Discoidin domain Receptor 2: A determinant of metabolic syndrome-associated arterial fibrosis in non-human primates

Collagen accumulation and remodeling in the vascular wall is a cardinal feature of vascular fibrosis that exacerbates the complications of hypertension, aging, diabetes and atherosclerosis. With no specific therapy available to date, identification of mechanisms underlying vascular fibrogenesis is an important clinical goal. Here, we tested the hypothesis that Discoidin Domain Receptor 2 (DDR2), a collagen-specific receptor tyrosine kinase, is a determinant of arterial fibrosis. We report a significant increase in collagen type 1 levels along with collagen and ECM remodeling, degradation of elastic laminae, enhanced fat deposition and calcification in the abdominal aorta in a non-human primate model of high-fat, high-sucrose diet (HFS)-induced metabolic syndrome. These changes were associated with a marked increase in DDR2. Resveratrol attenuated collagen type I deposition and remodeling induced by the HFS diet, with a concomintant reduction in DDR2. Further, in isolated rat vascular adventitial fibroblasts and VSMCs, hyperglycemia increased DDR2 and collagen type I expression via TGF-β1/SMAD2/3, which was attenuated by resveratrol. Notably, gene knockdown and overexpression approaches demonstrated an obligate role for DDR2 in hyperglycemia-induced increase in collagen type I expression in these cells. Together, our observations point to DDR2 as a hitherto unrecognized molecular link between metabolic syndrome and arterial fibrosis, and hence a therapeutic target.


Ethics statement
Handling assessment, 14 male rhesus monkeys (7-13 years) were quasi-randomized into 2 groups: a HFS diet (n = 10); and the healthy standard diet (n = 4). The SD was a purified biscuit with 13% kcal from fat and less than 5% sucrose by weight. The HFS diet was a specially formulated purified ingredient diet with 42% kcal from fat and approximately 27% sucrose by weight (Harlan, Teklad, Madison, WI). The monkeys were gradually switched to the HFS diet over a 3-week period. All groups received 2 meals per day of the specified diet for 2 years in allotments that represented ad libitum feeding, ensuring iso-caloric consumption across groups.
The study also included a resveratrol group (n = 10), provided with a daily dietary supplementation of resveratrol (80 mg during the 1st year and 480 mg during the 2nd year) along with the HFS diet. Animal experiments were performed under conditions described earlier [21].

Isolation of rat vascular adventitial fibroblasts and VSMCs
Vascular adventitial fibroblasts and VSMCs were isolated from young adult male Sprague-Dawley rats following protocols previously standardized by us [22]. Aorta was removed from left subclavian origin, washed in HBSS solution and then transferred in a sterile hood into clean 100 mm dish with fresh HBSS. Fat was removed gently under a microscope and incubated in a 35 mm dish containing 2 ml Digestion Medium I for 30 min at 37˚C. Tissue was then placed in fresh HBSS, cut open, separated into the outermost adventitial layer and middle VSMCs after scrapping out the endothelium, and incubated in Digestion Medium II for 90 min at 37˚C. After incubation, an equal volume of M199 containing 10% FBS was added to the dish and mixed well for proper dissociation. Cells were collected by filtering through a cell strainer (70 μm) and centrifuged at 1500 rpm for 5 min. The cell pellet was collected and resuspended in 0.5 ml of M199 containing 10% FBS and seeded on 12-well plate. At 24 h after isolation, the supernatant containing unattached cells was discarded and the cells were incubated with M199 containing 10% FBS in a CO 2 incubator at 37˚C. Cultures of adventitial fibroblasts and VSMCs were characterized and checked for cross-contamination by immunocytochemistry. Adventitial fibroblasts stained positive for DDR2 and negative for Desmin (S1

Real-time polymerase chain reaction analysis
Sub-confluent cultures of adventitial fibroblasts and VSMCs were subjected to the indicated treatments and total RNA was isolated using TRI reagent, according to the manufacturer's instructions. Following DNase I treatment, 2 μg of total RNA was reverse transcribed to cDNA with random primers and M-MLV reverse transcriptase. TaqMan quantitative RT-PCR analysis was carried out using the ABI prism 7500 Sequence Detection System (Applied Biosystems, CA) with specific FAM-labeled probes. PCR reactions were performed over 40 cycles, as per the manufacturer's instructions. DDR2 and collagen mRNA expression levels were normalized to β-actin and 18s rRNA, respectively.

