Rac1 mediates Cadherin-11 induced cellular pathogenic processes in aortic valve calcification

Abstract

Background

Calcific aortic valve disease (CAVD), a major cause for surgical aortic valve replacement, currently lacks available pharmacological treatments. Cadherin-11 (Cad11), a promising therapeutic target, promotes aortic valve calcification in vivo, but direct Cad11 inhibition in clinical trials has been unsuccessful. Targeting of downstream Cad11 effectors instead may be clinically useful; however, the downstream effectors that mediate Cad11-induced aortic valve cellular pathogenesis have not been investigated.

Approach and Results

Immunofluorescence of calcified human aortic valves revealed that GTP-Rac1 is highly upregulated in calcified leaflets and is 2.15 times more co-localized with Cad11 in calcified valves than GTP-RhoA. Using dominant negative mutants in porcine aortic valve interstitial cells (PAVICs), we show that Cad11 predominantly regulates Runx2 nuclear localization via Rac1. Rac1-GEF inhibition via NSC23766 effectively reduces calcification in ex vivo porcine aortic valve leaflets treated with osteogenic media by 2.8-fold and also prevents Cad11-induced cell migration, compaction, and calcification in PAVICs. GTP-Rac1 and Trio, a known Cad11 binding partner and Rac1-GEF, are significantly upregulated in Nfatc1Cre;R26-Cad11Tg/Tg (Cad11 OX) mice that conditionally overexpress Cad11 in the heart valves by 3.1-fold and 6.3-fold, respectively. Finally, we found that the Trio-specific Rac1-GEF inhibitor, ITX3, effectively prevents Cad11-induced calcification and Runx2 induction in osteogenic conditions.

Conclusion

Here we show that Cad11 induces many cellular pathogenic processes via Rac1 and that Rac1 inhibition effectively prevents many Cad11-induced aortic disease phenotypes. These findings highlight the therapeutic potential of blocking Rac1-GEFs in CAVD.

Graphical Abstract

1. Introduction

Valvular disease is an increasingly prevalent, debilitating condition that is currently estimated to affect 2.5% of the population[1]. Aortic valve disorders were associated with 34,408 deaths in 2014¹ and responsible for 26,414 transcatheter aortic valve replacement, commonly called TAVR, procedures alone between the years of 2011-2014[2]. Taken together, the clinical data reflect how aortic valve disorders pose a major economic burden and are difficult to treat at the end-stages. This burden is exacerbated by the lack of biomarkers for aortic valve disorders forcing diagnosis only at the late to end-stages via echocardiography once significant aortic valve remodeling and pathogenesis have already occurred.

Calcific aortic stenosis is the major cause for aortic valve replacement in the U.S. as well as Europe[3]. Calcific aortic valve disease (CAVD) is characterized by gradual aortic valve leaflet calcification and thickening, which lead to aortic valve stenosis and insufficiency[4-6]. CAVD was previously thought to occur due to the passive, natural deterioration of the aortic valve. However, CAVD is an active disease process in which fibroblasts transition into pathological phenotypes and contribute to inflammation, fibrosis and calcification of the aortic valve leaflets[4]. The most current research focuses on a systematic omics approaches to uncover the most relevant molecular mechanisms that drive disease[7-9]. However, despite the progress in studies of the mechanisms that regulate CAVD, useful molecular targets for the treatment of CAVD remain elusive. The development of CAVD pharmacological therapies will improve patient outcomes and reduce the economic burden created by invasive treatments.

CAVD is thought to occur in a series of early-stage cellular pathogenic events during which the disease is generally asymptomatic followed by a series of late-stage macroscopic changes resulting in valve thickening, stiffening and stenosis[4]. In echocardiography, these late-stage changes manifest as increased ejection velocity and regurgitation[4]. However, the goal of future pharmacological therapy is to slow the progress of early-stage disease before the damage to the valve becomes irreversible. Early stage cellular pathogenesis involves activation of fibroblasts to myofibroblasts that secrete type I collagen[10]. In wound healing, myofibroblasts are important for stimulating cell migration and compaction for contraction of the wound, after which they apoptose[11]. However, in fibrotic disease, myofibroblasts persist and initiate major remodeling of the valve matrix and increase valve stiffness[4,11]. Therefore, it is vital that pharmacological inhibitors can prevent the signaling processes that induce cell migration and compaction of these activated fibroblasts.

In the past, only a few studies had investigated changes in gene expression that occur in calcified human valve leaflets. In 2001, the largest study using human CAVD samples evaluated the histopathological presence of fully differentiated osteogenic cell types including osteoblasts/osteoclasts and detected these cells macroscopically in 13% of calcified samples[12]. Only BMP2/4 protein expression was evaluated and found to be increased[12]. However, it became increasingly clear that activated fibroblasts may initiate osteogenic expression programs to induce calcification whether or not the cells themselves become fully differentiated[13]. This was later evaluated by another large human study evaluating specific osteoblastic markers via RT-PCR for osteopontin, bone sialoprotein, osteocalcin, alkaline phosphatase, and the osteoblast-specific transcription factor Cbfa1 also known as Runx2 in calcified human aortic valve leaflets[14]. This study showed that human aortic valve calcification was associated with increased expression of osteopontin, bone sialoprotein, osteocalcin, and Runx2. Alkaline phosphatase did not have increased mRNA levels but had increased protein expression in the tissue. These findings illustrated that CAVD is not a degenerative process and actually involves active bone formation/remodeling via initiation of osteogenic gene programs. The recategorization of CAVD as an active disease process is generally accepted in the current literature[4]. Following these findings, studies have shown that a subset of these activated fibroblasts can differentiate into osteoblast-like cells and initiate the valve calcification process[4,13]. Here we focus on the osteoblast-specific differentiation transcription factor, Runx2, as the primary marker of active bone formation.

