Prmt3 ameliorates aortic valve calcification in ApoE^−/− mice independent of metabolic regulation.

Pharmacologic Inhibition of Prmt3 (SGC707) Alleviates Aortic Valve Calcification in Mice

To evaluate the therapeutic potential of pharmacologic Prmt3 inhibition in CAVD, HCD-fed ApoE^−/− mice were treated with the selective Prmt3 inhibitor SGC707.³² SGC707 treatment did not alter metabolic measures, including glucose and lipid profiles (Figure S3A), but continuous-wave Doppler echocardiography demonstrated improved aortic valve hemodynamics, as evidenced by reduced peak transvalvular jet velocity and mean pressure gradient, as well as an increased aortic valve area (Figure 4A–4D). Histologic analysis further revealed that SGC707 markedly attenuated valve leaflet thickening and calcium deposition (Figure 4E–4G). In addition, SGC707 treatment significantly reduced expression of the osteogenic markers Opn and Osx in aortic valve tissues (Figure 4H–4J). These findings indicated a general feasibility of pharmacologic PRMT3 inhibition for mitigating the progression of aortic valve calcification, highlighting PRMT3 as a promising target for CAVD intervention.

Figure 4.

Targeting PRMT3 suppresses aortic valve calcification development in mice. A through D, Echocardiographic assessment of the severity of aortic valve stenosis in high-cholesterol diet–fed ApoE^−/− mice treated with vehicle or SGC707. Pulsed-wave Doppler across the aortic valve was performed, and the peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area were computed. Bar graphs show the quantification of the peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area (n=10; unpaired Student t test). E, Hematoxylin & eosin (H&E)–based analysis of the thickness of aortic valve leaflets. The bar graph shows the quantification of leaflet thickness (n=10; unpaired Student t test). F, Alizarin Red staining of aortic valves and the total calcified area were analyzed accordingly. The bar graph shows the quantification of the total calcified area (n=10; unpaired Student t test). G, The bar graph shows the quantification of the calcium content in aortic valve leaflets (n=10; unpaired Student t test). H through J, The protein levels of Opn (osteopontin) and Osx (Osterix) in mouse aortic valves were measured by Western blotting. Bar graphs show the quantification of Opn and Osx protein levels (n=10; unpaired Student t test). K through N, Echocardiographic assessment of the severity of aortic valve stenosis in high-cholesterol diet–fed ApoE^−/−mice treated with vehicle or compound 11. Pulsed-wave Doppler across the aortic valve was performed, and the peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area were computed. Bar graphs show the quantification of peak transvalvular jet velocity, mean transvalvular pressure gradient, and aortic valve area (n=10; unpaired Student t test). O, H&E–based analysis of the thickness of aortic valve leaflets. The bar graph shows the quantification of leaflet thickness (n=10; unpaired Student t test). P, Alizarin Red staining of aortic valves and the total calcified area were analyzed accordingly. The bar graph shows the quantification of the total calcified area (n=10; unpaired Student t test). Q, The bar graph shows quantification of the calcium content in aortic valve leaflets (n=10; unpaired Student t test). R through T, The protein levels of Opn and Osx in the mouse aortic valves were measured by Western blotting. Bar graphs show the quantification of Opn and Osx protein levels (n=10; unpaired Student t test).

PROTAC-Mediated PRMT3 Degradation as a Novel Strategy for CAVD Treatment

Targeted protein degradation, which offers high efficacy, specificity, and a favorable safety profile, has shown therapeutic potential in preclinical models of cardiovascular disease.³³ A recent study identified compound 11 as a first-in-class PRMT3-targeting PROTAC (proteolysis-targeting chimera) that selectively degrades PRMT3 and inhibits leukemia cell proliferation.¹⁸ Therefore, we used compound 11 to investigate whether PRMT3 degradation could be a novel approach for treating CAVD. Initial in vitro assays revealed that compound 11 exhibited minimal cytotoxicity in hVICs, even at high concentrations (Figure S3B). Treatment with 5 µM compound 11 resulted in robust PRMT3 degradation within 12 hours (Figure S3C and S3D), and this effect was abolished by the proteasome inhibitor MG132 (Figure S3E); these results indicate that compound 11 effectively inhibits PRMT3 expression in hVICs without significantly affecting cell viability.

To assess the therapeutic efficacy of compound 11 in vivo, it was administered to HCD-fed ApoE^−/− mice. Metabolic measures, including glucose, total cholesterol, LDL, and triglyceride levels, remained unaffected after treatment (Figure S3F). Echocardiographic analysis revealed that compound 11 significantly reduced the peak transvalvular jet velocity and mean pressure gradient and increased the aortic valve area (Figure 4K–4N). Moreover, compound 11 treatment also attenuated aortic valve leaflet thickening and calcium deposition (Figure 4O–4Q). Consistent with these findings, compound 11 treatment significantly downregulated the osteogenic markers OPN and Osx in valve tissues, mirroring the effects observed in the PRMT3-deficient models and after pharmacologic inhibition of PRMT3 (Figure 4R–4T). These results highlighted that PRMT3 degradation by compound 11 is a novel and effective approach for treating CAVD in mice.