Western blot analysis
Sub-confluent cultures of adventitial fibroblasts and VSMCs in serum-free M199 were treated with high glucose, and relative collagen and DDR2 protein expression was determined by western blot analysis following standard protocols, using β-actin as loading control. Enhanced chemiluminescence reagent was used to detect the proteins, and protein expression was quantified by densitometric scanning (Bio-Rad Laboratories). The following protocol was employed for western blot analysis on monkey aortic tissue homogenate. Immediately after harvesting, monkey abdominal aortic tissue was rinsed in 1X PBS, frozen in liquid nitrogen and pulverized in the presence of T-PER Tissue Protein Extraction Reagent. Lysates were centrifuged for 10 min to remove debris and 40 μg of total homogenous arterial protein was resolved by SDS-PAGE and transferred onto PVDF membrane. The transferred membranes were incubated in 3% non-fat milk containing primary antibodies at 4˚C overnight. HRP-conjugated IgG were used as secondary antibodies and detected with Super Signal West Pico Chemiluminescent Substrate. Densitometric analysis was performed on the protein bands from the aortae of 3 monkeys from each group. Beta-actin immunoblotting was used as a protein loading control.

RNA interference
RNA interference protocol was as reported by us earlier. Briefly, cells were seeded on 12-well plates at 8×10 4 cells/well. After 24 h, the cells were incubated in Opti-MEM with Ambion predesigned Silencer Select siRNA [5 pmol DDR2, 5 pmol collagen α1 type1, 5 pmol TGF-β1 or scrambled siRNA, control] and Lipofectamine 2000 (2 μl) for 19 h. Following an additional incubation in M199 with 10% FBS for 12 h, the cells were treated with high glucose for the indicated duration. Cell lysate was prepared in SDS lysis buffer, denatured and used for western blot analysis.

Over-expression of DDR2
Strong constitutive expression of DDR2 was achieved under a pCMV promoter, purchased as cDNA clone from Sinobiologicals, China (Rat DDR2/CD167b Gene ORF cDNA). The DDR2 clone size was verified by PCR amplification of the primers provided in the kit. The size of the plasmid was further analyzed by single site restriction digestion at the Mlu-1 site. DDR2 overexpression plasmid cocktail was prepared with 1 μg plasmid and 3 μl Lipofectamine 2000 in 100 μl Opti-MEM, which was added to the cultures for incubation in M199+10% FBS for 8 hours. After a recovery period of 12 hours, cells were pre-treated with the TGF-β1 inhibitor (SB431542 hydrate) or resveratrol followed by HG treatment. Cultures of adventitial fibroblasts and VSMCs were transfected with FLAG-tagged DDR2 plasmid vector. The number of cells staining positive for FLAG in 4-5 fields was analyzed by immunocytochemistry (S3A and S3B Fig) to ascertain overexpression efficiency, which was found to be >80%.

Histochemistry, immunohistology and semi-quantification analysis
In 4-6 consecutive 5 μm paraffin embedded aortic cross-sections of monkey abdominal aortae, we performed Elastica van Gieson (EVG) (to determine elastin degradation), picrosirius red (to determine fibrosis), collagen (to determine collagen deposition), collagen type I and DDR2 antibodies stain (to determine collagen type I deposition and DDR2 expression, respectively), and Masson Trichrome stain (to determine extracellular matrix remodeling). In 4-6 consecutive 10 μm frozen aortic cross-sections, we performed Oil Red O (to determine fat deposits) and Alizarin Red staining (to determine calcium deposits), according to the modified protocols as reported in prior studies [21]. 4 fields each from 4 abdominal aortic cross-sections from each animal were analyzed. Methods and criteria of semi-quantification and quantification of histochemical and immunobiological staining are shown in Table 1.

Statistical analysis
Data were expressed as Mean±S.D. Statistical analysis was performed using Student's t test. p�0.05 was considered significant. Data were also analyzed by one-way ANOVA and p�0.05 was considered statistically significant. The data presented are biological replicates that are representative of three independent experiments where n = 3. In vivo, experiments were statistically analyzed using One-way ANOVA and Kruskal-Wallis with Post-hoc Dunn's test and p�0.05 was considered statistically significant. For in vivo experiments, data were expressed as Mean±S.E.M [SD (n = 4),HFS (n = 10), and HFS+RESV (n = 10)].