Cadherin-11 (Cad11), also known as OB-cadherin (osteoblastic), is a cell adhesion protein that regulates many of the cellular and tissue pathogenic processes that promote CAVD. Cad11 has been shown to promote calcification in porcine aortic valve interstitial cells (PAVICs) in vitro using two different models of aortic valve pathogenesis: TGF-beta induced calcification and osteogenic media (OGM) induced calcification[15,16]. Cad11 overexpression also induces many of the cellular processes that promote aortic valve pathogenesis. This includes promoting cell migration, matrix compaction, as well as myofibroblastic and osteogenic gene expression[16]. In vivo, conditional Cad11 overexpression in murine aortic valves induces calcification, extracellular matrix remodeling, as well as aortic regurgitation[16]. Together, the data strongly suggest Cad11 as a promising, potential therapeutic target for CAVD.

Due to the significant mediating role of Cad11 in aortic valve pathogenesis, direct Cad11 targeting via small molecule Cad11 inhibitors or blocking antibodies have been considered for CAVD therapy. SYN0012, a murine anti-Cad11 blocking antibody, effectively reduces aortic valve leaflet hyperplasia, calcification, inflammatory signals and tissue stiffening in vivo induced by Notch haploinsufficiency[17]. This antibody had already been in clinical trials for rheumatoid arthritis, a disease that is also induced by Cad11-mediated pathogenesis. Unfortunately, the human equivalent RG6125 has recently been pulled following phase II clinical trials for rheumatoid arthritis as it was found to show little to no difference in efficacy compared to the placebo treatment[18]. It is currently unclear whether the antibody will play a significant role in CAVD treatment. The search for effective CAVD therapies is therefore still a significant area of interest. Due to the significant role that gradual tissue pathogenesis plays in the progression of CAVD, investigation of the efficacy of therapeutic treatments will require validation of targeting on a molecular and cellular level.

Understanding the cellular mechanisms that regulate CAVD is important for developing compounds for effective pharmacological targeting of pathways that maintain CAVD pathogenesis. Cad11 promotes many pathogenic cellular phenotypes including increased calcific nodule formation, matrix deposition, myofibroblastic and osteogenic gene expression, cell migration and compaction[16,19]. Cad11 is known to promote small GTPase activity such as RhoA and Rac1 in neural crest cell migration[20]. We previously showed that the RhoA/ROCK pathway is a driver of in vitro calcification downstream of Cad11 overexpression in PAVICs [16]. Rac1 has not yet been investigated in calcific aortic valve disease; however, it has been shown to potentially play a role in diseases of similar pathophysiology[21,22]. Therefore, in this study, we investigate Cad11 downstream signaling to identify an intracellular means of targeting Cad11-induced cellular and tissue pathogenesis.

2. Methods

2.1. Procurement, Histology and Immunofluorescence of Human and Mouse Aortic Valves

Wildtype (WT) Nfatc1^Cre; R26-Cad-11^+/+ and valve-specific Cad-11 overexpressing Nfatc1^Cre; R26-Cad-11^Tg/Tg (Cad11 OX) mice were genotyped and handled as previously described[16]. Animal work was conducted according to the relevant national and international guidelines. The details of this study were reviewed and approved by the Cornell IACUC (Protocol #2008-0011). Calcified human aortic valves (CHAV) were selected, obtained and processed as previously described [16]. All samples were obtained with informed consent from patients. All procedures were approved by the Institutional Review Boards at Cornell University, NY Presbyterian Hospital, and Robert Packer Hospital. Adult mouse hearts were fixed, processed and sectioned as previously described[16].

Human and Mouse sections were then processed and analyzed for immunohistochemistry with the same secondary antibodies and DRAQ-5 nuclear stain as previously described[16]. Primary antibodies (1:500 dilution) used included Cadherin-11 (rabbit, Invitrogen), GTP-Rac1 (mouse, NewEast Biosciences), GTP-RhoA (mouse, NewEast Biosciences), β-catenin (rabbit, Cell Signaling Technologies), and Trio (rabbit, Santa Cruz Biotechnology). Zeiss 710 confocal microscopy (Cornell University Life Sciences Core Laboratories Center) were used for signal detection and image collection. Immunoreactivity of proteins stained in tissue sections was measured using ImageJ as previously described[16]. Quantification involved normalization of fluorescent intensity to the total area of the valve. Experimental samples were then normalized to controls as demonstrated previously[23,24]. Nuclear localization was measured using NIH ImageJ software using the ImageJ plugin Colocalization for the protein of interest and DRAQ5-nuclear staining[24]. Channels with the protein of interest and nuclei were each thresholded and co-localized regions were measured using the ImageJ plugin Colocalization[16] GTP-Rac1/GTP-RhoA co-localization and Cad11/Trio were measured using the ImageJ plugin Colocalization.

