Ucma/GRP inhibits phosphate-induced vascular smooth muscle cell calcification via SMAD-dependent BMP signalling

Vascular calcification (VC) is the process of deposition of calcium phosphate crystals in the blood vessel wall, with a central role for vascular smooth muscle cells (VSMCs). VC is highly prevalent in chronic kidney disease (CKD) patients and thought, in part, to be induced by phosphate imbalance. The molecular mechanisms that regulate VC are not fully known. Here we propose a novel role for the mineralisation regulator Ucma/GRP (Upper zone of growth plate and Cartilage Matrix Associated protein/Gla Rich Protein) in phosphate-induced VSMC calcification. We show that Ucma/GRP is present in calcified atherosclerotic plaques and highly expressed in calcifying VSMCs in vitro. VSMCs from Ucma/GRP−/− mice showed increased mineralisation and expression of osteo/chondrogenic markers (BMP-2, Runx2, β-catenin, p-SMAD1/5/8, ALP, OCN), and decreased expression of mineralisation inhibitor MGP, suggesting that Ucma/GRP is an inhibitor of mineralisation. Using BMP signalling inhibitor noggin and SMAD1/5/8 signalling inhibitor dorsomorphin we showed that Ucma/GRP is involved in inhibiting the BMP-2-SMAD1/5/8 osteo/chondrogenic signalling pathway in VSMCs treated with elevated phosphate concentrations. Additionally, we showed for the first time evidence of a direct interaction between Ucma/GRP and BMP-2. These results demonstrate an important role of Ucma/GRP in regulating osteo/chondrogenic differentiation and phosphate-induced mineralisation of VSMCs.

In physiological conditions VSMCs exist in the vessel wall as contractile cells and regulate vascular tone. However, VSMCs are known to have a high degree of phenotypic plasticity 12 . In response to stress and injury, VSMCs lose expression of contractility-related genes such as SM22α, calponin (CNN1), and myosin light chain (MLC) [13][14][15][16] . When vascular injury is persistent, the phenotypic transition is dysregulated and VSMCs can undergo an unfavourable transdifferentiation into cells with characteristics of osteoblasts or chondrocytes 4,12,[17][18][19][20] . This is termed osteo/chondrogenic differentiation. Many bone mineralisation-regulating proteins were found to be expressed in the calcifying blood vessel, such as BMP-2 Runx2, MGP, osteocalcin (OCN), osteopontin (OPN) 18,21,22 amongst others and have been shown to play an active part in regulating VC. Additionally, calcifying VSMCs have been shown to release extracellular vesicles in a mechanism similar to release of matrix vesicles from chondrocytes 23 . However, the exact molecular mechanisms regulating VSMC osteo/chondrogenic transdifferentiation are unknown.
Ucma (Upper zone of growth plate and Cartilage Matrix Associated protein; also known as Gla Rich Protein -GRP) is a novel mineralisation inhibitor, which was first reported in cartilage 24,25 and later in the vasculature 26,27 . Ucma/GRP-deficient mice did not develop a clear phenotype 28 , however in vitro studies revealed that Ucma/ GRP regulates differentiation of chondrocytes and osteoblasts [29][30][31] . Using immunohistochemistry, Ucma/GRP was shown to be present at sites of VC. Moreover, when added exogenously, Ucma/GRP inhibited calcification of aortic rings in vitro 26,27,29 .
Based on the involvement of Ucma/GRP in differentiation of osteoblasts and its role in VC, we hypothesized that Ucma/GRP is involved in osteo/chondrogenic differentiation of VSMCs. To test our hypothesis we examined expression of Ucma/GRP during phosphate-induced calcification of VSMCs. We also examined osteo/ chondrogenic gene expression and calcification of Ucma/GRP −/− VSMCs exposed to elevated levels of inorganic phosphate. Our findings indicate that Ucma/GRP reduces VC by inhibiting osteo/chondrogenic VSMC transdifferentiation through a BMP-2-regulated pathway.

Treatment of VSMCs in vitro with phosphate induces calcification.