DDR2 mediates TGF-β1-dependent increase in collagen type I gene expression in advential fibroblasts exposed to high glucose
The effect of 25 mM glucose (HG) on collagen type I alpha1 mRNA and protein expression in vascular adventitial fibroblasts was determined by real-time PCR and western blot analyses, respectively. A 2-fold increase in collagen type I alpha 1 mRNA ( Fig 1A) and protein expression ( Fig 1B and 1C) was observed in cells exposed to hyperglycemic conditions for 12 h.
HG also caused a 2-fold increase in DDR2 mRNA expression ( Fig 1D) and a 2.5-fold increase in DDR2 protein expression (Fig 1E and 1F) in these cells. The effect of HG on collagen type 1 alpha 1 and DDR2 expression was sustained at 48 h (S4A and S4B Fig).
To test the possibility that a regulatory relationship exists between DDR2 and collagen type 1 in HG-stimulated adventitial fibroblasts, cells were transfected with DDR2 siRNA as described under Methods. After confirming knockdown, collagen type I alpha1 expression was determined following treatment with HG. RNA interference-based inhibition of DDR2 attenuated the stimulatory effect of HG on collagen type I alpha 1 expression (Fig 1G and 1H), indicating a regulatory role for DDR2 in collagen type I gene expression in HG-treated adventitial fibroblasts. The regulatory role of DDR2 in collagen was also observed under basal conditions (S4E Fig).
Inhibition of TGF-β1 using SB431542 abolished HG-induced increase in collagen type I alpha 1 and DDR2 expression (Fig 2A-2D), which was further confirmed by RNA interference-based TGF-β1 knockdown (Fig 2E and 2F). Knockdown of TGF-β1 also attenuated basal expression levels of collagen type I alpha 1 and DDR2 (S4G Fig). Importantly, however, the decrease in collagen type I alpha 1 expression upon TGF-β1 inhibition was not observed in cells over-expressing DDR2 (Fig 2G and 2H). Together, these data show that TGF-β1 acts via DDR2 to enhance collagen type I expression in adventitial fibroblasts exposed to HG. Inhibition of SMAD 2 and 3 using PD169316 and SIS3, respectively, was found to attenuate HG-induced collagen type I alpha 1 (Fig 3A and 3B) and DDR2 (Fig 3C and 3D) expression in adventitial fibroblasts, indicating the regulatory role of TGF-β1/SMAD2/3 signaling in HGinduced increase in DDR2 and collagen expression. ERK1/2 inhibition using PD98059 attenuated HG-induced expression of collagen type 1 alpha 1 (Fig 4A and 4B) but not DDR2 (Fig 4C  and 4D). Further, DDR2 knockdown prevented HG-induced ERK1/2 phosphorylation (Fig 4E  and 4F), indicating that DDR2-dependent ERK1/2 activation enhances collagen type I expression in cells exposed to HG.

05). (B and C)
Protein was isolated and subjected to western blot analysis for detection of collagen type I alpha 1, with β-actin as loading control. � p< 0.05 vs. control, ns-not significant vs. control (One-way ANOVA p<0.05). (D-F) Sub-confluent quiescent cultures of vascular adventitial fibroblasts in M199 were stimulated with HG (25mM). (D) DDR2 mRNA levels were determined by Taqman quantitative real-time PCR analysis at 12 h of HG treatment, with β-actin as loading control. � p< 0.05 vs. control, ns-not significant vs. control (One-way ANOVA p< 0.05). (E and F) Protein was isolated and subjected to western blot analysis for detection of DDR2, with β-actin as loading control. � p< 0.05 vs. control, ns-not significant vs. control (One-way ANOVA p< 0.05). (G and H) Regulatory relationship between DDR2 and collagen type I in HG-treated adventitial fibroblasts. Adventitial fibroblasts were transiently transfected with DDR2 siRNA or scrambled siRNA. Following exposure of the transfected cells to HG for 12 h, collagen type I alpha 1 protein expression was examined by western blot analysis and normalized to β-actin. � p< 0.05 vs. control, ## p< 0.01 vs. HG (One-way ANOVA p< 0.05). Data are representative of three independent experiments, n = 3. Error bars represent SD.