PRMT3 Promotes the Osteogenic Differentiation of hVICs Through Its Enzymatic Activity

Because osteogenic differentiation of hVICs is a key cellular mechanism in aortic valve calcification pathogenesis,³⁴ we next investigated the role of PRMT3 in this process. hVICs were isolated from human aortic valve leaflets (see Table S2 for patient characteristics) and initially maintained in a quiescent state. At baseline, PRMT3 expression was relatively low at both mRNA and protein levels (Figure 5A and 5B). After osteogenic induction for 14 days, a time-dependent increase in PRMT3 expression was observed, in parallel with upregulation of the osteogenic marker OPN (Figure 5A and 5B). Next, we investigated the role of PRMT3 in the osteogenic differentiation of hVICs in vitro. We found that PRMT3 knockdown suppressed the OM-induced increases in the expression of the osteogenic markers OPN and Osx as well as calcified nodule formation (Figure 5C–5E). Because PRMT3 is a type 1 arginine methyltransferase, we next investigated whether the pro-osteogenic effects of PRMT3 depend on its enzymatic activity. PRMT3-knockdown hVICs were reconstituted with either PRMT3-WT or a catalytically inactive variant (PRMT3-3M; G263A/C264A/G265A).³⁵ Reexpression of PRMT3-WT, but not the enzymatically inactive PRMT3-3M, restored the expression of osteogenic markers and calcification capacity in PRMT3-silenced hVICs under conditions of OM stimulation (Figure 5C–5E). Conversely, overexpression of PRMT3 enhanced OM-induced osteogenic responses in hVICs, whereas cotreatment with the selective PRMT3 inhibitor SGC707 effectively reversed these effects (Figure 5F–5H). These results demonstrated that PRMT3 facilitates the osteogenic differentiation of hVICs in vitro through its enzymatic activity.

Figure 5.

PRMT3 promotes osteogenic differentiation of human aortic valve interstitial cells through its enzymatic activity. A, The mRNA levels of PRMT3 in human aortic valve interstitial cells (hVICs) at different time points (days 0, 1, 3, 5, 7, and 14) after culture in osteogenic medium (OM) were measured by quantitative polymerase chain reaction. The bar graph shows the quantification of the PRMT3 mRNA levels (n=6; 1-way ANOVA followed by the Dunnett multiple-comparisons test). B, The protein levels of PRMT3 (protein arginine methyltransferase 3) and OPN (osteopontin) in hVICs at different time points (days 0, 1, 3, 5, 7, and 14) after culture in OM were measured by Western blotting. Bar graphs show the quantification of the PRMT3 and OPN protein levels (n=6; 1-way ANOVA followed by the Dunnett multiple-comparisons test). C, The protein levels of OPN and Osx (Osterix) in hVICs cultured with OM were measured by Western blotting. Bar graphs show the quantification of OPN and Osx protein levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). D and E, Alizarin red staining of mineralization nodules in calcified hVICs. The bar graph shows the quantification of calcium deposition levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). F, The protein levels of OPN and Osx in hVICs cultured with OM were measured by Western blotting. Bar graphs show the quantification of OPN and Osx protein levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). G and H, Alizarin Red staining of mineralization nodules in calcified hVICs. The bar graph shows the quantification of calcium deposition levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). EV indicates empty vector.

PRMT3 Methylates PCSK9 at R582

Because PRMT3 functions through its arginine methyltransferase activity,³¹ we performed liquid chromatography–tandem mass spectrometry analysis of proteins that coimmunoprecipitated with PRMT3 in OM-cultured hVICs and hCAVs to identify potential PRMT3 substrates. The experimental workflow is shown in Figure 6A. After immunoglobulin-derived peptides were removed, 236 PRMT3-interacting proteins were identified in OM-cultured hVICs (Table S4, and Figure 6A), and 350 were identified in hCAVs (Table S5, and Figure 6A). Cross-analysis of both data sets revealed a shared subset of putative PRMT3 substrates, including PCSK9, MYH9 (myosin heavy chain 9), VIM (vimentin), and PABPC1 (poly[A]-binding protein cytoplasmic 1; Figure 6A). Among these putative substrates, PCSK9 has been implicated in CAVD pathogenesis.⁸ Repeated immunoprecipitation of human and mouse valvular interstitial cell (VIC) lysates followed by liquid chromatography–tandem mass spectrometry analysis verified that PCSK9 is a PRMT3-binding protein (Figure 6B and Figure S4A). Coimmunoprecipitation experiments also confirmed the interaction of endogenous PCSK9 and PRMT3 in hVICs (Figure 6C). Next, we determined whether PRMT3 catalyzes arginine methylation of PCSK9. hVICs were treated with the pan-PRMT inhibitor adenosine dialdehyde, a PRMT3-selective inhibitor (SGC707), or PRMT3-targeting shRNA (shPRMT3). Immunoprecipitation of PCSK9 followed by asymmetric dimethylation-specific immunoblotting revealed significantly reduced asymmetric dimethylation levels across all the treatment groups (Figure 6D–6F), suggesting that PRMT3 directly mediated the asymmetric dimethylation of PCSK9. To assess whether PRMT3 enzymatic activity was required for PCSK9 asymmetric dimethylation, hVICs were transfected with either PRMT3-WT or a catalytically inactive variant PRMT3-3M (G263A/C264A/G265A). Only PRMT3-WT increased the asymmetric dimethylation of PCSK9 (Figure 6G), indicating that the PRMT3-mediated asymmetric dimethylation of PCSK9 depended on its enzyme activity. To further elucidate the underlying mechanism, we used the PRmePRed tool (https://bioinfo.icgeb.res.in/PRmePRed) to predict potential arginine methylation sites on PCSK9. Five candidate residues were identified: R582, R469, R160, R434, and R165. R582 was the only asymmetric dimethylation site of PCSK9 (Figure 6H). Immunoprecipitation with an anti-PCSK9 antibody followed by mass spectrometry further confirmed that PRMT3 methylated PCSK9 at R582 in mice and humans (Figure 6I and Figure S4B). To assess site specificity, the PCSK9 variants R582K and R469K (arginine-to-lysine variant) were overexpressed in hVICs. Only the R582K variation attenuated the PRMT3-mediated asymmetric dimethylation of PCSK9 (Figure 6J), confirming that R582 is the methylation site. Next, we explored whether PRMT3-mediated methylation of PCSK9 at R582 depends on its enzymatic activity. PCSK9-WT, PCSK9-R582K, PRMT3-WT, and the catalytically inactive variant PRMT3-3M were purified from hVICs, and subjected to in vitro methylation assays in the presence of the methyl donor S-adenosylmethionine. PRMT3-WT, but not PRMT3-3M, induced asymmetric dimethylation of PCSK9-WT. PRMT3-WT failed to methylate the PCSK9-R582K variant, suggesting the enzymatic dependency and site specificity of PRMT3 activity (Figure 6K). To assess the effects of the PRMT3-mediated asymmetric dimethylation of PCSK9, we silenced PRMT3 or PCSK9 in hVICs. PRMT3 knockdown reduced the PCSK9 protein levels (Figure 6L), whereas PCSK9 silencing had no effect on PRMT3 expression (Figure 6M), suggesting that PRMT3-mediated asymmetric dimethylation inhibits PCSK9 expression. Protein expression is regulated in part by degradation, primarily through the ubiquitin-proteasome or autophagy-lysosome pathways.³⁶ To determine the mechanisms by which PRMT3 limits PCSK9 degradation, PRMT3-knockdown hVICs were treated with the proteasome inhibitor MG132 or the lysosomal inhibitor bafilomycin A1 (BafA1). MG132, but not BafA1, restored the PCSK9 protein levels, indicating that PRMT3 reduced PCSK9 degradation through the ubiquitin-proteasome pathway (Figure 6N). Moreover, intracellular trafficking and compartmentation of PCSK9 are important for overall PCSK9 function.³⁷ To explore how this is affected by PRMT3-induced methylation, and how this affects PCSK9 degradation, the Golgi apparatus, endoplasmic reticulum, and lysosome, which are key organelles of hVICs involved in PCSK9 protein synthesis and degradation,³⁸ were isolated by using extraction kits for further testing. We found that the expression of PCSK9 was reduced in OM-induced hVICs with PRMT3 knockdown (Figure S5A), whereas its subcellular distribution across Golgi apparatus, endoplasmic reticulum, and lysosome showed no significant differences (Figure S5B–S5D). These results indicate that PRMT3-induced methylation has minimal effect on intracellular trafficking, compartmentalization, or lysosomal degradation of PCSK9. Collectively, these findings suggested that PRMT3 catalyzes asymmetric dimethylation of PCSK9 at R582 and inhibits its ubiquitination-mediated degradation.