We first set out to examine calcification of primary mouse VSMCs in vitro. WT VSMCs were cultured in medium with elevated phosphate as described before 8,32 (osteogenic medium, OM), which induced mineralisation in vitro, as observed by quantification of deposited calcium phosphate crystals (Fig. 1A). To control for spontaneous mineral precipitation on collagen, collagen-coated wells were incubated with OM in the absence of VSMCs. Under these a-cellular conditions, no calcification was observed (data not shown). To examine osteo/chondrogenic differentiation in mineralisation of VSMCs in vitro, Runx2 and Osteocalcin expression was measured using qPCR after 12 days of incubation with osteogenic medium (Fig. 1B and C). Expression of both markers was significantly increased, suggesting that the observed mineralisation is an active process involving osteo/chondrogenic differentiation. Moreover, treatment of human VSMC with osteogenic medium increased secretion of extracellular vesicles (Fig. 1D), which have recently been shown to mediate VC 23 .
Ucma/GRP is expressed in atherosclerotic plaques of ApoE −/− mice. Ucma/GRP has previously been shown to be expressed in calcified human aortas and aortic valves 27 . Here we demonstrated that Ucma/ GRP is expressed in advanced atherosclerotic lesions of ApoE −/− mice. UcmaGRP was expressed specifically in atherosclerotic plaques ( Fig. 2A and B) in cells which show similarities to chondrocyte morphology. As shown in Fig. 2C and D, these cells also expressed osteo/chondrogenic markers BMP-2 and Runx2. This suggests that Ucma/GRP is involved in osteo/chondrogenic differentiation and calcification of VSMCs in atherosclerotic plaques. Interestingly, not all cells with chondrocyte-like morphology visibly expressed Ucma/GRP. Calcification of mVSMCs was associated with increased expression of osteogenic genes Runx2 and osteocalcin measured by qPCR, n = 3. (D) Calcification of human VSMCs increased exosome secretion measured using a bead capture assay, n = 3. All graphs show mean + SD. Statistical significance was assessed using t-test. **p < 0.001-0.01, ***p < 0.001.
SCIENTIFIC RePoRts | (2018) 8:4961 | DOI:10.1038/s41598-018-23353-y Ucma/GRP expression in VSMCs in vitro increases in response to osteogenic medium. The presence of Ucma/GRP in murine VSMCs (mVSMCs) in vitro was confirmed using RT-PCR (Fig. 3A). Additionally, immunocytochemistry (Fig. 3B) has shown expression of Ucma/GRP in scattered cytoplasmic foci, which is consistent with its subcellular localization in other cell types 31 . Next, in order to examine whether Ucma/GRP has a role in mineralisation of VSMCs in vitro, expression was measured in response to treatment with osteogenic medium (Fig. 3C). Ucma/GRP expression, measured by qPCR, increased in a time-dependent manner with a 20-fold increase after 12 days of culture. Immunocytochemistry analysis confirmed this result and showed that control medium does not induce a similar increase (Fig. 3D).
Ucma/GRP-deficient VSMCs are more prone to calcification in response to osteogenic medium. Aortas were harvested from Ucma/GRP −/− mice (KO) and wild type (WT) counterparts generated as previously described 28 . The knock-out of Ucma/GRP was confirmed by immunohistochemical staining of the epiphyseal plate (Supplemental Fig. 1A). VSMCs were isolated and their identity was verified by immmunocytochemical staining for VSMC markers (αSMA, CNN1, p-MLC) (Supplemental Fig. 1B). We and others 26,27 have shown that Ucma/GRP is upregulated in response to calcification and that its expression is confined to osteo/chondrogenic cells in atherosclerotic plaques (Figs 2A and 3A,B). Therefore, we hypothesized that Ucma/GRP is involved in controlling osteo/chondrogenic differentiation of VSMCs and consequently calcification. To test this, WT and Ucma/GRP −/− VSMCs were incubated in osteogenic medium over a time span of 15 days ( Fig. 4A and B). Both WT and Ucma/GRP −/− VSMCs calcified in response to osteogenic medium, with calcification becoming detectable after 6 days. However, Ucma/GRP −/− VSMCs calcified approximately twice as much as WT cells after 9, 12 and 15 days. Neither Ucma/GRP −/− nor WT VSMCs calcified in control medium (Supplemental Fig. 1C). Ucma/GRP knock-down in WT cells using siRNA resulted in significantly increased calcification compared to cells transfected with a scramble control (Supplemental Fig. 1D), confirming that the observed effects are specific to Ucma/GRP. In response to osteogenic medium alkaline phosphatase (ALP) activity was increased in Ucma/GRP −/− cells but not in WT cells, suggesting increased osteo/chondrogenic potential of Ucma/GRP −/− cells (Fig. 4C).