2.2. Procurement and Histology of ex vivo porcine aortic valve leaflets.

Porcine aortic valve leaflets were harvested, transported, and cultured as previously described[25]. Valves were incubated for 11 days in regular growth media, after which they were changed to osteogenic differentiation medium (control medium with 10 mmol/L β-glycerophosphate, 50μg/mL ascorbic acid, and 10 nmol/L dexamethasone). 25μM NSC23766 (2161, Tocris) were added to aliquots of sterile OGM and used to treat a subset of ex vivo valves for 10 days. Media was changed every other day. After 10 days, valve leaflets were fixed in 4% paraformaldehyde (PFA) overnight at 4°C, dehydrated through an ethanol series, paraffin embedded and sectioned at 8μm thickness. Sections were mounted on slides and stained with Alizarin Red stains as previously described[16,25].

2.3. PAVIC Plasmid Transfection and Calcification Analysis

Porcine aortic valve cells were isolated and subsequently cultured as previously described[26-28]. PAVICs were used between passages 2-8. For calcification, mRNA, and protein analysis, cells were seeded in 24-well plates at 0.05×10⁶ cells/well, and after incubation for 24 hours, were transfected with either pmax-GFP empty vector (VDF-1012, Amaxa Biosystems) or a Cad11 Human cDNA ORF Clone (NM_001797, Origene) using X-tremeGENE9 Transfection Reagent (Roche). Then cells were left for 24 hours in minimal medium supplemented with DMEM, 10% FBS, and without antibiotics as previously described[16]. Then cells were supplemented with osteogenic growth media (OGM, control medium with 10 mmol/L β-glycerophosphate, 50μg/mL ascorbic acid, and 10 nmol/L dexamethasone as previously described[16]). A subset of Cad11 OX cells was additionally treated with 10μM Rac1 inhibitor, NSC23766 (2161, Tocris) or 10μM of the Trio-specific Rac1-GEF inhibitor, ITX3 (I1411, MilliporeSigma). Cells were cultured for 10 days with the media changed every 48 hours. For regular media calcification, the same method as above was repeated but regular growth media was used instead of osteogenic growth media. Cells were fixed with 4% PFA and were incubated in 40nmol/L Alizarin Red S (ARS) dye for 20 minutes and then washed with 1X PBS to identify calcific nodules. Alizarin Red absorbance was detected as previously described[29]. Nodules were counted using ImageJ software as previously described and normalized to the control[16]. Transfection with CA-Rac1 (Q61L, Addgene), dn-Rac1 (T17N, Addgene), dn-RhoA (T19N, Addgene) and pmax control PAVICs was conducted using the same method as described above. Cells were fixed 48 hours after transfection for cell immunofluorescence. PAVICs were transfected with Cad11 protein domain deletion variants using the same method described above and treated with OGM for 10 days prior to fixation and Alizarin Red staining.

2.4. PAVIC mRNA Analysis

RNA was extracted using the RNeasy Mini Kit (Qiagen, Valencia, CA) and reverse transcribed into cDNA using the qScript cDNA SuperMix kit (Quantabio) using manufacturer’s instructions. Cad-11 overexpression were confirmed real-time PCR (qPCR) with custom human-specific primers (Table SI) and SYBR Green PCR master mix as previously described [16]. Only mRNA analysis was log transformed in order to show data points reasonably on using the same graphing scale.

2.5. PAVIC Cell Immunofluorescence

For cell immunofluorescence, PAVICs were seeded at 0.05 x 10⁶ cells/well in 35mm uncoated dishes with 10mm microwell and No. 1.5 coverslips and were transfected with pmax-GFP, Cad11, dn-Rac1, dn-RhoA, and CA-Rac1 plasmids as described above. These cells were also incubated with DMEM supplemented with 10% FBS, and 1% penicillin/streptomycin. After fixation with 4% PFA overnight, wells were washed 3 times with 1X PBS and rocked for 15 minutes in between. Samples were then permeabilized with 0.2% Triton-X 100 (VWR International, West Chester, PA) for 10 minutes, and washed another 3 times with 1X PBS. Samples were incubated in 10% goat serum diluted in 1% BSA for 30 minutes and stained for GTP-Rac1 (mouse, NewEast Biosciences), GTP-RhoA (mouse, NewEast Biosciences), β-catenin (rabbit, Cell Signaling Technologies) or Runx2 (mouse, Abcam) and DRAQ5 at a 1:200 dilution. Images were obtained with Zeiss 710 confocal microscopy and immunofluorescence analysis was conducted as described above.