Figure 6.

PRMT3 methylates PCSK9 at R582. A, Schematic illustration of the experimental design of the 2 interactomes that were used for PRMT3 (protein arginine methyltransferase 3) substrate screening. Human aortic valve interstitial cells (hVICs) were transfected with either the Flag-vector or Flag-PRMT3 after osteogenic medium (OM) stimulation. hCAVs were lysed and divided into 2 portions. The bound proteins were extracted, digested into peptides, and analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS). The list shows the potential PRMT3 substrates that were identified from the interactomes. B, Protein lysates from hVICs were immunoprecipitated with an anti-PRMT3 antibody, followed by LC-MS/MS. The bar graph shows PCSK9 peptides derived from the mass spectrometric analysis. C, Coimmunoprecipitation shows the interaction between endogenous PCSK9 (proprotein convertase subtilisin/kexin type 9) and PRMT3. Protein lysates from hVICs were immunoprecipitated with an anti-PCSK9 antibody and immunoblotted with an anti-PRMT3 antibody (upper panel) or immunoprecipitated with an anti-PRMT3 antibody and immunoblotted with an anti-PCSK9 antibody (lower panel). D through F, Coimmunoprecipitation shows the asymmetric demethylation (ADMA) levels of PCSK9. OM-cultured hVICs were treated with vehicle, adenosine dialdehyde, SGC707, shNC, shPRMT3, or MG132. Protein lysates from hVICs were immunoprecipitated with an anti-PCSK9 antibody and immunoblotted with an anti-ADMA antibody. G, Coimmunoprecipitation shows the asymmetric dimethylation levels of PCSK9. OM-cultured hVICs were transfected with wild-type (WT) PRMT3 or PRMT3-3M and treated with MG132. Protein lysates from hVICs were immunoprecipitated with an anti-PCSK9 antibody and immunoblotted with an anti-ADMA antibody. H, Prediction of PCSK9 protein methylations by PRmePRed. I, Protein lysates from hVICs were immunoprecipitated with an anti-PCSK9 antibody, followed by LC-MS/MS. The bar graph shows the asymmetric dimethylation levels of PCSK9 at R582. J, Coimmunoprecipitation shows the asymmetric dimethylation levels of PCSK9. OM-cultured hVICs were transfected with Myc-PCSK9-WT, Myc-PCSK9-R469K, or Myc-PCSK9-R582K and treated with MG132. Protein lysates from hVICs were immunoprecipitated with an anti-Myc antibody and immunoblotted with an anti-ADMA antibody. K, Myc-PCSK9-WT, Myc-PCSK9-R582K, GST-PRMT3-WT, and GST-PRMT3-3M (G263A/C264A/G265A) proteins were overexpressed and purified for PRMT3 in vitro methylation assays. L, The protein levels of PCSK9 and PRMT3 in hVICs with or without PRMT3 knockdown were measured by Western blotting. M, The protein levels of PCSK9 and PRMT3 in hVICs with or without PCSK9 knockdown were measured by Western blotting. N, The protein levels of PCSK9 in hVICs treated with vehicle, MG132, BafA1, shNC, or shPRMT3 were measured by Western blotting. EV indicates empty vector; and IgG, immunoglobulin G.