Ucma/GRP deficiency results in increased osteo/chondrogenic gene expression. Since osteo/ chondrogenic differentiation is an important mechanism regulating VC and expression of osteo/chondrogenic markers is found in atherosclerotic plaques, we set out to examine whether there were any differences in osteo/ chondrogenic marker expression between WT and Ucma/GRP −/− VSMCs. We observed that Ucma/GRP −/− cells expressed increased levels of BMP-2 mRNA in baseline conditions (CM, Fig. 5A). Additionally, when challenged with osteogenic medium, Ucma/GRP −/− VSMCs expressed higher levels of OPN, Osteocalcin and BMP-2, compared to WT VSMCs (Fig. 5A,B and C). MGP mRNA expression was significantly downregulated in Ucma/ GRP −/− compared to WT VSMCs, both in baseline and osteogenic conditions, in line with the increased calcification potential of these cells (Fig. 5D). Upon treatment with osteogenic medium MGP expression increased, both in WT and Ucma/GRP −/− VSMCs. Western blot analysis demonstrated an increase in β-catenin and phosphorylated SMAD1/5/8 expression, suggesting the involvement of SMAD signalling pathways ( Fig. 5E,F,G) known to be involved in osteoblast mineralization and VC. Additionally, expression of Runx2 and ALP was higher in Ucma/ GRP −/− cells compared to WT (Fig. 5E,H and I). Importantly, we also observed that Ucma/GRP −/− VSMCs displayed reduced expression levels of contractile phenotype marker αSMA compared to WT VSMCs (Supplemental Fig. 1E). Taken together these results show that Ucma/GRP −/− VSMCs undergo increased osteo/chondrogenic differentiation, compared to WT VSMCs, both at baseline as well as in osteogenic conditions.
Further to that, to investigate the involvement of specific downstream targets of BMP-2 we treated cells with dorsomorphin, an inhibitor of SMAD1/5/8 activation (AMPK inhibitor). Dorsomorphin decreased calcification of WT VSMCS and restored levels of calcification in Ucma/GRP −/− cells to the levels observed in WT VSMCs (Fig. 6C). The reduction in calcification was reflected by a decrease in p-SMAD1/5/8 and ALP expression (Fig. 6D,E,F). Taken together, these results suggest that increased osteo/chondrogenic and calcification potential of Ucma/GRP −/− VSMCs is due to impaired inhibition of SMAD-dependent BMP signalling.
As Erk signalling has been known to also activate SMAD signalling 33 , we investigated whether inhibiting Erk phosphorylation has an effect on calcification of Ucma/GRP −/− VSMCs. Treatment of WT and Ucma/GRP −/− VSMCs with Erk phosphorylation inhibitor U0126 did not change calcification levels of cells of either genotype, suggesting that Erk phosphorylation and Ucma/GRP are not interacting to inhibit calcification of VSMC cultures (Supplemental Fig. 2B).

