Differential BMP Signaling Mediates the Interplay Between Genetics and Leaflet Numbers in Aortic Valve Calcification

Visual Abstract

There is currently no medical therapy to prevent or reverse CAVD. 2 The absence of appropriate animal models that mimic human CAVD is a major limitation to understanding CAVD pathophysiology and the development of effective medical therapy. 3 Bicuspid aortic valve (BAV) is a major risk factor for CAVD, and patients with BAV have earlier, more aggressive disease progression compared with tricuspid aortic valve (TAV). Interestingly, relatives of patients with familial BAV who have a TAV are at increased risk for developing CAVD, pointing to the presence of common factor(s) that predispose to BAV development and CAVD in both BAV and TAV. 4,5 The pathogenesis of CAVD is complex and involves the differentiation of quiescent fibroblast-like valvular interstitial cells (VICs) to myofibroblast and osteoblast-like interstitial cells. Like in bone mineralization, bone morphogenic protein (BMP), and more specifically, BMP2 signaling is implicated in valvular calcification. However, existing evidence indicates that BMP2 is insufficient by itself and additional factors are required to induce aortic valve calcification. 6 It is possible that local changes in blood flow in BAV leads to biomechanical alterations that contribute to the accelerated progression of CAVD. However, in the absence of representative animal models, the molecular mechanisms of the interplay between the leaflet numbers and genetic factors that predispose to CAVD remain unclear. 3 Discoidin, CUB and LCCL Domain Containing 2 (DCBLD2, also known as endothelial and smooth muscle cell-derived neuropilin-like protein [ESDN]) is a transmembrane protein 7,8 implicated in the regulation of growth factor signaling, vascular remodeling, and angiogenesis. 9-13 A recent human aortic valve gene profiling study suggested that DCBLD2 transcripts are reduced in stenotic aortic valves. 14,15 In the course of our studies aimed at evaluating the role of DCBLD2 in vascular remodeling, we noticed the presence of BAV with thickened leaflets in a subset of Dcbld2 À/À mice. This led us to investigate the potential role of DCBLD2 in aortic valve pathology and to take advantage of this model to address how the interplay between the genetic background and leaflet numbers determines aortic valve calcification and stenosis. Here, we show that DCBLD2 expression is reduced in human CAVD, and Dcbld2 À/À mice develop BAV and CAVD with typical features of human disease. In vitro and in vivo studies link the more prominent valvular calcification of BAV to BMP2 signaling, which despite similar levels of Bmp2 expression between bicuspid and tricuspid valves, is enhanced in BAV. ANIMAL MODELS. The generation of Dcbld2 À/À mice on a C57BL/6 background and endothelial-specific conditional knockout mice (Cdh5-Cre/Dcbld2 fl/À ) were reported previously. 10   To explore the role of DCBLD2 in aortic valve pathobiology, we first evaluated its expression in wild-type (WT) and Dcbld2 À/À murine aortic valves.
Like in human aortic valves, in the aortic valves of WT mice, DCBLD2 was detectable by immunostaining throughout the leaflet, including CD31-positive ECs ( Figure 1C, Supplemental Figure 1C). As expected, Dcbld2 À/À aortic valves did not express any DCBLD2 ( Figure 1C). Importantly, about 50% of Dcbld2 À/À mice of both sexes were noted to have BAV ( Figures 1D and   1E). The presence of BAV was not associated with any unrelated structural cardiac abnormality. Echocardiography in 9-to 12-month-old animals showed no difference in left ventricular ejection fraction or outflow tract velocity between Dcbld2 À/À BAV, Dcbld2 À/À TAV, and WT mice. However, the left ventricular mass was significantly higher in Dcbld2 À/À To investigate the development of CAVD in Dcbld2 À/À mice, we assessed the key features of CAVD, namely, valvular fibrosis and calcification in 9to 12-month-old mice in comparison with agematched WT animals. Dcbld2 À/À BAV leaflets were significantly thicker than WT and Dcbld2 À/À TAV leaflets (WT: 26.5 AE 3.8 mm; Dcbld2 À/À TAV: 34.3 AE 12.1 mm; Dcbld2 À/À BAV: 100.6 AE 41.6 mm, n ¼ 5 to 9; P < 0.01 for BAV vs WT and P < 0.05 for BAV vs TAV) ( Figures 1I and 1J). Interestingly, not all BAV leaflets were thickened, and a small subset of animals with TAV had leaflet thickening ( Figure 1J). Masson's trichrome staining suggested that fibrosis contributes to leaflet thickening in Dcbld2 À/À mice (Supplemental Figure 2). Indeed, the collagen type I content, as detected by immunostaining, was significantly higher in Dcbld2 À/À BAV compared with WT leaflets (Figures 2A and 2B). The Dcbld2 À/À TAV leaflets without thickening contained less collagen with a patchy distribution.
Alizarin red staining of aortic valves showed multiple calcified nodules in Dcbld2 À/À BAV leaflets, especially in the hinge area similar to human CAVD, that spared the aortic sinus. No calcification was detected in WT aortic valves. Quantification of Alizarin red staining showed significantly higher