2.6. PAVIC Western Blotting

Cells were trypsinized and spun down to a cell pellet and washed three times with 1X PBS post-experiment. 2X Laemmmli Sample Buffer (Bio-Rad Laboratories) was diluted 1:1 in DepC water and β-mercaptoethanol was added at 5% of final buffer volume. After, cell pellets were diluted in Laemmmli Sample Buffer Solution. After loading protein into 4-15% Mini-PROTEAN Precast gels (Bio-Rad Laboratories), gels were run for 20 minutes at 120V in 25mM Tris, 0.2M Glycine, 1% SDS running buffer. The Trans-blot Turbo Transfer System (Bio-Rad Laboratories) was used for western blot transfer to nitrocellulose membrane (Thermo Scientific, Rockford, IL) which was performed for 10 minutes at 25 Volts in Trans-Blot Turbo 5x Transfer Buffer (Bio-Rad Laboratories. Membrane was stained with Ponceau Red to ensure even transfer of protein and then blocked for 1 hour in Odyssey Blocking Buffer at room temperature, which was diluted 1:1 in 1X PBS with 0.1% Tween-20 (PBST). Runx2 (mouse, Abcam), αSMA (rabbit, Abcam, dilution 1:5000), and GAPDH (mouse, Abcam, dilution 1:5000) were used to detect protein expression in 1:1 Odyssey buffer and PBST overnight at 4°C. Immunoblots were then washed 4 times for at least 15-minute intervals in PBST. They were then incubated with 1:10,000 dilutions of anti-mouse and anti-rabbit secondary antibodies (Li-Cor IRDye) for 1 hour at room temperature. Blots were washed 4 times for at least 15-minute intervals and subsequently imaged with the Odyssey Infrared system (Li-Cor).

2.7. PAVIC Compaction, Migration Assays

3D collagen hydrogels were created, imaged and analyzed as previously described[16] For migration analysis, PAVICs were seeded at 0.05×10⁶ cells in 24-well plates and scratch wound assay was conducted as previously described[16].

2.8. Cad11 Protein Domain Deletion Variants

Cad11 deletion variants were constructed from full-length Cad11 Human cDNA ORF Clone (NM_001797, Origene) using restriction enzyme cutting site modification using previously described methods[20]. Primers were designed for amplification of the desired region with BamHI, XhoI, or KpnI restriction enzyme recognition sequences added on to the ends of the primer sequence, without changing the amino acid sequence in synthesized oligonucleotides. The PCR products were digested with BamHI/XhoI, BamHI/KpnI or KpnI/XhoI double digests and were then ligated into the same pmax-GFP vector plasmid for transfection into cells. All insert sequences were confirmed by DNA sequencing. Cad11 deletion variants expression was using qRT-PCR 24 hours after transfection. PAVIC seeding, transfection, calcification and immunofluorescence was conducted as described above.

3. Theory/Calculations

3.1. Quantification of Histological Changes and Statistical Analysis

All data are reported as means (n≥3 independent experiments per condition) with error bars representing standard error of the mean (SEM) as previously described[16]. All statistical analysis was performed on RStudio Version 1.0.153 or Graphpad PRISM 8. Statistical significance was determined using the Student’s t-test, or one-way ANOVA with Tukey’s post hoc paired tests for treatment effects. Outliers which were defined as data points outside 1.5x margins of the interquartile range were excluded from analysis. ANOVA assumptions were tested with Bartlett’s/Levene’s test and the Shapiro-Wilks test. If assumptions were not met, Welch’s ANOVA/Games-Howell or Kruskal-Wallis/Dunn’s Test were used. Differences between means were considered significant at p<0.05.

4. Results

4.1. GTP-Rac1 is highly upregulated and co-localized with Cad11 in calcified human aortic valve leaflets

In order to test the relevance of Rac1 in Cad11-induced calcification, GTP-Rac1 levels were assessed in control and calcified human aortic valve leaflets. Immunofluorescence staining revealed a 1.9-fold increase in GTP-Rac1 expression in calcified human valve leaflets relative to the control (p=0.035) (Fig. 1, A-C). Furthermore, matched immunofluorescence samples illustrated that GTP-Rac1 is 2.15 times more co-localized with Cad11 expression than GTP-RhoA is co-localized with Cad11 (p=0.049) (Fig. 1 D-L). This data not only suggests that Rac1 may play an active role in calcification but also that Rac1 may be more closely tied to Cad11-induced pathogenic processes in the aortic valve than RhoA.

Figure 1. GTP-Rac1 is upregulated in calcified human aortic valves and highly colocalized with Cad11.

A-B, Immunofluorescence of GTP-Rac1 in control and the fibrosa of calcified human aortic valves. Scale bar = 100μm. C, GTP-Rac1 expression control and calcified human aortic valves normalized to the Control. (CHAV, n = 5. Control, n=3). D-K, Immunofluorescence of matched calcified human aortic valve (CHAV) leaflets for colocalization of Cad11 with GTP-Rac1 and GTP-RhoA. Colocalization with Cad11 (coloc) is shown in white in panels G, K. Scale bar = 100μm. L, Relative colocalization Cad11 with GTP-Rac1 and GTP-RhoA. (CHAV, n=4. Control n=3). *p<0.05.