Discussion
VC is an important predictor of cardiovascular mortality. In particular, CKD patients have been shown to be at an increased risk of VC 5,34 . However, the mechanisms through which VC develops remain incompletely understood. In the present study we demonstrate that Ucma/GRP plays a role in the attenuation of osteo/chondrogenic differentiation of VSMCs through a pathway involving SMAD-dependent BMP signalling. Osteo/chondrogenic switching is accelerated in the absence of Ucma/GRP and this results in enhanced calcification. We also provide evidence that high phosphate concentrations, which is a common finding in CKD patients 7 , induce upregulation of Ucma/GRP in VSMCs. Our findings reveal a novel pathway through which Ucma/GRP inhibits phosphate-induced SMAD signalling and osteo/chondrogenic differentiation, thereby reducing the propensity for calcification of the vessel wall.  29 . Ucma/GRP is known to be present in low levels in calcified aortic valves and in the media. Ucma/GRP inhibits calcification of aortic rings cultured in calcifying medium, when added exogenously to cultures, this was accompanied by an increase in αSMA expression and a decrease of osteopontin expression 27 . In our study we confirmed the presence of Ucma/GRP in the vasculature and for the first time reveal that Ucma/GRP expression is markedly increased in osteo/chondrogenic cells in atherosclerotic plaques of ApoE −/− mice, which implicates its role in this vascular pathology. CKD patients show accelerated atherosclerosis, which suggests that our in vivo findings are relevant for calcification in CKD-related atherosclerosis 35,36 .
BMP-2 is a potent inducer of calcification and regulator of osteo/chondrogenic differentiation of VSMCs, known to cause phosphate influx into cells 21 . BMP-2 acts by binding its cell surface receptors which transduce the signal to the cytoplasm by phosphorylating pathway-restricted SMADs (SMAD1 and SMAD5) for BMPs 37 . This leads to heterodimerization of pathway-restricted SMADs with SMAD4, a common-mediator SMAD, and translocation of the complex to the nucleus, where it binds directly to DNA 38 . As a result, changes in gene expression occur, such as upregulation of Runx2, OCN and ALP. Additionally, BMP-2 has been shown to induce accumulation of β-catenin 39 . β-catenin has been demonstrated to activate Runx2 expression in the context of phosphate-induced VSMC calcification 40 . BMP and Wnt pathways have been shown to be co-activated in calcifying VSMCs 41 , where SMAD1 and β-catenin interacted and regulated gene expression together. Additionally, dorsomorphin homologue 1 has been recently shown to inhibit phosphate-induced osteogenic differentiation of hVSMCs by inhibiting BMP-2 42 . Our data are in line with these findings, as we demonstrate increased Here we demonstrate for the first time that Ucma/GRP interacts with this BMP-SMAD signalling pathway by binding BMP-2 and that inhibiting BMP-2/-4 signalling with noggin and SMAD1/5/8 phosphorylation with dorsomorphin decreased calcification of VSMCs and osteo/chondrogenic gene expression. Interestingly, previous reports show that externally added BMP-2 decreased Ucma/ GRP expression in chondrocytes 29 , suggesting an interaction between BMP-2 signalling and Ucma/GRP. In addition to that, in osteoblasts Ucma/GRP is under direct transcriptional control of Runx2 and Osterix 30 , however this has not been investigated in VSMCs.
In our study we confirm that high levels of phosphate induce hallmarks of osteogenic differentiation in VSMCs, such as osteo/chondrogenic gene expression and extracellular vesicle release, and that Ucma/GRP expression is increased in VSMCs treated with osteogenic medium in vitro. This, together with the presence of Ucma/GRP in advanced atherosclerotic lesions, suggests that calcifying cells increase the expression of this inhibitor in a bid to protect themselves from adverse phenotype changes. This has been shown to happen with MGP, another vitamin K-dependent calcification inhibitor, by us in the present study and others 43 . Additionally, insufficient carboxylation of MGP due to warfarin administration is one of the factors contributing to VC in CKD 44 . Uncarboxylated Ucma/GRP, similar to uncarboxylated MGP, could contribute to these processes. Ucma/ GRP is a vitamin K-dependent protein, as it undergoes posttranslational carboxylation 45 . It has been previously shown that uncarboxylated Ucma/GRP accumulates at sites of pathological calcification in human aortic valves and media and only the carboxylated form of Ucma/GRP was able to inhibit calcification 27 . Our results are in line with this, as we show that carboxylated Ucma/GRP has high affinity to BMP-2, whereas uncarboxylated Ucma/ GRP binds BMP-2 only weakly. However, we have not investigated the carboxylation status of Ucma/GRP in the atherosclerotic plaques which we examined and this warrants further investigation.
