|Home | About | Journals | Submit | Contact Us | Français|
Accumulating experimental evidence implicates β-catenin signaling and enzyme transglutaminase 2 (TG2) in the progression of vascular calcification, and our previous studies have shown that TG2 can activate β-catenin signaling in vascular smooth muscle cells (VSMCs). Here we investigated the role of the TG2/β-catenin signaling axis in vascular calcification induced by warfarin.
Warfarin-induced calcification in rat A10 VSMCs is associated with the activation of β-catenin signaling and is independent from oxidative stress. The canonical β-catenin inhibitor Dkk1, but not the Wnt antagonist Wif-1,prevents warfarin-induced activation of β-catenin, calcification, and osteogenic trans-differentiation in VSMCs. TG2 expression and activity are increased in warfarin-treated cells, in contrast to canonical Wnt ligands. Vascular cells with genetically or pharmacologically reduced TG2 activity fail to activate β-catenin in response to warfarin. Moreover, warfarin-induced calcification is significantly reduced on the background of attenuated TG2 both in vitro and in vivo.
TG2 is a critical mediator of warfarin-induced vascular calcification that acts through the activation of β-catenin signaling in VSMCs. Inhibition of canonical β-catenin pathway or TG2 activity prevents warfarin-regulated calcification, identifying the TG2/β-catenin axis as a novel therapeutic target in vascular calcification.
Medial arterial calcification is common in patients with chronic kidney disease and diabetes mellitus. It leads to arterial stiffening and thus contributes to the development of hypertension. One of the major risk factors for vascular calcification, especially in the setting of uremia and hyperphosphatemia, is warfarin – a staple in anticoagulant therapy administered as a long-term treatment to millions of people. Warfarin treatment correlates with cardiovascular calcification in human patients 1-4 and causes rapid and extensive medial calcification in animal models 5-7. Molecular mechanisms of warfarin-induced calcification are emerging. Warfarin inactivates several regulators of tissue calcification, including Matrix Gla Protein (MGP) 8 and growth arrest-specific gene 6 protein 9, through inhibition of their vitamin K-dependent carboxylation. This pharmacodynamic effect of warfarin is based on its chemical similarity to vitamin K 10. In addition, in a manner similar to vitamin K2, warfarin can directly activate the transcription factor PXR 10, thus affecting the expression of osteogenic genes that are up-regulated during the osteoblast-like transformation of calcifying VSMCs 11-14. Moreover, warfarin binding can affect protein conformation 15 and may therefore regulate the activities of various other proteins. Hence, the overall biological effects of warfarin may be mediated by diverse, non-mutually exclusive mechanisms.
Calcifying vascular cells consistently demonstrate signs of osteochondrogenic trans-differentiation, suggesting that similarities exist in molecular regulation between bone development and vascular calcification. Enzyme transglutaminase 2 (TG2) has been identified as a potent regulator of osteo-chondrogenic differentiation and bone formation 16-20. A role for this protein in vascular calcification is suggested by the reported accumulation of either TG2-mediated cross-links or the TG2 protein itself in the calcifying arteries of warfarin-treated rats 21 and MGP–/– mice 22. Moreover, experiments using cultured VSMCs and aortic rings have shown that TG2 enhances calcification in elevated inorganic phosphate (Pi) 23, 24, while genetic ablation of TG2 prevents this 24, identifying this enzyme as a key regulator of Pi-induced calcification in vitro, although no mechanistic insights have been specified.
The underlying molecular mechanisms of TG2-mediated calcification remain to be determined; however, our earlier research revealed that TG2 activates β-catenin signaling in VSMCs concomitant with increased calcium deposition 23. β-catenin signaling is a key pathway in osteogenesis 25 that has been implicated in diabetes-associated medial calcification 26 and in valvular calcification 27. Accretion of the major components of the canonical β-catenin signaling pathway, including LRP5, Wnt3a, and β-catenin protein, suggests activation of this conduit in calcifying heart valves 27. Activation of β-catenin signaling in calcifying medial VSMCs has also been observed in the diabetic model of LDLR–/– mice 26, and both calcium accrual and β-catenin signaling are inhibited by activation of PTH signaling in the vessel walls of these animals 28.