Resveratrol prevents hyperglycemia-induced DDR2 and collagen type I in vascular adventitial fibroblasts
Pre-treatment of cells with 50 μmol/L resveratrol significantly reduced HG-induced increase in collagen type I alpha 1 expression (Fig 5A and 5B). Resveratrol was also found to prevent HG effect on DDR2 expression (Fig 5C and 5D). It attenuated basal expression levels of collagen type I alpha 1 and DDR2 in vascular adventitial fibroblasts (S4H and S4I Fig). Notably, resveratrol inhibited HG-induced (Fig 5E and 5F) and basal (S4J Fig) TGF-β1 expression in these cells. Interestingly, however, the inhibitory effect of resveratrol on HG-stimulated collagen type I expression was not observed in DDR2-overexpressing cells, showing that attenuation of HG-stimulated collagen type I expression by resveratrol is due to its negative regulation of DDR2 (Fig 5G and 5H). Moreover, exogenous addition of recombinant TGF-β1 (10ng/mL) restored DDR2 and collagen type I alpha 1 expression in resveratrol-treated cells, suggesting that resveratrol inhibits TGF-β1 to attenuate DDR2 and collagen type I expression in response to HG (Fig 5I and 5J).
Hyperglycemia stimulates collagen type I and DDR2 expression in VSMCs. To check whether similar mechanisms exist in VSMCs, cells were exposed to hyperglycemic conditions and DDR2 and collagen type I alpha 1 mRNA and protein expression patterns were analyzed. Significant up-regulation of collagen type I and DDR2 expression (Fig 6A-6F) was observed in VSMCs as well. Moreover, the effect of HG on collagen type I and DDR2 expression was sustained at 48 h (S4C and S4D Fig). Interestingly, the regulatory relationship between DDR2 and collagen type I seen in adventitial fibroblasts was observed in VSMCs exposed to HG (Fig 6G  and 6H) and under basal conditions (S4F Fig). Further, as in adventitial fibroblasts, resveratrol (50 μM) abolished HG-induced increase in collagen type I alpha 1 (Fig 7A and 7B) and DDR2 in VSMCs (Fig 7C and 7D). Resveratrol also attenuated the basal expression levels of collagen type I alpha 1 and DDR2 in VSMCs (S4K and S4L Fig). While resveratrol attenuated HG-stimulated collagen type I expression in VSMCs, the effect was not observed in DDR2-overexpressing cells, showing that attenuation of HG-stimulated collagen type I expression by resveratrol is due to DDR2 inhibition (Fig 7E and 7F). Moreover, while HG-induced TGF-β expression in VSMCs was reduced by resveratrol (Fig 7G and 7H), exogenous addition of recombinant TGF-β-1 (10 ng/mL) restored DDR2 and collagen type I alpha 1 expression in resveratroltreated cells, suggesting that resveratrol inhibits TGF-β1 to downregulate DDR2 and collagen type I expression in HG-treated cells (Fig 7I and 7J).
Association of DDR2 with collagen deposition and remodeling in the abdominal aorta in a monkey model of metabolic syndrome. A significant increase in body weight and cholesterol, loss of endothelial cell integrity, lipid and macrophage infiltration and elevated blood pressure were earlier reported by us in adult male Macaca mulatta fed a HFS diet for 2 years [21]. In the present study, the HFS diet was found to induce adverse abdominal aortic remodeling, as evidenced by extensive degradation of the elastic laminae (Fig 8A and 8B), calcification (Fig 8C and 8D) enhanced staining for fat deposition (Fig 8E and 8F), and ECM remodeling events (Fig 9A-9F).
A significant increase in ECM remodeling (Fig 9A and 9B) along with total collagen levels ( Fig 9C and 9D) was observed in the abdominal aortic wall in animals fed the HFS diet for 2 DDR2 in metabolic syndrome-associated arterial fibrosis in non-human primates years. Importantly, the changes in collagen type I expression levels (Fig 9E and 9F) and total collagen levels (Fig 9C and 9D) were associated with a marked increase in DDR2 in the arterial wall in animals fed the HFS diet (Fig 9G and 9H). The study explored the effect of resveratrol on HFS diet-induced changes in the animals. A significant reduction in the markers of arterial remodeling and atherogenic progression, including elastin degradation, calcification and fat deposition, was observed in the resveratrol-fed animals (Fig 8A-8F). Resveratrol also produced a marked decrease in DDR2 expression that correlated with a reduction in ECM remodeling, collagen remodeling and collagen type I immunostaining in these animals (Fig 9A-9H). Consistent with these observations, western blot analysis showed a significant increase in the expression of collagen types I and III in the abdominal aortic tissue from the HFS-fed group, which correlated with an increase in the expression of DDR2. These changes were not observed in the resveratrol-group (Fig 9I-9N). The colocalization of DDR2 in adventitial fibroblasts, VSMCs and endothelial cells was analyzed (Fig 10A and 10B). DDR2 was found to colocalize with α-SMA in fibroblasts and VSMCs (Fig 10A). Colocalization of DDR2 was not observed in endothelial cells (Fig 10B). Fig 10C provides a schematic representation of DDR2-dependent increase in collagen type 1 gene expression in vascular cells exposed to hyperglycemic conditions.