4.2. Osteogenic marker induction via Cad11 is primarily Rac1-dependent

In order to assess the relative importance of Rac1 and RhoA in Cad11-induced calcification, we performed loss of function experiments in porcine aortic valve interstitial cells (PAVICs) co-overexpressing Cad11 (Cad11 OX) with the dominant negative form of each protein independently (dnRac1 and dnRhoA). Cad11 activity was confirmed to be significantly elevated in all PAVICs transfected with a Cad11 plasmid (p<0.0005). While Cad11 expression significantly increased Runx2 expression and nuclear localization by 5.1% which is a 16.9 fold change(p<0.0005), co-transfection of Cad11 with dnRhoA still produced a significant amount of Runx2 nuclear localization relative to the control, which was elevated by 3.8% (p<0.0005). However, co-transfection of Cad11 with dnRac1 significantly reduced Cad11-induced Runx2 nuclear localization to 1.2% (Fig. 2, A-F). The Cad11 OX + dnRac1 condition was significant relative to the Cad11 OX condition (p<0.0001) but not significant to the Control condition (p=0.08). This suggests Cad11-induced Runx2 nuclear localization is primarily Rac1-dependent. Furthermore, transfection of PAVICs with a constitutively-active Rac1 plasmid upregulated Runx2 expression and induced Runx2 nuclear localization by 3.9% (p=0.01) which is an 18.1 fold change (Fig. 2, G-I), suggesting that Rac1 activity is sufficient to induce Runx2 nuclear localization.

Figure 2. Cad11-induced Runx2 nuclear localization is primarily Rac1-dependent.

A-D, Runx2 expression in Control, Cad11 OX, Cad11OX + dominant negative Rac1 (dnRac1), and Cad11OX + dominant negative RhoA (dnRhoA) conditions 48 hours after transfection. Scale bar = 250μm. White box denotes area that is magnified for visualization. Nuclear Runx2 is indicated by white arrows. E, %Nuclear Area positive for Runx2 in Control, Cad11 OX, Cad11OX + dnRac1, and Cad11OX + dnRhoA conditions, n≥4. F, Cad11 expression for Control, Cad11 OX, Cad11OX + dnRac1, and Cad11OX + dnRhoA conditions normalized to the Control, n≥4. G-H, Runx2 expression in Control PAVICs and PAVICs transfected with constitutively-active Rac1 (CA-Rac1). Scale bar = 100μm. I, %Nuclear Area positive for Runx2 for Control and CA-Rac1 conditions, n=4. *p<0.05, **p<0.005, ***p<0.0005.

4.3. The Rac1-GEF inhibitor, NSC23766, effectively prevents Cad11-induced cellular pathogenesis in ex vivo porcine aortic valve leaflets and PAVICs

Immunofluorescence staining of 10-month Nfatc1^cre; R26-Cad11^+/+ (WT) and Nfatc1^cre; R26-Cad11^tg/tg (Cad11 OX) murine aortic valves revealed Rac1-GTP was upregulated 3-fold in Cad11 OX valves, suggesting that Rac1 activation occurs downstream of Cad11 (p=0.01) (Fig. 3, A-C). The widely used NSC23766 compound is an effective Rac1 inhibitor targeting the two major Rac1-GEFs, Trio and Tiam1[30]. Trio operates downstream of Cad11 and binds directly to the intermembrane domain (IMD) on the cytoplasmic side of Cad11[20]. Direct links between Tiam1 and Cad11 have not been established; however, Tiam1 frequently operates downstream of growth factors/signals and is a mediator for inducing cell growth, migration, adhesion, etc. via Rac1[31,32]. Here we use the widely-used Rac1-GEF inhibitor, NSC23766, to effectively block all major sources of Rac1 activation under valve pathogenic conditions.

Figure 3. Treatment of ex vivo porcine aortic valve leaflets with NSC23766 reduces calcification.

A-B, GTP-Rac1 expression in the aortic valves of 10 month Nfatc1^cre; R26-Cad11^+/+ (WT) and Nfatc1^cre; R26-Cad11^tg/tg (Cad11 OX) aortic valves. Scale bar = 50μm. C, GTP-Rac1 expression in control and Cad11 OX aortic valves normalized to the Control, n=6. D-E, Alizarin Red stain of ex vivo porcine aortic valve leaflets cultured in osteogenic media (OGM) and treated with NSC23766 (25μM) for 10 days. Scale Bar = 0.5mm. F, Quantification of Fraction Calcified Area for ex vivo porcine aortic valve treated with NSC23766 (25μM) in OGM for 10 days, n≥3. *p<0.05, **p<0.005, ***p<0.0005.

We primarily tested the usefulness of Rac1 targeted therapy using an ex vivo porcine aortic valve leaflets and PAVICs cultured in osteogenic media (OGM). We have shown previously that cultured ex vivo porcine leaflets can be treated to effectively mimic many of the properties of valve pathogenesis and can be used to assess the immediate effectiveness of pharmacological treatments in a histopathological context[25,33]. Cad11 is also an important mediator of OGM-induced calcification[16]. Alizarin Red staining revealed that treatment of ex vivo porcine aortic valve leaflets cultured in osteogenic media (OGM) with 25μM NSC23766 effectively reduces OGM-induced calcification by 2.8-fold as assessed via Alizarin Red (p=0.01) (Fig. 3, D-F).