Together, in vitro and in vivo data imply that β-catenin activation is a likely mediator of vascular calcification; however, the potential involvement of this signaling in warfarin-induced medial calcification remains elusive. In this paper, we demonstrate that canonical β-catenin signaling is essential in warfarin-induced vascular calcification. Further, we demonstrate that TG2 is critical for the regulation of both β-catenin signaling and warfarin-dependent calcification in VSMCs. Our results support a key role for the TG2/β-catenin signaling axis in arterial calcification.
Animal used were CB57/B6 and TG2–/– mice (a kind gift from Robert Graham, Victor Chang Cardiovascular Institute, New South Wales, Australia). All procedures were approved by the institutional animal care and use committee at the University of Maryland Medical School and were conducted in compliance with NIH guidelines for the care and use of laboratory animals. Arterial calcification was induced by warfarin-vitamin K diet following the described protocol 7.
The A10 clonal embryonic rat aortic smooth muscle cell line (A10 VSMC; ATCC) was used. Cells were cultured for 6 days in DMEM, 1% FBS, 1.6 mM inorganic phosphate and 10 μM warfarin (Sigma) unless otherwise stated in the text. Calcium content was determined colorimetrically using the Calcium (CPC) liquicolor kit (Stanbio) in 0.1M HCl extracts from cultured cells or aortic tissues.
Cellular oxidative stress was detected using the cell-permeable fluorogenic probe CellROX (5 μM; Molecular Probes).
A stable β-catenin reporter A10 cell line was established using the CIgnal Lenti TCF/LEF Luciferase Reporter (SA Biosciences). Luciferase assay (Promega) was performed in a 96-well plate format and normalized to total lactate dehydrogenase in cell lysates.
TG2 activity in total cell lysates was assayed in an ELISA-like assay as previously described 29 by incorporation of biotinylated pentylamine, EZ-link (Pierce), into N,N’-Dimethylcasein (Sigma). . Incorporated EZ-link was detected with ExtrAvidin-Peroxidase (1:5000, Sigma) and Super AquaBlue peroxidase substrate (eBioscience), and measured at 405nm on a Polarstar Optima plate reader. Purified guinea pig liver transglutaminase 2 (Sigma-Aldrich) was used as a standard. TG2 activity in live cells was assayed by stable incorporation of the rhodamine-labeled synthetic peptide Pro-Val-Lys-Gly (SY2011) 30 and detected by fluorescence microscopy
Western blots were analyzed by densitometry using Scion Image (Scion Corp.). Real-time PCR and luciferase activity data were quantified using Excel (Microsoft Corp.). Results are graphed as means ± standard error (SEM). Statistics (t-test and ANOVA) were performed using Vassar Stats software (Vassar College, NY). A value of p<0.05 was considered statistically significant.
Previous studies have shown that warfarin induces calcification of the elastin-rich arterial wall in vivo and in vitro, suggesting a direct effect of warfarin on vascular tissue 31. Here, we examined whether warfarin can induce calcification in VSMC cultures which lack the elastic lamellae. High levels of inorganic phosphate (Pi) are critical for warfarin-induced ex vivo calcification in aortic rings 31, 32, and Pi alone is sufficient to induce calcification in cultured VSMCs in a time- and dose-dependent manner in the 1.6mM to 3.0mM range 33. To allow for the detection of warfarin-induced calcification, we selected relatively low Pi levels to promote moderate calcification, and serum was used at 1% since higher serum levels have been shown to inhibit in vitro calcification 34.
In cultured rat A10 VSMCs, moderate calcification was induced by 1.6mM Pi in 1% FBS medium (17.7±6.4 compared to 0.12±0.03 μg Ca2+/mg protein in normal 1% medium containing ~1 mM Pi, p<0.01). This was significantly increased by the addition of 10μM warfarin, up to 62±15 μg Ca2+/mg protein, p<0.01; however, lower levels of Pi, including 1.0 and 1.2 mM, did not support warfarin-induced calcification (Fig.1A). This result implies Pi-induced sensitization of VSMCs towards the pro-mineralizing effect of warfarin and thus, supports the important role of phosphate in vascular calcification 31, 33, 35, 36. In addition, warfarin elicited deposition of calcified matrix by VSMCs in a dose-dependent manner (Fig. 1B), highlighting the specificity of the observed effect.