Discussion
Metabolic syndrome, which is a cluster of conditions including elevated blood pressure, high blood sugar, dyslipdemia and abdominal obesity, is a significant contributor to the pathogenesis of vascular fibrosis [23]. Risk factors associated with metabolic syndrome alter the phenotype of adventitial fibroblasts and VSMCs in favor of enhanced collagen production to promote vascular fibrosis and its sequelae [18]. The underlying mechanisms, however, remain largely unexplored.
Our earlier studies demonstrated that HFS diet induces a marked increase in central arterial stiffening, as evidenced by an increase in pulse wave velocity, and enhanced signatures of inflammatory markers driven by oxidative stress in the thoracic aorta of adult male rhesus monkeys [21]. The present study employed this model of metabolic stress to focus on ECM remodeling in the abdominal aortic wall, which is prone to atherosclerotic remodeling in response to metabolic insult. Here, we show a significant increase in collagen type I deposition within the medial and adventitial layers and adverse collagen remodeling in the HFS-fed group (Fig 9A-9F). Notably, the increase in collagen deposition and adverse ECM remodeling correlated well with increased expression of DDR2 within adventitial fibroblasts and VSMCs in the HFS-fed group (Fig 9G and 9H), suggesting a possible link between DDR2 and collagen production in a setting of metabolic syndrome. Further, the present investigations on isolated vascular cells indicated an obligate role for DDR2 in collagen type I gene expression in adventitial fibroblasts and VSMCs exposed in vitro to hyperglycemia, which is an important risk factor associated with metabolic syndrome. The finding is in accord with a previous report from our laboratory that Angiotensin II-stimulated collagen type I expression in cardiac fibroblasts DDR2 in metabolic syndrome-associated arterial fibrosis in non-human primates is DDR2-dependent [24]. Together, the findings point to the shared characteristic of a regulatory relationship between DDR2 and collagen in cells of mesenchymal origin that express both DDR2 and collagen, which may be an important determinant of fibrosis in the cardiovascular system. DDR2 in metabolic syndrome-associated arterial fibrosis in non-human primates Degeneration of compliant elastin fibers, and deposition of stiffer collagen, is considered an important cause of arterial stiffening. The HFS diet was found to induce elastin degradation in the arterial wall, which can contribute significantly to loss of arterial resilience and consequent increase in arterial stiffness. In this regard, DDR2 is known to trigger matrix metalloproteinase-2 activation [18], which in turn is closely associated with elastolysis, calcification and atherosclerotic plaque vulnerability, the hallmark of metabolic arterial syndrome [25]. It is likely that, in addition to promoting collagen deposition and remodeling, DDR2 may also have a role in inducing other structural changes in the arterial wall via MMP-2 activation.
It is pertinent to note that previous studies on DDRs in relation to collagen remodeling during vascular injury and progression of atherosclerosis were focused largely on DDR1. For example, Hou et al. (2001) demonstrated a positive regulatory role for DDR1 in collagen expression and deposition in a rat model of carotid artery injury [26]. However, another study by the same group reported a negative correlation between DDR1 and collagen production within plaques of atherosclerotic mice, which may contribute to increased plaque stability [27]. The study also demonstrated an increase in pro-collagen type I and III mRNA expression in VSMCs lacking DDR1, indicating an inverse relationship between DDR1 and collagen. On the other hand, a link between DDR2 and vascular fibrosis has not hitherto been demonstrated although the regulation of DDR2 in vascular cells has been addressed earlier [28]. A study by Ferri et al. (2004) on atherosclerosis-associated collagen remodeling in non-human primate (C) Schematic representation of DDR2-dependent increase in collagen type 1 gene expression in vascular cells exposed to hyperglycemic conditions. models showed that early