In addition, HIF-1α is an important procalcific factor in valvular interstitial cells and vascular smooth muscle cells,^39,40 and it can also be methylated by PRMT3 in various cell types.^21,41 To explore whether HIF-1α plays a role in PRMT3-induced aortic valve calcification, we performed coimmunoprecipitation assays in aortic valve leaflets and hVICs from hCAVs, which confirmed a physical interaction between PRMT3 and HIF-1α (Figure S6A). We further found that asymmetric dimethylation of HIF-1α was increased in aortic valve leaflets and hVICs from hCAVs, consistent with the elevated PRMT3 expression (Figure S6B and S6C and Figure 1C). However, HIF-1α expression showed only a slight increase (Figure S6B and S6C). To explore the role of PRMT3 in the asymmetric dimethylation and expression of HIF-1α during aortic valve calcification, we induced osteogenic differentiation of hVICs with PRMT3-selective inhibitor (SGC707) or PRMT3-targeting shRNA (shPRMT3) treatment in vitro. PRMT3 inhibition or silencing alleviated the asymmetric dimethylation of HIF-1α, but had no significant effect on its expression (Figure S6D and S6E). We also found that PRMT3 overexpression modestly increased HIF-1α expression in OM-cultured hVICs, an effect that was reversed by SGC707 (Figure S6F and S6G). These findings suggest that PRMT3-mediated methylation of HIF-1α may contribute to a slight increase in HIF-1α expression during aortic valve calcification. To evaluate the contribution of HIF-1α in the process of PRMT3-induced osteogenic differentiation of hVICs, OM-induced hVICs were overexpressed with PRMT3 and treated with PX-478 (an HIF-1α inhibitor) or shHIF1A.⁴² We found that neither pharmacologic inhibition nor genetic knockdown of HIF-1α had a significant effect on the procalcific effects mediated by PRMT3 (Figure S6H–S6M). These results demonstrated that HIF-1α is not involved in PRMT3-mediated aortic valve calcification.

Biglycan, a leucine-rich proteoglycan and key extracellular matrix component, accumulates in aortic valves from patients with CAVD.⁴³ Previous studies have demonstrated that hVIC-derived biglycan contributes to aortic valve calcification.⁴⁴ To investigate whether PRMT3 affects biglycan during aortic valve calcification, aortic valve leaflets and hVICs from hCAVs and noncalcified controls were tested. Asymmetric dimethylation of biglycan showed no significant difference between the 2 groups (Figure S7A and S7B). These results may be attributable to the distinct localization of biglycan, which is predominantly extracellular, whereas PRMT3 is primarily expressed intracellularly in VICs.^44,45 In addition, consistent with previous research findings, we found increased biglycan levels in aortic valve leaflets and hVICs from hCAVs, as well as in osteogenically differentiated hVICs (Figure S7A–S7C).⁴⁴ For further investigation, hVICs were induced to undergo osteogenic differentiation and treated with either the PRMT3 inhibitor SGC707 or PRMT3-targeting shRNA. Neither PRMT3 inhibition nor PRMT3 silencing altered biglycan levels in OM-induced hVICs (Figure S7D and S7E). These findings suggested that PRMT3-promoted aortic valve calcification is independent of biglycan.

Elevated plasma lipoprotein(a) [Lp(a)] is an independent causal risk factor for cardiovascular disease and CAVD.⁴⁶ Lp(a) formation is restricted to humans, apes, old-world monkeys, and an unrelated version in European hedgehogs.⁴⁷ To study how PRMT3 interacts with Lp(a), WT (Prmt3^+/+) or Prmt3 haploinsufficiency (Prmt3^±) male mice were fasted for 3 hours and injected in the tail vein with 25 μg of human Lp(a) in 100 μL of saline according to previous published studies.⁴⁸ No significant difference in the kinetics of Lp(a) clearance was observed between Prmt3^+/+ and Prmt3^± mice (Figure S8A). In addition, to further examine the potential involvement of PRMT3 in Lp(a)-dependent calcification, we supplemented Lp(a) during osteogenic induction of VICs.^46,49 Lp(a) treatment did not significantly alter PRMT3 expression, nor did it substantially affect PCSK9 expression or ADMA levels (Figure S8B and S8C). These findings suggest that PRMT3 plays a minimal role in in Lp(a)-dependent calcification pathways under the experimental conditions tested.

PCSK9 R582 Methylation Inhibits CHIP-Mediated K575 Ubiquitination and Degradation

Because PRMT3 has not been previously implicated in direct regulation of the ubiquitin-proteasome system, we hypothesized that additional factors, such as E3 ligases or deubiquitinases, might cooperate with PRMT3 to inhibit PCSK9 degradation. To explore this possibility, immunoprecipitation of hVIC lysates was performed using an anti-PCSK9 antibody or control IgG (immunoglobulin G), followed by liquid chromatography–tandem mass spectrometry analysis to identify potential interacting proteins (Figure 7A). Among the candidates (Table S6), the E3 ligase CHIP is the only ubiquitination-related enzyme that was identified to interact with PCSK9 (Figure 7B). This interaction in hVICs was further confirmed by coimmunoprecipitation assays (Figure 7C). To further explore the effects of CHIP on the expression of PCSK9, we silenced or overexpressed CHIP in hVICs. CHIP knockdown increased PCSK9 protein levels without affecting its mRNA expression (Figure 7D and Figure S9A). In contrast, CHIP overexpression, but not enzymatically inactive CHIP variant (CHIP-H260Q) overexpression, reduced PCSK9 protein levels, which was reversed by the proteasome inhibitor MG132 (Figure 7E and Figure S9B). Moreover, CHIP knockdown prolonged the half-life of PCSK9 (Figure 7F) and suppressed its polyubiquitination (Figure 7G). In contrast, overexpression of CHIP-WT, but not CHIP-H260Q, shortened the PCSK9 half-life and increased its polyubiquitination (Figure S9C–S9E). Together, these results identified CHIP as a bona fide E3 ligase that promotes PCSK9 ubiquitination and degradation.

Figure 7.