Because vascular calcification can potently be promoted by oxidative stress 37, we examined the possibility that warfarin induced oxidative stress in VSMCs. Reactive oxygen species were detected by the incorporation of a fluorogenic probe (ROX) in live A10 VSMCs exposed to either 10 μM warfarin or to 100 μM menadione (vitamin K3), an oxidative stress inducer used as a positive control. Activation of the ROX probe, resulting in bright red fluorescence, was detected only in the vitamin K3-treated cells (Fig. 1C). In cells exposed to warfarin no fluorescence was observed, indicating that warfarin had no oxidative effect on VSMCs. In agreement with this, expression of the oxidative stress markers HIF-1a, Keap-1 and Nrf-2 was not significantly affected by warfarin treatment (Fig. 1D). These results indicate that warfarin-induced calcification in VSMCs is independent of oxidative stress.
Recent studies indicate that warfarin may affect cells via diverse mechanisms, some that involve the inhibition of protein γ-carboxylation and others that do not 10. Taking into consideration the proposed role of β-catenin signaling in vascular calcification associated with various clinical conditions 26, 27, we examined the potential involvement of this signaling pathway in warfarin-induced calcification in vitro. β-catenin-dependent luciferase reporter activity showed a significant 2-fold increase in response to warfarin treatment (p<0.01) (Fig. 2A). This activation was specific to warfarin and was not detected in the 1.6mM Pi-medium that supports moderate calcification (Fig. 2A). To confirm the luciferase reporter results, three major hallmarks of β-catenin activation were analyzed: accumulation of β-catenin protein, its nuclear localization, and transcription of endogenous β-catenin target genes. In the pro-mineralizing conditions warfarin-induced a 2-fold increase in total β-catenin protein (p<0.05) (Fig. 2B), caused its nuclear translocation (Fig. 2C), and activated transcription of β-catenin target genes, including axin2, cyclin D1 and Tcf4 (p<0.01) (Fig. 2D). Notably, the canonical β-catenin activator Wnt3a induced nuclear translocation of β-catenin in VSMCs to a similar extent as warfarin (Fig. 2C) indicating substantial activation of β-catenin in our model.
Activation of β-catenin through the canonical pathway is mediated by LRP5/6 receptors, while secreted Dikkopf proteins (Dkks) act to inhibit this pathway by binding to the LRP5/6 receptors, targeting them for endocytosis 38. Recombinant Dkk1 prevented the warfarin-induced activation of the β-catenin reporter (Fig. 3A) and the warfarin-dependent increase in calcification in VSMCs (Fig. 3B), demonstrating a critical role for the LRP5/6-mediated canonical pathway in this process. Further, Dkk1 attenuated warfarin-induced activation of osteogenic genes (Fig. 3C), indicating that canonical β-catenin signaling plays a key role in the osteogenic trans-differentiation of calcifying VSMCs. Of note, expression of Msx2 mRNA did not increase with warfarin treatment suggesting that activation of β-catenin in VSMCs was not via Msx2, similar to the minor role for Msx2 noted in the calcifying tunica media of the MGP-/- mice 39 and in agreement with activation of Msx2 restricted to adventitia in the diabetic LDLR-/- mice 26.
