Reduced sox9 function promotes heart valve calcification phenotypes in vivo

Abstract

Rationale: Calcification of heart valve structures is the most common form of valvular disease and is characterized by the appearance of bone-like phenotypes within affected structures. Despite the clinical significance, the underlying etiology of disease onset and progression is largely unknown and valve replacement remains the most effective treatment. The SRY-related transcription factor Sox9 is expressed in developing and mature heart valves, and its function is required for expression of cartilage-associated proteins, similar to its role in chondrogenesis. In addition to cartilage-associated defects, mice with reduced sox9 function develop skeletal bone prematurely; however, the ability of sox9 deficiency to promote ectopic osteogenic phenotypes in heart valves has not been examined.

Objective: This study aims to determine the role of Sox9 in maintaining connective tissue homeostasis in mature heart valves using in vivo and in vitro approaches.

Methods and results: Using histological and molecular analyses, we report that, from 3 months of age, Sox9(fl/+);Col2a1-cre mice develop calcific lesions in heart valve leaflets associated with increased expression of bone-related genes and activation of inflammation and matrix remodeling processes. Consistently, ectopic calcification is also observed following direct knockdown of Sox9 in heart valves in vitro. Furthermore, we show that retinoic acid treatment in mature heart valves is sufficient to promote calcific processes in vitro, which can be attenuated by Sox9 overexpression.

Conclusions: This study provides insight into the molecular mechanisms of heart valve calcification and identifies reduced Sox9 function as a potential genetic basis for calcific valvular disease.

Figure 1. Sox9 expression is detected in developing and mature heart valve leaflets

Sox9 (red) is expressed in the mitral (mv) (A) and aortic valve (Ao) leaflets at E17.5 (arrows) (B). X-gal staining shows Col2a1-cre specific recombination (blue) in a subset of cells of the mitral (C) and aortic valves (D) in E18.5 Col2a1-cre × Rosa26R reporter mice. In Sox9^fl/+;Col2a1-cre mice (F), Sox9 expression (red) is reduced in recombined cells along the edge of the valve leaflets (arrows) compared to adjacent valve interstitial cells (arrowheads) not expressing the Col2a1-cre transgene. In control leaflets (E), cells with strong Sox9 expression are observed both at the leaflet surface and in the interstitium. Sox9 signal appears reduced in cells at the leaflet surface in Sox9^fl/+;Col2a1-cre mice (F). Pecam (green) indicates endothelial cells (E, F). Nuclei are stained with DAPI (A,B,E,F).

Figure 2. Von Kossa reactivity reveals increased calcium deposition in valve leaflets from Sox9^fl/+;Col2a1-cre mice

Von Kossa reactivity indicates calcium deposits (black, arrowheads) in aortic (Ao) (B) and mitral valve (mv) (D) leaflets adjacent to blood flow (red arrows) in tissue sections from Sox9^fl/+;Col2a1-cre mice at 12 months of age. Von Kossa reactivity was not significantly detected in littermate Sox9^fl/+ controls (A, C). Inset (D), no von Kossa treatment. Alcian blue counterstain defines the valve area (A–D), * indicates the base of the aortic valve, adjacent to the myocardium (A, B). Quantification of von Kossa reactivity as a percentage of valve area demonstrates significant increases in calcium deposits in Sox9^fl/+;Col2a1-cre mitral valve leaflets with a trend toward increasing severity and variability with age (E, F). m, months. Low magnification images (A–D) are available in Online Figure I.

Figure 3. Transcript levels of bone-related genes are increased in 12 month old Sox9^fl/+;Col2a1-cre mice

(A) Normalized fold changes in transcripts levels of osteogenic-signaling genes Runx2, Osteonectin (ON), Osteopontin (OP), Osteoprotegrin (OPG) and Smpd3 in Sox9^fl/+;Col2a1-cre versus Sox9^fl/+ control mice at 3, 6, and 12 months as determined by TaqMan Low Density Array (TLDA). *p<0.05. Osteonectin, undetected by in situ hybridization in control mitral valve leaflets (B), is detected in a 12 month Sox9^fl/+;Col2a1-cre mouse mitral valve leaflet (arrows, C). Low magnification images (B, C) are available in Online Figure II.

Figure 4. Inflammation and ECM remodeling-related genes are increased in 12 month old Sox9^fl/+;Col2a1-cre mice

(A) Inflammation-associated genes Vcam-1, Csfr1, and Tlr2 are increased inSox9^fl/+;Col2a1-cre mice relative to controls at 3, 6, and 12 months (TLDA analysis). (B) Increases in ECM remodeling and fibrosis-related genes Col1a1, Col1a2, Timp1, and Tenascin-C (Ten-C) are also observed. *p<0.05 relative to age-matched controls, #p<0.05 relative to 3 month Sox9^fl/+;Col2a1-cre mice. Elastin (green) is detected on the atrial surface of the mitral valve in control Sox9^fl/+ mice (arrows, C). Elastin fibers are fragmented and disorganized in Sox9^fl/+;Col2a1-cre mitral valve leaflets (arrows, D).

Figure 5. Sox9 knockdown increases calcification phenotypes in mouse valve explants

(A, B) von Kossa reactivity in Sox9^fl/fl neonatal mouse valve explants infected with GFP-adenovirus (Ad-GFP) (A) or Cre-adenovirus (Ad-Cre) (B). (C) Quantification of von Kossa reactivity in Ad-GFP and Ad-Cre infected explants as a percentage of total area defined by Alcian blue. (D) Sox9 transcript levels are significantly decreased following infection with Ad-Cre, while Runx2 and Osteopontin (OP) are increased relative to GFP-treated controls. *p>0.05.

Figure 6. Retinoic acid treatment reduces Sox9 expression and promotes matrix mineralization

Cultured E10 chick mitral valve explants were treated with DMSO (A, C, E) or RA (B, D, F) and subject to infection with Ad-GFP (C, D) or Ad-Sox9 (E, F), or cultured without virus (A, B). (G) Quantitation of von Kossa reactivity in treated explants, normalized to area (Alcian blue staining). *p<0.05 compared to relative DMSO control, #p<0.05 compared to RA and Ad-GFP infection. (H) Normalized fold change in endogenous chicken Sox9 (cSox9) and adenoviral mouse Sox9 (Ad mSox9). *p<0.05 compared to respective DMSO- or RA-treated controls. #p<0.05 compared to DMSO+Ad-GFP.