PCSK9 R582-asymmetric dimethylation inhibits CHIP-mediated K575 ubiquitination and degradation. A, Silver-stained gel of total protein from human aortic valve interstitial cells (hVICs) after immunoprecipitation with IgG (immunoglobulin G) or an anti–PCSK9 (proprotein convertase subtilisin/kexin type 9) antibody. B, The bar graph shows CHIP (carboxyl terminus of Hsc70-interacting protein) peptides derived from mass spectrometry. C, Coimmunoprecipitation shows the interaction between CHIP and PCSK9. Protein lysates from hVICs were immunoprecipitated with an anti-PCSK9 antibody and immunoblotted with an anti-CHIP antibody (upper panel) or immunoprecipitated with an anti-CHIP antibody and immunoblotted with an anti-PCSK9 antibody (lower panel). D, The protein levels of PCSK9 and CHIP in hVICs with or without CHIP knockdown were measured by Western blotting. E, The protein levels of PCSK9 and CHIP in hVICs were measured by Western blotting. hVICs were transfected with or without wild-type (WT) Flag-CHIP and treated with or without MG132. F, Western blotting analysis of PCSK9 expression in hVICs transfected with shCHIP or shNC for 48 hours, followed by treatment with CHX (cycloheximide) for 0, 4, 8, or 16 hours. The bar graph shows the quantification of PCSK9 protein levels (n=5; 2-way ANOVA followed by the Sidak multiple-comparisons test). G, Coimmunoprecipitation shows the ubiquitination of PCSK9. hVICs were cotransfected with ShNC, ShCHIP, or HA-Ub and treated with MG132. Protein lysates were immunoprecipitated with an anti-PCSK9 antibody and immunoblotted with the indicated antibodies. H, The protein levels of PCSK9 and CHIP in hVICs were measured by Western blotting. hVICs were transfected with ShNC, ShPRMT3, or ShCHIP (upper panel). hVICs were transfected with ShNC or ShCHIP with or without SGC707 treatment (lower panel). I, The protein levels of PCSK9 and CHIP in hVICs were measured by Western blotting. hVICs were transfected with Flag-CHIP-WT, Myc-PCSK9-WT, or Myc-PCSK9-K575A. J and K, Western blotting analysis of PCSK9 expression in hVICs transfected with Myc-PCSK9-WT, Myc-PCSK9-K575A, or Myc-PCSK9-R582K for 48 hours, followed by treatment with CHX for 0, 4, 8, or 16 hours. The bar graph shows the quantification of PCSK9 protein levels (n=8; 2-way ANOVA followed by the Sidak multiple-comparisons test). L, Coimmunoprecipitation shows the ubiquitination of PCSK9. hVICs were cotransfected with Myc-PCSK9-WT, Myc-PCSK9-K575A, Myc-PCSK9-R582K, Flag-CHIP-WT, or HA-Ub and treated with MG132. Protein lysates were immunoprecipitated with an anti-Myc antibody and immunoblotted with the indicated antibodies. M, Coimmunoprecipitation shows the ubiquitination of PCSK9. hVICs were cotransfected with Myc-PCSK9-WT, Myc-PCSK9-K575A, or HA-Ub and treated with MG132. The cells were then treated with or without SGC707. Protein lysates were immunoprecipitated with an anti-Myc antibody and immunoblotted with the indicated antibodies. EV indicates empty vector.

To explore the role of PRMT3 in CHIP-mediated PCSK9 ubiquitination and degradation, hVICs were subjected to PRMT3 knockdown or PRMT3-specific inhibitor treatment, with or without concurrent CHIP silencing. CHIP depletion markedly reversed the PRMT3 inhibition-induced PCSK9 reduction (Figure 7H), indicating that CHIP is required for PCSK9 degradation upon PRMT3 inhibition. The full-length PCSK9 sequence (UniProt ID: Q8NBP7) was subsequently analyzed using the UbPred tool (http://ubpred.org). Among the predicted sites, K575, which is located adjacent to the R582 methylation site, was identified as a potential polyubiquitination target (Figure S9F and S9G). To determine whether CHIP mediates PCSK9 ubiquitination and degradation through residue K575, hVICs were transfected with CHIP together with either PCSK9-WT or a lysine-to-alanine variant at position 575 (PCSK9-K575A). CHIP overexpression significantly reduced PCSK9-WT protein levels, whereas PCSK9-K575A was resistant to CHIP-mediated degradation (Figure 7I). Similarly, overexpression of PCSK9-K575A, but not PCSK9-WT, prolonged the protein half-life of PCSK9 (Figure 7J and 7K), indicating that CHIP-mediated PCSK9 degradation depends on K575 ubiquitination. Moreover, we found that the asymmetric dimethylation-deficient variant PCSK9-R582K shortened the half-life of the PCSK9 protein (Figure 7J and 7K). To elucidate the interplay between PRMT3-mediated arginine methylation and CHIP-mediated PCSK9 K575 ubiquitination, hVICs were cotransfected with CHIP and PCSK9-WT, PCSK9-K575A, or PCSK9-R582K. CHIP overexpression markedly enhanced the polyubiquitination of PCSK9-WT but not PCSK9-K575A (Figure 7L). Notably, polyubiquitination of the PCSK9-R582K variant was elevated compared with that of PCSK9-WT, suggesting that PRMT3-mediated asymmetric dimethylation at R582 antagonizes the ubiquitination at K575. Finally, we found that either PRMT3 knockdown or PRMT3 inhibitor treatment increased the polyubiquitination of PCSK9-WT but had minimal effects on the K575A variant (Figure 7M and Figure S9H). These findings suggested that PRMT3 stabilizes PCSK9 by catalyzing asymmetric dimethylation at R582, which in turn attenuates CHIP-mediated ubiquitination at K575 and degradation.