MGP is known to inhibit VC by binding mineral and interacting with BMP-2, preventing it from binding its receptor 7,9,23 . It is therefore tempting to speculate that vitamin K-dependent inhibitors complement each other with respect to regulating VC: calcium-induced calcification is regulated by MGP and whereas phosphate-induced calcification is regulated by Ucma/GRP. This is further supported by previous studies showing that phosphate increases osteo/chondrogenic differentiation of VSMC in vitro, whereas this was not demonstrated for calcium yet 8,12,46 . Based on our observations, we propose a model of how Ucma/GRP inhibits VC (Fig. 7). When VSMCs are exposed to elevated phosphate concentrations in the extracellular environment, as is the case in CKD patients, osteo/chondrogenic differentiation (as evidenced by increased Runx2, β-catenin, osteocalcin, osteopontin and ALP expression) is initiated via increased BMP-2-SMAD signalling. We propose that Ucma/GRP can ameliorate the effect of increased phosphate by inhibiting this signalling pathway and thus reduce calcification of the vessel wall.
Taken together, we demonstrate that Ucma/GRP, a member of a range of inherent calcification inhibitors present in the vasculature 47 , plays an important role in regulating osteo/chondrogenic differentiation of VSMCs. Our findings have implications for vascular pathologies which lead to VC and represent an advancement in our knowledge about the steps resulting in VC. One of the main goals in the treatment of patients with CKD is to reduce the substantially increased risk of cardiovascular comorbidity and mortality, improving both quality and length of life. The increase in CKD stage and cardiovascular risk is accompanied by increased phosphate levels. As shown by our results, the phosphate-Ucma/GRP-BMP axis holds promise for novel interventions. Pharmacological manipulation of Ucma/GRP levels could represent a novel therapeutic strategy to prevent VC.
Stock, University of Erlangen, Germany, with the approval of the local ethics authorities (University of Erlangen-Nuremberg and the Government of Mittelfranken, Ansbach, Germany) and according to the regulations of the animal facilities in Germany 28 . Mice were maintained in a specific pathogen-free environment with free access to water and western type diet (Arie Block, Woerden, the Netherlands) and sacrificed by portal vein puncture after 12 weeks. Aortas were harvested for immunohistochemistry and VSMC isolation.
VSMC isolation and culture. Endothelium and adventitia were removed from aortas of Ucma/GRP −/− and WT littermates. Tissue was dissected into ± 5 mm² pieces and digested with 3 mg/ml collagenase (Sigma, Zwijndrecht, the Netherlands) and 1 mg/ml elastase (Sigma) for 4 hours at 37 °C in DMEM (Gibco, Bleiswijk, the Netherlands). The cells were washed, resuspended in growth medium (DMEM with 10% FCS, Gibco, 100 U/ ml penicillin and 100 mg/ml streptomycin, Gibco) and transferred to culture dishes. Success of the isolation was determined by immunofluorescent staining for αSMA, SM22α, phosphorylated myosin light chain 2 (method below) and western blotting for αSMA. Cells at passage 4-12 were used for experiments. Cells were maintained in culture medium and passaged when 90% confluent. Cell cultures were not supplemented with additional vitamin K. The following inhibitors were added to media for treatments: 2 µM of U0126 (Merck, Amsterdam, the Netherlands), 100 ng/ml noggin (R&D systems, Abingdon, United Kingdom), 2 µM dorsomorphin dihydrochloride (sc-361173, Santa Cruz). All inhibitors were added to cells for the full duration of the experiment, as indicated in figure legends. Medium was refreshed every 2-3 days. Human primary VSMCs (hVSMCs) were used for exosome quantification assay and co-immunoprecipitation. They were maintained in M199 with 20% FCS, 100 U/ml penicillin and 100 mg/ml streptomycin (Gibco).