We next sought to identify endogenous activators of the canonical β-catenin pathway that may mediate the effect of warfarin. mRNA levels of four endogenous canonical Wnt ligands Wnts 2, 3a, 7b and 8b expressed in cultured VSMCs 40 were not significantly affected by warfarin (Supplemental Fig. I A), and Wif-1, a known antagonist to Wnt ligands 41, did not mitigate the activation of β-catenin or the calcification induced by warfarin (Supplemental Fig. I B,D). Together these data indicate a Wnt-independent modulation of canonical β-catenin signaling induced by warfarin. Enzyme transglutaminase 2 (TG2) has been shown in our previous work to bind LRP5, activate β-catenin signaling, and enhance VSMC calcification induced by chondrocyte-conditioned medium 23. To examine whether calcification and β-catenin activation induced by warfarin can be mediated by TG2, we analyzed TG2 levels and activity in our system. We found that warfarin caused a significant 6.8±0.4 fold (p<0.01) increase in TG2 mRNA levels (Fig. 4A). This increase in expression was accompanied by a similar elevation in total transamidating activity, evaluated by an enzymatic ELISA-like activity assay measuring pentylamine incorporation into immobilized casein 29 (Fig. 4B). The dihydroisoxazole TG2 inhibitor, ERW-1069 42, dramatically attenuated the warfarin-dependent induction of transamidating activity (Fig. 4B), suggesting that this increase is almost entirely owing to the TG2 protein. This was confirmed by the analysis of arterial tissues from wild-type (WT) and TG2–/– (TG2-KO) mice, in which several TG isoforms are expressed (Supplemental Fig. II). While warfarin significantly increased total transamidating activity in VSMCs from the wild-type mouse, this increase was not observed in the TG2–/– cells (Fig. 4C). Interestingly, in addition to increasing TG2 expression, we found that warfarin activates the TG2 protein. When added to purified TG2 in the enzymatic assay, warfarin increased transamidating activity in a dose-dependent manner (Fig. 4D)
Catalytic cross-linking activity of TG2 is regulated allosterically at the GTP-binding domain 43 and it is believed to be suppressed both intracellularly and at the cell surface 19, 44. To further confirm that warfarin increased cellular TG2 activity in VSMCs, cultured cells were exposed to rhodamine-labeled peptide substrate Pro-Val-Lys-Gly (SY2011) 30, 45 for four hours, followed by an overnight recovery in regular growth medium to allow for exocytosis of unincorporated substrate. Multiple sites of SY2011 incorporation were observed in warfarin-treated cells in contrast to vehicle-treated control cells (Fig. 4E) further confirming induction of TG2 activity by warfarin.
Our data demonstrate that warfarin increases TG2 levels and activity, and our previous work has shown that elevated TG2 can activate β-catenin in VSMCs 23. We, therefore, hypothesized that increased TG2 mediates the warfarin-induced activation of β-catenin associated with enhanced vascular calcification. To test this hypothesis, we analyzed the effects of TG2 inhibition on the warfarin-mediated activation of β-catenin in A10 VSMCs. The dihydroisoxazole TG2 inhibitors attenuated warfarin-induced accumulation of β-catenin protein (Fig. 5A), as well as activation of β-catenin transcriptional activity (Fig. 5B, Supplemental Fig. III A). Further, genetic ablation of TG2 prevented the warfarin-dependent increase of β-catenin protein in ex vivo cultures of the TG2-/- arterial tissue (Fig. 5C), while exogenous purified TG2 added to the culture medium partially rescues this phenotype by supporting a significant 1.5-fold increase of β-catenin levels (p<0.01) (Fig. 5C). At the same time, the wild-type mouse arterial tissue responds to warfarin treatment ex vivo similar to rat VSMCs with a 2.5-fold increase (p<0.01) in the levels of β-catenin protein (Supplemental Fig. III B). These data confirm the critical role for TG2 in the activation of β-catenin signaling by warfarin. Of note, the addition of exogenous purified TG2 had no significant effect on β-catenin protein levels in TG2–/– tissue in the absence of warfarin (Fig. 5C, grey bars). Together with the above observation that warfarin increases the activity of TG2 (Fig. 4D), this strengthens the notion that only catalytically-active TG2 enhances vascular β-catenin signaling.
Lastly, we analyzed whether the increase in catalytic activity of TG2 was required for warfarin-induced calcification. In vitro, the TG2 inhibitor ERW-1069 completely blocked the increase in calcium deposition induced by warfarin (Fig. 6A) and attenuated activation of osteogenic gene expression by warfarin in VSMCs (Fig. 6B). Next, to examine the role of TG2 in warfarin-induced calcification in vivo, wild-type and TG2–/– mice were treated for 3 weeks with warfarin-vitamin K regimen 7, at which time point significant calcification (approximately 2 μg Ca2+/mg dry weight) has been reported in rats 32. The basal levels of arterial calcification in both genotypes are very similar (0.12±0.005 μg Ca2+/mg dry weight in wild type aortas versus 0.1±0.008 μg Ca2+/mg dry weight in the TG2–/– aortas). Warfarin treatment resulted in a massive 18-fold increase in aortic calcification in the wild-type mice (up to 1.8±0.35 μg Ca2+/mg dry weight), but in a significantly lower 5-fold increase in calcification in the TG2–/– mice (up to 0.69±0.12 μg Ca2+/mg dry weight) (Fig. 6C, p<0.01). These findings demonstrate the critical role of TG2 in warfarin-induced vascular calcification.