PRMT3 Exerts Procalcific Effects Through PCSK9 Upregulation

Next, to determine whether the procalcific effects of PRMT3 are mediated through PCSK9, rescue experiments were conducted in hVICs. PRMT3 knockdown significantly attenuated osteogenic differentiation, as demonstrated by the reduced expression of the osteogenic markers OPN and Osx and decreased calcium deposition. These effects were reversed by PCSK9 reexpression, as shown by restored osteogenic marker levels and calcium deposition (Figure S10A–S10C). Conversely, PCSK9 silencing abrogated the procalcific effects of PRMT3 overexpression (Figure 8A–8C), suggesting that PRMT3 promotes the osteogenic differentiation of hVICs in a PCSK9-dependent manner. PRMT3 overexpression promoted in vitro calcification in hVICs expressing PCSK9-WT but not in those expressing the asymmetric dimethylation-deficient PCSK9-R582K variant (Figure S10D–S10F). Taken together, these results indicated that PRMT3 promotes aortic valve calcification by stabilizing and upregulating PCSK9 through asymmetric dimethylation at R582.

Figure 8.

PRMT3 promotes aortic valve calcification through the upregulation of PCSK9.A, The protein levels of PCSK9 (proprotein convertase subtilisin/kexin type 9), OPN (osteopontin), and Osx (Osterix) in human aortic valve interstitial cells (hVICs) were measured by Western blotting. hVICs were cotransfected with PRMT3, ShNC, or ShPCSK9. The bar graph shows the quantification of PCSK9, OPN, and Osx protein levels (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). B and C, Alizarin Red staining of mineralization nodules in calcified hVICs. hVICs were cotransfected with PRMT3, ShNC, or ShPCSK9. The bar graph shows the quantification of calcium deposition levels (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). D, The protein levels of LOX-1 (lectin-like oxidized low-density lipoprotein receptor 1) in aortic valve leaflets from human calcified aortic valves (hCAVs) and noncalcified controls were measured by Western blotting. The bar graph shows the quantification of LOX-1 protein levels (n=10; unpaired Student t test). E, The protein levels of LOX-1 in hVICs from hCAVs and noncalcified controls were measured by Western blotting. The bar graph shows the quantification of LOX-1 protein levels (n=10; unpaired Student t test). F, The protein levels of LOX-1 in hVICs were measured by Western blotting. hVICs were cotransfected with PCSK9, ShNC, or ShPCSK9. The bar graph shows the quantification of LOX-1 protein levels (n=6; 1-way ANOVA followed by the Tukey multiple-comparisons test). G and H, Representative confocal images show DIL-ox-LDL (red) puncta in hVICs treated with DIL-Ox-LDL (10 µg/mL) for 4 hours. hVICs were cotransfected with PCSK9, ShNC, or ShLOX1. The bar graph shows the quantification of DIL-ox-LDL uptake in hVICs (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). I and J, Alizarin Red staining of mineralization nodules in calcified hVICs. hVICs were cotransfected with PCSK9, ShNC, or Sh LOX1. The bar graph shows the quantification of calcium deposition levels (n=6; 2-way ANOVA followed by the Tukey multiple-comparisons test). EV indicates empty vector; and OM, osteogenic medium.

Next, we investigated the mechanisms by which PCSK9 promotes calcification. Oxidized LDL (ox-LDL) is a recognized driver of aortic valve calcification through induction of VIC osteogenic differentiation, and its uptake is mediated in part by the scavenger receptor LOX-1 (lectin-like oxidized low-density lipoprotein receptor 1).^50,51 PCSK9 has been reported to upregulate LOX-1 expression, thereby enhancing ox-LDL uptake.⁵² Thus, we hypothesized that PCSK9 may promote VIC osteogenic differentiation by increasing LOX-1–dependent ox-LDL uptake. To test this, we collected aortic valve leaflets and hVICs from hCAVs and from noncalcified controls. Both the mRNA and protein levels of PCSK9 and LOX-1 were increased in aortic valve leaflets and hVICs from hCAVs (Figure 8D and 8E and Figure S11A–S11F). We further examined the expression of PCSK9 and LOX-1 during osteogenic differentiation of hVICs in vitro. Osteogenically differentiated hVICs exhibited higher expression of LOX-1 than controls (Figure 8F). In addition, PCSK9 knockdown reduced LOX-1 expression in hVICs, whereas PCSK9 overexpression led to an increase in LOX-1 expression (Figure 8F). These findings suggest that PCSK9 promotes the expression of LOX-1 in hVICs during osteogenic differentiation. We then examined whether PCSK9-induced LOX-1 enhances the ox-LDL intake of hVICs. PCSK9 overexpression increased ox-LDL intake in osteogenic differentiated hVICs, which was attenuated with LOX-1 knockdown (Figure 8G and 8H). These results demonstrate that PCSK9 increases the ox-LDL intake of hVICs by promoting the expression of LOX-1. We next explored the role of the PCSK9/LOX-1/ox-LDL axis in osteogenic differentiation of hVICs. PCSK9 overexpression enhanced OM-induced osteogenic responses in hVICs, whereas LOX-1 knockdown damped these effects (Figure 8I and 8J). Taken together, our results indicated that PRMT3-mediated arginine methylation stabilizes PCSK9; then, PCSK9 increases ox-LDL intake through LOX-1 in VICs, thereby facilitating VIC osteogenic differentiation (Figure S11G).

DISCUSSION

CAVD is a multifactorial disease with high spatiotemporal heterogeneity⁵³ that involves complex processes including changes in genetic factors, inflammation, lipoprotein deposition, and osteogenic transition of VICs.^54,55 There are no effective interventions that prevent CAVD progression.² Thus, the identification of new targets for treating CAVD is urgently needed. This is the first study to show that PRMT3, a type 1 protein arginine methyltransferase, plays a procalcific role in aortic valve calcification by stabilizing PCSK9. RUNX2 recruits the histone acetyltransferase P300 to induce PRMT3 transcription through H3K27ac in hVICs during aortic valve calcification. Inhibition of PRMT3 attenuates aortic valve calcification both in vitro and in vivo. We identified a novel interplay between arginine methylation and ubiquitination: PRMT3 catalyzes the asymmetric dimethylation of PCSK9 at R582, which suppresses CHIP-mediated ubiquitination at K575, thereby stabilizing PCSK9. PRMT3 exerted procalcific effects by upregulating PCSK9. Overall, our study defines the PRMT3-CHIP-PCSK9 axis as an important amplifier and therapeutic target in CAVD (Figure S12).