Calcification assays. 4 × 10 4 cells were seeded per well of a 12-well plate. After 24 hours medium was changed to osteogenic medium (growth medium +2.6 mM PO 4 3− , OM) or growth medium (control medium, CM). Quantification of deposited calcium was carried out using a calcium determination kit (Randox, London, United Kingdom) according to the manufacturer's instruction, after solubilising mineral deposits in 0.1 M HCl. Calcium measurements were normalised to protein content using micro BCA assay (Thermo Scientific, Bleiswijk, the Netherlands). Calcification was additionally visualised using Alizarin Red S staining adapted from Gregory et al. 49 .
RNA isolation and quantitative real-time PCR. Total RNA was isolated using TRI Reagent (Sigma). RNA concentration and quality were determined using a Nanodrop 2000 spectrophotometer (Thermo Scientific, Bleiswijk, the Netherlands) and agarose-denaturing gel electrophoresis. 1 µg of total RNA was treated with DNAse I (Promega, Leiden, the Netherlands) and reverse transcribed using M-MLV reverse transcriptase (Invitrogen, Bleiswijk, the Netherlands), an oligo dT adapter (Eurogentec, Maastricht, the Netherlands) and RNAse Out (Invitrogen, Bleiswijk, the Netherlands). Real-time qPCR was performed using the Quantitect SYBR green PCR kit (Qiagen, Venlo, the Netherlands) in a LightCycler 480 II (Roche, Woerden, the Netherlands) with 50 ng of cDNA and 0. Fluorescence curves were analyzed with LightCycler 480 Software (Version 1.5) and relative quantification was performed with the 2 −ΔΔCt method. Ucma/GRP was detected as described previously 45 . All samples were assayed in triplicate.
Quantification of exosome release. Exosomes were quantified using a bead-capture assay as described before 23 . Briefly, anti-human CD63 antibody (556019, BD Bioscience) was immobilized on aldehyde-sulfate functionalized beads (Invitrogen). Human primary VSMC culture media was incubated with anti-CD63-coated beads on a shaker overnight at 4 °C. VSMCs were stained with Hoechst 33342 (Thermo Fisher) and counted using a Cytation 3 live-cell imager. Beads were washed and incubated with anti-CD81-PE antibodies for 1 h at room temperature. Then beads were washed with PBS with 2% BSA and analyzed by flow cytometry (Accuri C6, BD). Arbitrary units (AU) were calculated as mean fluorescence units times percentage of positive beads and normalized to the number of cells.

Co-immunoprecipitation.
Co-IP assays were performed using conditioned media from hVSMCs cultured in the presence of 500 ng/ml of recombinant BMP-2 (Preprotech) for 24 h in M199 with 10% FBS, and the Dynabeads Co-immunoprecipitation Kit (Novex, Life Technologies), according to manufacturer's recommendations. The BMP-2/4 monoclonal (Santa Cruz Biotechnology, sc-137087) and negative IgG antibodies were used as capture antibodies in the Co-IP reactions, and Western blot analysis of the eluted proteins were performed using CTerm-Ucma/GRP (GenoGla Diagnostics, Faro, Portugal) and BMP-2/4 antibodies as described above.
Alkaline phosphatase activity. Cells were lysed in 1% Triton X-100 in PBS, subjected to 2 freeze-thaw cycles and centrifuged at 13000 g for 5 minutes. ALP activity in the supernatants was measured at 405 nm using 4-Nitrophenyl phosphate disodium salt hexahydrate (Sigma) as substrate. Enzyme activity (U) was normalised to protein concentrations. Statistical analysis. Data shown are mean ± SD. All data were verified in ≥3 independent experiments. Statistical analysis was performed by 1-way ANOVA with Bonferroni post hoc test or Student's t-test, as stated in figure legends, using PRISM software (GraphPad). Values of P < 0.05 were considered statistically significant.
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
Disclosures. Cees Vermeer is the CSO of VitaK BV, a spin-off company fully owned by Maastricht University.
Dina C. Simes and Carla S. B. Viegas are cofounders of GenoGla Diagnostics.