This study presents the first direct evidence for the critical role of TG2-mediated activation of the β-catenin signaling pathway in warfarin-induced medial calcification. While under normal physiological condition vascular β-catenin signaling is generally inactive 46, despite the expression of several canonical Wnt ligands in VSMCs 40, its activation associates with vascular injury and remodeling 47, 48. Previous studies have implicated the β -catenin signaling pathway in vascular calcification 49 based on the reported accumulation of the pathway proteins in calcific valve disease 27 and in diabetic calcifying lesions 26, 28, 50. In addition, circumstantial evidence for β-catenin signaling in medial calcification is provided by the demonstrated abilities of MMPs to activate β-catenin 51 and promote calcification associated with elastin degradation 52, 53. However, the mechanisms regulating the β-catenin pathway in valvular and vascular calcification have yet to be characterized 49, 54, and the contribution of this pathway to calcification has not been demonstrated.
In this study, warfarin-induced calcification and osteogenic gene expression were studied in cultured aortic smooth muscle (A10) VSMCs, independent of the inflammatory response that commonly accompanies calcification in vivo. Earlier studies investigated inhibition of the γ-carboxylation pathway by warfarin in cultured VSMCs and in arterial tissue ex vivo 31, 34. Here, we examined whether other mechanisms relevant to vascular calcification, such as oxidative stress and β-catenin signaling, are affected by warfarin and thus may mediate its effect. We did not observe induction of oxidative stress in VSMCs by warfarin, indicating that this mechanism is likely not involved in warfarin-mediated calcification in vitro. However, we did find that warfarin potently activates β-catenin signaling. Further, we observed that the LRP5/6 antagonist Dkk1, which has been shown to inhibit osteogenic differentiation in mesenchymal cells and myofibroblasts 26, 55, attenuates warfarin-induced β-catenin activity, expression of osteogenic genes, and calcium deposition in VSMCs. These findings provide novel, direct evidence for the key role of the LRP5/6-mediated canonical β-catenin pathway in osteoblastic transformation and calcification in VSMCs, and thus expand previous observations implicating β-catenin signaling in diverse types of vascular calcification 49.
There is only a limited understanding of endogenous activators of β-catenin signaling in vascular calcification in general, and in warfarin-induced calcification in particular. Wnts 2, 3a, 7b and 8b are expressed by VSMCs 40 and Wnt7b and Wnt3a are increased in the calcifying arteries of the diabetic atherosclerotic LDLR-/- mice and in calcified heart valves 26, 27, 49, indicative of the potential role of these canonical ligands in β-catenin activation. However, the levels of these Wnt ligands are not changed in calcification of the MGP-/- arteries 39 or during attenuation of β-catenin activity and calcification in the LDLR-/- mice by PTH 28, and we found that they are also not affected by the calcification-inducing warfarin treatment of VSMCs. Further, warfarin-induced activation of β-catenin is not prevented by Wif-1, an antagonist of Wnt ligands. Therefore, other factors responsible for regulation of the canonical β-catenin pathway in vascular calcification should be considered.
Here we show that warfarin induces up-regulation and activation of TG2 – an enzyme that can bind canonical LRP5 receptor, activate β-catenin signaling, and enhance phosphate-induced calcification in VSMCs 23. Using pharmacological and genetic approaches we show that catalytic activity of TG2 is critical for warfarin-induced activation of β-catenin signaling, osteoblast-like trans-differentiation of VSMCs, and calcium mineral deposition in vitro and in vivo, demonstrating the key role for the TG2/β-catenin axis in warfarin-induced calcification. Moreover, in addition to increased TG2-dependent protein cross-linking in the calcifying aortas of warfarin-treated rats 21, previous studies showed the requirement for catalytic activity of TG2 in phosphate-induced aortic calcification ex vivo 24 and accumulation of TG2 in arterial walls of the MGP-/- mice 22, indicating that such a mechanism may be involved in diverse variants of medial vascular calcification, although involvement of the β-catenin pathway in these cases still needs to be defined. Interestingly, in contrast to medial calcification, in atherosclerotic plaques elevated TG2 is associated with inflammatory cells and acts to regulate plaque stability, showing little correlation with vascular calcification 56-59. Thus, biological activities of TG2 may be determined by the local milieu of various regulators that differ between the vascular wall layers.