PRMTs have emerged as attractive therapeutic targets in various developmental and pathophysiologic processes,^56–59 but their roles in CAVD remain uncharacterized. For the first time, we demonstrated that elevated PRMT3 expression in calcified aortic valves is conserved across both mice and humans during CAVD. We further showed that PRMT3 expression strongly correlates with calcification markers and aortic stenosis severity, supporting its potential utility for CAVD diagnosis and risk stratification. Transcription factor RUNX2 is a well-established marker of calcification^8,60; however, its role in the transcriptional regulation of key procalcific factors remains incompletely defined. In this study, we identified PRMT3 as a direct transcriptional target of RUNX2. RUNX2 bound to the PRMT3 promoter and promoted its transcriptional activity in hCAVs. RUNX2 recruited the histone acetyltransferase P300 to transactivate the PRMT3 gene through H3K27ac. These findings reveal a novel mechanism by which RUNX2 drives PRMT3 upregulation during osteogenic differentiation in hVICs, amplifying the pathogenesis of CAVD.

In addition to evaluating the expression of PRMT3 and the mechanism by which PRMT3 expression is regulated, its role in the pathogenesis of CAVD was also confirmed. Prmt3 haploinsufficiency significantly alleviates HCD-induced aortic valve calcification in ApoE^−/− mice. In parallel, in vitro studies further confirmed the role of PRMT3 in promoting the osteogenic reprogramming of hVICs. Thus, in this study, we determined the therapeutic value of PRMT3 in CAVD. To our knowledge, this is the first report to establish a functionally relevant role of PRMT3 in CAVD, substantially advancing our understanding of protein arginine methylation in the disease. The degradation of key proteins through PROTACs provides a novel strategy for the treatment of human diseases.^61–64 A first‐in‐class PRMT3‐targeting PROTAC, compound 11, has been developed, which selectively induces PRMT3 degradation through recruitment of the MDM2 E3 ubiquitin ligase.¹⁸ Therefore, developing innovative PROTAC systems to treat CAVD is an attractive strategy. As expected, we found that compound 11 effectively induced PRMT3 protein degradation and significantly attenuated aortic valve calcification in vivo. Thus, future clinical studies evaluating compound 11 as a potential therapy for CAVD are warranted.

Previous studies have shown that PCSK9 is closely regulated by different types of PTMs, including N-glycosylation,¹² Tyr-sulfation,¹³ Ser-phosphorylation,¹⁴ and S-palmitoylation.¹⁵ However, the PTM-mediated molecular mechanisms that regulate PCSK9 expression in CAVD remain unexplored. Our study revealed that PCSK9 is a novel substrate of PRMT3 and that its methylation at R582 was essential for PRMT3-mediated procalcific effects. Although PCSK9 is known to undergo multiple PTMs that regulate its function,^12–15 this is the first study to identify arginine methylation as a novel PTM of PCSK9 and to delineate its physiologic relevance. Elucidating the mechanism by which PRMT3‐mediated methylation regulates PCSK9 fills a knowledge gap in understanding the roles of PCSK9 and PRMT3 in aortic valve calcification and may uncover novel therapeutic strategies for CAVD. Arginine methylation regulates the biologic functions of many proteins.^65,66 Indeed, a recent publication suggested that PRMT3 represses the innate immune response during viral infection by methylating cytosolic RNA and DNA sensors, such as RIG-I (retinoic acid-inducible gene-I), MDA5 (melanoma differentiation-associated protein 5), and cGAS (cyclic GMP-AMP synthase).³¹ Notably, our proteomic analysis also revealed multiple additional PRMT3-interacting proteins, which could be PRMT3 substrates. Because PCSK9 knockdown in PRMT3-overexpressing hVICs did not fully abrogate the pro-osteogenic effects of PRMT3, these findings suggest that additional PRMT3 substrates may contribute to the pathogenesis of CAVD. Future studies are needed to determine whether other PRMT3 substrates contribute to the functions of PRMT3 in CAVD.

Increasing evidence suggests that ubiquitination plays an important role in the pathogenesis of CAVD.^67–69 We showed for the first time that the E3 ubiquitin ligase CHIP bound to PCSK9, thereby directly catalyzing its ubiquitination and degradation in CAVD. Consistent with these findings, CHIP promotes the degradation of other proteins, including INSR (insulin receptor),⁷⁰ STX17 (Syntaxin 17),⁷¹ and SIRT6 (Sirtuin 6).⁷² Given the broad role of ubiquitination and the relatively limited number of E3 ligases, it is plausible that CHIP targets additional proteins that are involved in CAVD. Further investigations are warranted to elucidate whether CHIP-mediated ubiquitination contributes more broadly to the pathogenesis of CAVD. In addition, whether other E3 ubiquitin ligases target PCSK9 for ubiquitin-mediated degradation requires further exploration. Crosstalk between different PTMs has been increasingly recognized as a crucial modulator of various developmental and pathophysiologic processes.^73–75 For example, ZDHHC18 (zinc finger DHHC-type palmitoyltransferase 18) can palmitoylate MDH2 (malate dehydrogenase 2), preventing its ubiquitination and further increasing its protein stability.⁷⁶ The interaction between arginine symmetric dimethylation and ubiquitylation regulates KLF4 (Krüppel-like factor 4)–mediated genome stability and carcinogenesis.⁷⁷ In this study, we identified a novel interplay between asymmetric dimethylation and ubiquitylation in regulating PCSK9 expression in CAVD. We showed that PRMT3 directly interacted with PCSK9 and catalyzed the asymmetric dimethylation of R582 on PCSK9. This PTM is proposed to antagonize CHIP-mediated PCSK9 ubiquitylation at K575 and degradation, thereby stabilizing the PCSK9 protein and promoting CAVD progression. Crosstalk between protein modifications has been well characterized, especially with respect to histones and p53.⁷⁸ We do not know whether true crosstalk between arginine methylation and ubiquitination of PCSK9 occurs; that is, whether the ubiquitination of PCSK9 also interferes with PCSK9 modification by PRMT3. Future studies will also address structural changes caused by PCSK9 methylation that inhibit the binding of CHIP. The simulations presented herein focused on the mechanism underlying the recognition of PCSK9 by PRMT3 and the catalysis of PCSK9 arginine methylation.