BMP INHIBITION AND VALVULAR CALCIFICATION IN
BAV. Next, we sought to address to what extent this activation of BMP signaling is responsible for enhanced valvular calcification in BAV. As a prelude to in vivo studies, we assessed the effect of LDN-193189, a type 1 BMP receptor inhibitor, 23 on SMAD1/ 5/9 phosphorylation in pVIC. LDN-193189 suppressed SMAD1/5/9 phosphorylation in both serum-exposed control cells (P < 0.05) and following siRNAmediated DCBLD2 down-regulation (P < 0.001), which significantly enhances SMAD1/5/9 phosphorylation (P < 0.001) (Figures 5E and 5F). Next, a group of 1-year-old Dcbld2 À/À mice were treated with LDN-193189 (6 mg/kg/day, intraperitoneal) for 1 week, and the effect of the inhibitor on BMP signaling was evaluated by immunostaining and RT-PCR. On postmortem tissue analysis, LDN-193189 significantly reduced aortic valve SMAD1/5/9 phosphorylation (P < 0.001) (Figures 5G and 5H) and osteocalcin expression (P < 0.001) ( Figure 5I) in animals with BAV to levels seen in WT and Dcbld2 À/À mice with TAV ( Figure 5D). In animals with TAV, LDN-193189 had no effect on low levels of osteocalcin observed in nontreated animals (Supplemental Figure 7). Combined, these data indicate that BMP signaling is enhanced in valvular calcification of animals with BAV.

DISCUSSION
We established that DCBLD2 is down-regulated in human CAVD and identified a high incidence of BAV in Dcbld2 À/À mice. Taking advantage of this high incidence, here we show that 1) DCBLD2 deficiency promotes the development of CAVD and aortic stenosis, which is more severe in animals with BAV, 2) the difference between Dcbld2 À/À bicuspid and tricuspid valves is especially striking with regard to valvular calcification, which like in humans, affects the leaflets in these animals, and 3) BMP signaling is DCBLD2 is a member of the DCBLD family of transmembrane proteins. 26 Along with its less characterized homolog, DCBLD1, the 715-amino acid DCBLD2 has a structure consisting of a signal sequence, followed by CUB, LCCL, and coagulation factor V/VIII type-C (also discoidin) domains. This resembles the domain structure of neuropilins, which possess 2 CUB and discoidin domains. Recent studies have implicated DCBLD2 in the regulation of growth factor signaling. Accordingly, DCBLD2 deficiency promotes platelet-derived growth factor and insulin signaling in vascular smooth muscle cells, and inhibits vascular endothelial growth factor signaling in EC. 8,10,12,13 In addition, DCBLD2 regulates epidermal growth factor-induced tumorigenesis. 11 Our study extends the scope of those observations to demonstrate that in addition to regulating growth factor signaling in vascular and cancer cells, DCBLD2 modulates BMP2 expression and signaling in EC and VIC.
This maybe through regulation of BMP receptor expression or phosphorylation, as shown previously for a number of growth factor receptors. 10,12,13 BMP2 is a key osteogenic differentiation factor in vascular and valvular calcification. 6,27 Canonical BMP2 signaling involves the binding of the ligand to a heterotetrameric complex of transmembrane type I and II serine/threonine kinase receptors, which triggers SMAD 1/5/9 phosphorylation and nuclear translocation to regulate gene expression. 28 Previous in vitro studies had suggested that BMP signaling is necessary, but not sufficient, for VIC calcification, which requires a pro-osteogenic milieu. 6 Our study confirmed the requirement for BMP signaling in VIC calcification in vitro, as Noggin, a BMP-specific inhibitor, 18   TRANSLATIONAL OUTLOOK 1: The Dcbld2 À/À mouse is a clinically relevant model to study the pathophysiology of CAVD. The absence of appropriate animal models of CAVD makes the Dcbld2 À/À model a powerful tool to study the pathophysiology of CAVD.

FUNDING SUPPORT AND AUTHOR DISCLOSURES
The Dcbld2 À/À model also provides the opportunity to develop innovative therapies for CAVD, a disease for which there is no effective medical therapy.
TRANSLATIONAL OUTLOOK 2: Our data support the notion that BMP2 inhibition is a therapeutic target in CAVD, and modulation of DCBLD2 may be a novel pathway to prevent CAVD.