A plausible model for the warfarin-induced calcification mediated by the TG2/β-catenin axis is illustrated in figure 7. The observed up-regulation of TG2 transcript awaits further elucidation. One of the possible mechanisms of this transcriptional activation may be owing to the activation of the nuclear receptor PXR (NR1I2) by direct binding of the warfarin molecule 10. Activated PXR may bind to the four putative PXR-binding sites 60 present in the 4-kb upstream promoter region of the mouse TG2 gene (our unpublished data, 2011). Direct activation of the TG2 catalytic activity by warfarin presents another level of regulation. These effects are specific for TG2, because other transglutaminases expressed in the TG2-/- arteries do not respond to warfarin treatment. The unique feature of the TG2 protein is allosteric inhibition of its catalytic activity 43, which may be released by the conformational changes upon warfarin binding to tyrosine residues (similar to the described interaction of warfarin with the vitamin K epoxide reductase complex subunit 1) 15. Yet another potential pathway, not shown in the figure, may involve inhibition of MGP function by warfarin leading to accumulation of TG2 in the vessel wall similar to the effect of MGP knockout 22, but the molecular mechanisms that may underlie this effect are not yet clear. At the next step, up-regulated and activated TG2 interacts with LRP5/6 receptors and stimulates canonical β-catenin signaling in VSMCs, leading to expression of osteoblastic genes and enhanced calcification.
It is possible that in vivo warfarin may also activate β-catenin in VSMCs through enhanced vascular BMP signaling, which is normally suppressed by γ-carboxylated MGP protein 61 but may be activated when warfarin inhibits vitamin K-dependent γ-carboxylation. While redox cues from immunity and inflammatory cells in the adventitia can also activate the BMP/Msx2-Wnt signaling in diabetic arteriosclerosis model 50, 61, in clonal VSMC culture these cells are absent, and warfarin-induced calcification and activation of theTG2/β-catenin axis is not coupled to oxidative stress or changes in Msx2 expression. It is tempting to speculate that the MGP-BMP pathway which mediates cross-talk between endothelium and VSMCs 61, the BMP/Msx2-Wnt cascade originating in the adventitia 49, and the TG2/β-catenin signaling axis which is central for osteogenic trans-differentiation in VSMCs, complement each other to mediate the activation of β-catenin and vascular calcification in diverse manifestations of this manifold disorder.
In vitro, warfarin acts as an amplifier rather than inducer of the osteochondrogenic trans-differentiation in VSMCs because it requires elevated levels of inorganic phosphate (Pi). The critical role of increased serum phosphate in vascular calcification is thoroughly supported by previous reports 31, 33, 35, 36. A new observation of our study is that warfarin augments destabilization of the VSMC phenotype induced by a modest increase in Pi from 1.2 to 1.6 mM, while earlier studies analyzed the effects of warfarin on arterial tissue in 2-3.8 mM Pi 5, 32. Taking into account that each 0.33 mM increase in serum phosphate within the “normal” clinical range increases the likelihood of arterial calcification by 34% 62, our findings indicate that warfarin-induced vascular calcification may substantially vary in the general population and can pose a significant risk in chronic kidney disease patients even in early stages of the disease.
In conclusion, this study provides the first direct evidence for the requirement of the canonical β-catenin pathway in vascular calcification, and newly identifies the important role of the TG2/β-catenin signaling axis in warfarin-induced vascular calcification in VSMCs, adding to the emerging list of pharmacotherapeutic targets in cardiovascular disease.
We gratefully acknowledge Dr. Chaitan Khosla for the TG2 inhibitors ERW-1069, CK-III-35 and KCC-009, and Dr. Dmitry Nurminsky for critical review and scientific insight.
Sources of Funding This study was supported by the National Institutes of Health (Grants R01HL093305 and R56DK071920 awarded to M.N.). K.B. is a postdoctoral fellow on T32AR007592.
Beazley, TG2/β-catenin axis in calcifying VSMCs
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.