PAVICs transfected with a Cad11 plasmid (Cad11 OX) in OGM reproducibly calcify and upregulate Runx2 mRNA expression[16]. Overexpression of Cad11 in PAVICs cultured in OGM (Cad11 OX) led to a 2.9 log-fold increase in Runx2 expression (p=0.02). (Fig. 4, D). Treatment of this condition with 10μM NSC23766 had only a 1.14 log-fold increase in PAVICs compared to the Control condition. This value was significantly different with respect to the Cad11 OX condition (p=0.02). Runx2 expression patterns correlated with data measuring Alizarin Red Activity. Cad11 overexpression in OGM-cultured PAVICs lead to a 2.7-fold increase in detected Alizarin Red Activity relative to the control (p=0.02) (Fig. 4, A-C, F), which was reduced to .92-fold relative to the control condition when treated with NSC23766. This was again a significant decrease relative to the Cad11 OX condition (p=0.01). While Cad11 overexpression does not affect OGM-induced nodule size, treatment of Cad11 OX PAVICs with NSC23766, reduced nodule size by 37% (Fig. 4, E). This was significant relative to the Control (p=0.03) and the Cad11 OX condition (p=0.02). In regular growth media, PAVICs do not calcify but Cad11 overexpression in these PAVICs induces nodule formation. Treatment of Cad11 OX PAVICs with NSC23766 completely prevented Cad11-induced nodule formation in regular growth media (Fig. SI). Together, the data suggest that NSC23766 is able to effectively decrease Cad11-induced calcification in terms of nodule number, calcified area and nodule size in ex vivo and in vitro models as well as in osteogenic and normal media conditions.

Figure 4. Rac1-GEF inhibition via NSC23766 prevents Cad11-induced calcification in PAVICs.

A-C, Alizarin red staining of PAVICs cultured in osteogenic growth media for 10 days with Control, Cad11 OX, Cad11 OX + NSC23766 (10μM) conditions. Scale bar = 1mm. D, Cad11 and Runx2 mRNA expression for Control, Cad11 OX, Cad11 OX + NSC23766 conditions cultured for 7 days determined, n=3. Data was log transformed to display reasonably using a single scale. E-F, Nodule Size and Normalized Alizarin Red Activity of Control, Cad11 OX, Cad11 OX + NSC23766 conditions normalized to the Control, n=8. *p<0.05, **p<0.005.

4.4. NSC23766 effectively prevents Cad11-induced PAVIC migration and compaction.

Scratch Wound Assays in regular growth media revealed that Cad11 overexpression in PAVICs stimulated a 6.2% increase in wound closure relative to the Control condition (p=0.008) (Fig. 5, A-G). Treatment of Cad11 OX PAVICs with NSC23766 prevented this increase in migration entirely indicating (p=0.005), suggesting that Cad11-induced migration is entirely Rac1 dependent. Cad11 OX PAVICs were observed to have less lone cells/scratch, 4.54±0.08 relative to the 11.2±.045 lone cells/scratch in the control condition, suggesting that Cad11 promotes collective migration (p<0.005) (Fig. 5, H). Treatment of Cad11 OX PAVICs with NSC23766 increased this value to 7.1±0.06 lone cells/scratch (p<0.0005), suggesting that Cad11-induced collective migration is somewhat Rac1-dependent. In addition to cell migration, we assessed the effect of Rac1 inhibition on Cad11-induced compaction. Compaction assays using type I collagen gels revealed that Cad11 overexpression in PAVICs stimulated 9.2% increase in matrix compaction (p=0.02), while treatment of Cad11 OX PAVICs with NSC23766 significantly reduced this number by 7.8% (p=0.04) (Fig. 5, I). This data suggests that Cad11-induced compaction is also mediated by Rac1. Together, the data show that Rac1 inhibition can effectively prevent Cad11-induced PAVIC migration and compaction, both of which are important cellular pathogenic processes in CAVD.

Figure 5. Cad11-mediated collective migration and compaction is Rac1-dependent.

A-F, Scratch Wound Assay of Control, Cad11 OX, Cad11 OX + NSC23766 conditions at hour 0 and hour 12. White arrows indicate lone migrating cells. Dotted white lines mark migratory front. Scale bar = 1mm. G, Percent wound closure after 12 hours, n=4. H, Number of lone migrating cells per condition, n=4. I, Compaction of PAVICs seeded in Type I collagen gels for Control, Cad11 OX, Cad11 OX + NSC23766 conditions in normal media over 7 days. J, Quantification of gel size as a fraction of initial area, n=4. *p<0.05, **p<0.005, ***p<0.0005.

4.5. Expression of only the Cad11 transmembrane and intracellular region is sufficient and necessary to induce calcification and increased Rac1 activity in PAVICs

Cad11 is an intermembrane protein containing 5 extracellular domains, a transmembrane domain and highly conserved cytoplasmic domain. The extracellular domains of Cad11 dimerize at EC1, the outermost extracellular domain, to promote cell adhesion. We created truncated variants of our original Cad11 plasmid with either a deletion of the entire extracellular region (Δe) or a deletion of the entire intracellular region (Δc). Both variants had the transmembrane region intact and a signal peptide for localization to the plasma membrane. Expression levels in PAVICs were validated via qRT-PCR (Fig. 6, A-B). Calcification assays revealed that expression of only the extracellular region (Δc) was not sufficient to produce any increase in calcification (Fig. 6, D-H). These cells produced an average of 54.8±11.9 nodules, which was similar to the 54.1 ±4.3 nodules observed in the Control condition (p=0.9999). Expression of only the intracellular region (Δe) produced 135.8±14.9 nodules (p<0.0001), similar to 115.2±9.5 nodules produced in the Cad11 OX PAVICs condition (p=0.0013). Only the Cad11 OX condition and the Δe condition produced a significantly higher nodule number relative to the Control condition. Post-hoc tests following ANOVA analysis revealed that the Cad11 OX and Δe condition were not significantly different from each other (p=0.5113). This data illustrates that expression of only the Cad11 intracellular regions is sufficient and necessary to induce calcification.