and late stage atherosclerotic plaques express DDR1 and DDR2 [29]. The study did not correlate DDR2 levels with collagen levels. However, based on the observation that over-expression of DDR1 or DDR2 in isolated VSMCs plated on polymerized collagen-coated dishes down-regulated collagen mRNA expression, the authors concluded that DDRs may promote plaque instability. Furthermore, another study [30] found no significant difference in basal levels of expression of collagen type I in VSMCs isolated from DDR2 knockout mice, compared to the wild-type, which could be due to compensatory mechanisms in the knockout model. In contrast to these reports, our findings from both the in vitro and in vivo models point to a positive correlation between DDR2 and collagen expression and remodeling, suggestive of a role for DDR2 in promoting arterial fibrosis associated with metabolic syndrome.
Our in vitro studies demonstrate the effect of hyperglycemia on collagen production, which is generally attributed to the action of Angiotensin II [31] and TGF-β1 [32]. We found that TGF-β1/SMAD2/3 signaling mediates the stimulatory effect of hyperglycemia on DDR2 that in turn enhances collagen type 1 expression via ERK1/2 MAPK activation (Figs 2-4). Notably, while TGF-β1 knockdown inhibited DDR2 and collagen expression, the effect of TGF-β1 inhibition on collagen expression was not observed in cells over-expressing DDR2 (Fig 2E-2H), clearly demonstrating the centrality of DDR2 in mediating TGF-β1-dependent collagen type I expression in these cells. Together, the data presented here suggest that DDR2 may be a key player in the progression of vascular fibrosis and an early marker of fibrotic vascular remodeling. In this context, it is pertinent to note that DDR2 has previously been linked to pulmonary fibrosis in mice [24].
Interestingly, Mattison et al. (2014) had earlier demonstrated the therapeutic benefits of resveratrol in preventing arterial wall stiffening through amelioration of vascular inflammation and oxidative stress in this model of metabolic syndrome [21]. Resveratrol is reported to affect several pathways in various disease models [33,34]. Although it has been shown to reduce collagen production in VSMCs [35] and in rodent models of fibrosis [36], neither its molecular target nor its beneficial effect in mitigating vascular fibrosis through down-regulation of collagen production has hitherto been demonstrated in a clinically-relevant model of metabolic syndrome. In the present study, resveratrol addition to the HFS diet reduced elastin degradation and markers of adverse arterial remodeling such as calcification and fat deposition within the arterial wall. Further, dietary supplementation of resveratrol attenuated the increase in collagen type I and adverse collagen remodeling in the vascular wall. Importantly, the reduction in collagen content and ECM remodeling was associated with a concomitant reduction in DDR2 within the aortic adventitia and media, raising the possibility that resveratrol may exert its beneficial effects via DDR2. Such a postulation is supported by our observations that resveratrol abolishes the stimulatory effect of hyperglycemia on DDR2 and collagen in isolated adventitial fibroblasts and VSMCs (Figs 5A-5D and 7A-7D). Notably, while resveratrol attenuated HGstimulated collagen expression, the effect was not observed in DDR2-overexpressing cells (Figs 5G, 5H, 7E and 7F). Considered in tandem, these observations suggest that DDR2 could be a target of resveratrol in ameliorating vascular fibrosis.
In summary, the findings from this study support a role for DDR2 in the pathogenesis of vascular fibrosis in a setting of metabolic syndrome. Since vascular fibroblasts and VSMCs are the predominant source of both DDR2 and collagen in the vasculature, the regulatory relationship between DDR2 and collagen in these cell types underscores the existence of a molecular circuitry involving these two pro-fibrotic factors that may drive tissue fibrosis in response to injury. Clearly, with its specific localization in adventitial fibroblasts and VSMCs in the vascular wall, DDR2 emerges as a potential drug target in the management of adverse fibrotic vascular remodeling.