PCSK9, which is highly expressed in adult hepatocytes, increases LDL cholesterol levels by promoting the degradation of the LDL receptor.⁷⁹ Patients with atherosclerosis benefit from decreases in LDL cholesterol levels due to treatment with PCSK9 inhibitors.^80,81 However, it is becoming apparent that the therapeutic effects of PCSK9 inhibition depend on more than just its capacity to reduce LDL levels.^82,83 The procalcific role of PCSK9 in CAVD has been revealed in recent studies.^7,8,84 The presence of lower LDL cholesterol levels observed in PCSK9 loss-of-function models does not fully account for its protection against CAVD.⁸⁵ In addition to AVCAPIR,⁸ our study revealed that PCSK9 also functions as a downstream effector of PRMT3 to mediate its procalcific effects. However, genetic depletion, pharmacologic inhibition, or PROTAC-mediated degradation of Prmt3 did not change the lipid profile of ApoE^−/− mice, suggesting that the lipid-lowering effects of PCSK9 inhibition are not involved in the alleviation of aortic valve calcification induced by PRMT3 deletion. We further demonstrated that PRMT3-mediated arginine methylation stabilizes PCSK9, which enhances LOX-1–dependent ox-LDL uptake in VICs, thereby promoting osteogenic differentiation and aortic valve calcification. Thus, our results, together with those of previous studies, suggest that, in addition to its lipid-regulating effects, PCSK9 is a promising therapeutic target for CAVD.

There are several limitations to this study. First, because of the limited access to normal human aortic valves, we used aortic valves from patients with dilated cardiomyopathy as the control group. However, the pathology of dilated cardiomyopathy may affect the gene expression of valve leaflets and constitute a nonnegligible confounder in our study. Second, the follow-up period was relatively brief and insufficient to show the lasting effect of PRMT3 knockdown on aortic valve calcification. Longer follow-up studies are needed to determine whether aortic valve calcification can occur at a later stage and confirm the effect of PRMT3 knockdown on cardiac remodeling. Third, the high hypercholesterolemia rate and other restrictions of the HCD-fed ApoE^−/− mouse model may preclude accurate simulation of the entire course of human CAVD. There are numerous widely used models with clear advantages, such as NOS3^−/−/Notch1^±,86 USP9X^fl/Y/Cre^+,87 and klotho-deficient mouse models.⁸⁸ Further research is needed to conclusively determine whether Prmt3 inhibition can ameliorate aortic valve calcification in these animal models. Fourth, our study identifies VIC-derived PCSK9, which is distinct from systemic (hepatic and circulating) PCSK9, as a key local amplifier of aortic valve calcification. Whether antibody-based PCSK9 inhibitors effectively neutralize intracellular or locally produced PCSK9 and attenuate LOX-1–mediated oxLDL uptake that promotes VIC osteogenic differentiation in vivo remains an important question for future study. Fifth, PCSK9 inhibition with antibodies consistently reduced circulating Lp(a) levels by up to 30%.⁸⁹ The mechanisms underlying PCSK9–Lp(a) interaction, and the potential impact of PRMT3-stabilized PCSK9 on Lp(a) levels, remain to be clarified. Sixth, emerging evidence indicates that dysregulated circular RNAs contribute to CAVD.⁹⁰ Whether circRNAs, such as circZBTB44, interact with PRMT3 during valve calcification warrants further investigation.

CONCLUSION

Our work demonstrates a procalcific role of PRMT3 and PCSK9 in regulating aortic valve calcification, primarily by amplifying osteogenic reprogramming in hVICs. Furthermore, we demonstrated that the beneficial effects of targeting PRMT3 in aortic valve calcification were associated with attenuated procalcific PCSK9 expression. Mechanistically, PRMT3 stabilized PCSK9 by catalyzing asymmetric dimethylation at R582, which in turn attenuated CHIP-mediated ubiquitination at K575 and degradation. These results may serve as a road map for future clinical studies to evaluate the latent therapeutic effects of PRMT3 inhibition in patients with CAVD. Overall, we identify a previously unrecognized posttranslational mechanism regulating PCSK9 stability in valve interstitial cells during CAVD and establish a link between PRMT3-mediated arginine methylation and valve-specific lipid–osteogenic coupling.

ARTICLE INFORMATION

Acknowledgments

The authors thank the medical subcenter of the Analytical and Testing Center, Huazhong University of Science and Technology, for technical support; and Linfang Yang, Sisi Li, and Fuyuan Cao, for expert assistance.

Sources of Funding

This work was supported by the National Natural Science Foundation of China (grants 82270380, 82422036, 82522036, 82271811, 82241217, 82470379, 82572044, and 825B2005), the China Postdoctoral Science Foundation (grants 2024T170310 and 2023M741285), the Natural Science Fund of Hubei Province (grants 2024AFA047 and 2025AFB477), and the National Key Research and Development Program of China (grant 2023YFC2706200).

Disclosures

None.

Supplemental Material

Methods

Figures S1–S12

Tables S1–S9

References 91–117

Supplementary Material

cir-153-1637-s001.pdf ^{(2.3MB, pdf)}

cir-153-1637-s002.pdf ^{(144.8KB, pdf)}

cir-153-1637-s003.pdf ^{(4.1MB, pdf)}