Figure 6. Expression of only intracellular Cad11 is sufficient to induce calcification and increased Rac1 activity.

A, Schematic of Cad11 truncated constructs lacking either the cytoplasmic domain (Δc) or the extracellular domains (Δe). All constructs contain the signal peptide for membrane localization and retain the transmembrane domain (TM). B, RT-qPCR confirmation of truncated Cad11 expression in PAVICs. Data was log transformed to display reasonably using a single scale. C. Significance notation key: ‘*’ denotes significant to the Control while ‘#’ denotes significance to the Cad11 OX condition. D-G, Alizarin red staining of PAVICs cultured in osteogenic growth media for 10 days with Control, Cad11 OX, Δc and Δe conditions. Scale bar = 2mm H, Quantification of nodule number, n=4. I, GTP-Rac1, and GTP-RhoA expression for Control, Cad11, Δc and Δe conditions. J, GTP-Rac1 and GTP-RhoA expression for Control, Cad11 OX, Δc and Δe conditions 48 hours after transfection normalized to the Control, n=4. (‘*’ denotes statistical significance to the Control condition, ‘#’ denotes significance to the Cad11 OX condition, *p<0.01, **p<0.005, ***p<0.0001, ^#p<0.001, ^##p<0.0005, ^###p<0.00005.

Immunofluorescence was used to quantify the relative activity levels of Rac1 and RhoA in each condition (Fig. 6, I-J). GTP-Rac1 levels were 1.38-fold higher in the Cad11 OX condition (p=0.042) and 1.37-fold higher in the Δe condition relative to the control condition (p=0.047). However, GTP-Rac1 levels in the Δc condition were similar to the Control condition (p=0.97). Together, the data suggest that GTP-Rac1 levels correlated with the level of calcification observed in each condition. GTP-RhoA levels, however, were significantly higher in the Δc condition (p<0.0001), which did not calcify significantly relative to the Control condition. GTP-RhoA was .37-fold lower in the Cad11 OX condition relative to the Control (p=0.02) and similar to the Control condition in the Δe condition (p=0.934). There was no correlation observed between GTP-RhoA levels and the level of calcification in each condition. As a whole, this data supports the hypothesis that intracellular Cad11 may be driving calcification via increased Rac1 activity. Interestingly, we observed a similar correlation in β-catenin levels as GTP-Rac1 levels throughout the four conditions (Fig. SII). β-catenin was significantly elevated by 1.9-fold (p=0.03) in Cad11 OX mice. This data suggests that Rac1 may potentially be regulating Runx2 expression via β-catenin.

4.6. Trio specific Rac1-GEF inhibition via ITX3 effectively prevents calcification and osteogenic gene induction

Since Trio is a known binding partner of intracellular Cad11 that primarily promotes Rac1 activation[20], we tested whether Trio inhibition via the Trio-specific selective Rac1-GEF inhibitor, ITX3, would be sufficient to prevent Cad11-induced calcification. Treatment of Cad11 OX PAVICs cultured in OGM with ITX3 indeed reduced calcification significantly relative to the control condition (Fig. 7, A-D). Quantification of Alizarin Red revealed 2.9-fold increase in calcification in the Cad11 OX condition relative to the Control condition (p=0.0013). Treatment of Cad11 OX PAVICs with 10 μM ITX3 reduced this to 1.6-fold relative to the Control (p=0.27), which was significantly lower than the Cad11 OX condition (p=0.033). Furthermore, western blotting revealed that ITX3 treatment also reduced Cad11-induced increases in Runx2 and αSMA expression (Fig. 7, E). This data suggests that Rac1-induced calcification downstream of Cad11 is Trio-dependent. This data was validated in our Cad11 OX mice as we observed 6.3-fold increase in Trio expression in Cad11 OX mice relative to wildtype mice (p=0.01) and was 8.9 times more co-localized with Cad11 in Cad11OX mice than wildtype mice (p=0.004) (Fig. SIII).

Figure 7. Selective inhibition of the Trio Rac1-GEF domain via ITX3 prevents Cad11-induced calcification in PAVICs.

A-C, Alizarin red staining of PAVICs cultured in osteogenic growth media for 10 days with Control, Cad11 OX, Cad11 OX +ITX3 (10μM) conditions. Scale bar = 1mm. D, Alizarin Red Absorbance (405nm) Control, Cad11 OX, Cad11 OX + ITX3 conditions, n=6. E, Runx2 and αSMA expression for Control, Cad11 OX, Cad11 OX +ITX3 conditions. *p<0.05, **p<0.005.

5. Discussion

In this study, we identify Rac1 as a downstream effector of Cad11 that plays an active role in Cad11-mediated aortic valve calcification. Our data suggests that Rac1 but not RhoA is required for Cad11-induced osteogenic calcification mediated by Runx2 nuclear activity. Our investigation further revealed that Rac1 inhibition is effective at preventing many pathogenic functions of Cad11 including calcification, migration, and compaction, which are thought to drive early-stage CAVD. Additionally, Rac1 inhibition via NSC23766 is sufficient to prevent Cad11-induced calcification in osteogenic and normal conditions. Our findings suggest that inhibition of the major Rac1-GEFs may be an effective means of preventing Cad11-induced calcification in CAVD.

The perception of CAVD as an active disease process rather than a degenerative one is a relatively recent development[4]. In light of these paradigm-shifting advances, the field has made great strides in uncovering the pathophysiology of CAVD[3,10]. Despite all the progress that has been made in the past couple of decades, pharmacological therapy for the disease remains elusive. Aortic valve replacement is still the sole treatment option after irreversible damage has occurred to the valves and they are no longer functional[3]. Until that end-stage of the disease is reached, nothing can be done to halt or slow the progression of CAVD.

Many molecular players have been discovered to play a role in CAVD; however, the identification of an effective pharmacological target involves selecting one that will not perturb normal biological signaling and functioning. With recent data showing RG6125 to have low efficacy in phase II trials for rheumatoid arthritis, the search for new potential pharmacological therapies is pressing[18]. We have also shown that Cad11 inhibitors Celecoxib and its Cox-2 independent analogs including Dimethylcelecoxib paradoxically increase aortic valve calcification rather than decrease it due to a mechanism independent of Cad11, prompting the need to investigate other Cad11 pathway inhibitors[34]. Like Cad11, aberrant Rac1 signaling has been linked to a number of diseases. Rac1 has been found to drive cellular pathogenesis in cancer, neurodegenerative disease and rheumatoid arthritis[35]. Furthermore, Rac1 has been shown to be a significant driver of cardiovascular diseases such as cardiac hypertrophy and atherosclerosis[36]. Despite Rac1 having already emerged as a promising target for pharmacological inhibition in the cardiovascular field, our study is the first to link Rac1 to aortic valve disease.

We have also uncovered a new question of RhoA/Rac1 antagonism in the context of CAVD. We show that both GTP-Rac1 and GTP-RhoA are both highly elevated in calcified human aortic valve leaflets and with in vitro experiments, both small GTPases do not show clear patterns of temporal antagonism[16]. Furthermore, both GTP-Rac1 and GTP-RhoA are highly elevated in Cad11 OX Mice[16]. However, it is clear that GTP-Rac1 is more correlated with Cad11 spatial localization than GTP-RhoA in calcified human aortic valve leaflets. These findings suggest that antagonism between GTP-RhoA and GTP-Rac1 need not be temporal but may in this context be spatial within the calcifying environment itself. There is ample literature to support spatial or even compartmentalized antagonism even though RhoA and Rac1 are co-expressed temporally[37-39]. Furthermore, recent investigation has shown that Rac1/RhoA antagonism may not be universal but may actually depend on the cell-specific context[40]. We previously showed that RhoA/ROCK inhibition effectively prevents Cad11-induced calcification and changes in gene expression. Here we show that GTP-Rac1 not only prevents Cad11-induced calcification and changes in gene expression but also prevents Cad11-induced cell migration and compaction. The role of RhoA/Rac1 antagonism in aortic valve calcification and the possible existence of synergistic activity between the two Rho GTPases remains unclear. Further studies are needed to define the relationship between GTP-RhoA and GTP-Rac1 in aortic valve calcification.

Interestingly, we found that the overexpression of the intracellular Cad11 domain also contributed to increased calcification. In other contexts, E-cadherin intracellular domains are used as dominant negative controls[41]. However, Kashef et al. demonstrated similar results with regards to the role of Cadherin-11 in neural crest cell migration showing that the Δe condition had preserved migration while the Δc condition did not, suggesting that intracellular Cad11 may be a significant contributor to signaling in a variety of contexts[20]. It is possible that the mechanism of intracellular signaling slightly differ among the different Cadherins.

Understanding the pathophysiology of potential CAVD pharmacological targets will inform the development of highly selective Rac1 inhibitors. Studies have suggested that the anti-atherosclerotic properties of statins may be in part due to indirect inhibition of the activity of the small GTPases[36]. Therefore, the development of selective small GTPase inhibitors is of significant interest to many areas within the cardiovascular fields. Our data suggest that selective inhibition of the Rac1-GEF domain of Trio may potentially be an effective method for selective inhibition of Cad11-Rac1 interactions in these disease contexts.

6. Conclusion

GTP-Rac1 is highly upregulated in calcified human aortic valve leaflets and is highly co-localized with Cad11. Rac1 is necessary for Cad11-induced Runx2 nuclear localization. Furthermore, Inhibition of Rac1-GEFs via NSC23766 prevents calcification in ex vivo leaflets and calcification, migration and compaction in porcine aortic valve interstitial cells. These data illustrate the potential for Rac1-GEF inhibition as a pharmacological therapy for CAVD.

Highlights.

• Rac1 is highly upregulated in calcified human aortic valve leaflets

• Rac1 is highly colocalized with Cad11 in calcified human aortic valve leaflets

• Rac1 mediates Cad11-induced aortic valve pathogenic processes

• Rac1 itself promotes downstream cell migration, compaction, and calcification

• The Rac1 inhibitor, NSC23766, prevents Cad11-induced calcification