|Home | About | Journals | Submit | Contact Us | Français|
Vascular calcification increasingly afflicts our aging, dysmetabolic population. Once considered only a passive process of dead and dying cells, vascular calcification has now emerged as a highly regulated form of biomineralization organized by collagenous and elastin extracellular matrices. During skeletal bone formation, paracrine epithelial-mesenchymal and endothelial-mesenchymal interactions control osteochondrocytic differentiation of multipotent mesenchymal progenitor cells. These paracrine osteogenic signals, mediated by potent morphogens of the BMP and Wnt superfamilies, are also active in the programming of arterial osteoprogenitor cells during vascular and valve calcification. Inflammatory cytokines, reactive oxygen species and oxylipids – increased in the clinical settings of atherosclerosis, diabetes, and uremia that promote arteriosclerotic calcification – elicit the ectopic vascular activation of osteogenic morphogens. Specific extracellular and intracellular inhibitors of BMP-Wnt signaling have been identified as contributing to the regulation of osteogenic mineralization during development and disease. These inhibitory pathways and their regulators afford the development of novel therapeutic strategies to prevent and treat valve and vascular sclerosis.
Arteriosclerotic calcification increasingly afflicts our aging and dysmetabolic population1. Vascular calcification was once thought to be a passive process of dead and dying cells. However, elegant studies by the bone and cartilage biologist H. Clarke Anderson in the 1970s identified alkaline phosphatase (ALP) - containing matrix vesicles as sites of atherosclerotic and medial vascular mineralization2, 3; this provided intriguing evidence that actively regulated osteogenic processes were participating in vascular mineralization. Another bone biologist and physician-scientist, Marshall Urist, had identified a “morphogenetic matrix for differentiation of bone tissue” earlier in that decade4, and in 1988 John Wozney, Vicky Rosen and colleagues cloned and characterized the first genes encoding bone morphogenetic proteins, including BMP25. In 1993, Boström, Demer and colleagues first demonstrated the expression of BMP2 in calcified human atherosclerotic plaques – and the capacity of BMPs to direct osteogenic programming of vascular mesenchymal progenitors of the pericyte lineage6. Thus, in a span of ~2 decades, a picture emerged in which the mineralizing pathobiology of vascular calcification is painted in part by secreted osteochondrogenic factors -- polypeptides also responsible for bone morphogenesis during skeletal development7 and fracture repair8.
Since 1993, the notion that osteochondrogenic morphogens participate in the biology of arterial calcification has been robustly confirmed and extended to encompass newly-discovered regulators of skeletal development9. In this manuscript, we review the regulation of valvular and vascular sclerosis by osteogenic morphogens. We highlight mechanistic insights afforded by these studies that provide the foundation for novel therapeutic approaches to treat arteriosclerotic valve and vascular disease.
Bone and cartilage are the two best recognized tissues of the vertebrate skeleton, the former arising in intimate association with the vasculature10. The skeleton contains two major anabolic cell types, the osteoblast and the chrondrocyte11. True bone formation or “ossification” occurs via two distinct mechanisms. Membranous ossification describes the mineralization that occurs in neural crest-derived craniofacial bones12, 13. Membranous ossification does not require a preceding cartilaginous template, but occurs directly within a type I collagen –based extracellular matrix elaborated and organized by osteoblasts. By contrast, most bones of the body mineralize via endochondral ossification. This second process requires an initial, avascular type II collagen-based extracellular matrix template deposited by chondrocytes. Subsequent chondrocyte hypertrophy at the growth plate, type X collagen deposition, VEGF-driven vascular invasion, chondrocyte apoptosis and cartilaginous matrix mineralization ensues. This is followed by osteoclast-mediated matrix resorption and osteoblast-mediated bone deposition (again, type I collagen based) that replaces the calcified cartilage template12, 13.
During skeletal development, paracrine epithelial-mesenchymal interactions control the osteogenic programming of skeletal anlage, best characterized during odontogenesis and endochondral ossification of the appendicular skeleton14. As first elucidated in studies of limb morphogenesis, proximal-distal, anterior-posterior, and dorsal-ventral morphogen gradients15 interact to control skeletal patterning, growth and ossification16. For example, reciprocal interactions between epithelial FGF4 / FGF8 in the apical ectodermal ridge (AER) and mesenchymal FGF10 maintain limb bud outgrowth. Ectodermal Wnt3a signaling is required for AER FGF8 production. Sonic hedgehog (Shh) production by posterior mesenchyme called the zone of polarizing activity (ZPA) maintains AER function via gremlin, a negative regulator of BMP signaling. Limb BMP4 tone provides a powerful ventralizing signal that is modulated by gremlin (preserves the AER) and by Wnt7a elaborated from the dorsal ectoderm. Dynamic expression of Dkk1 in ventral limb bud, anterior and posterior mesenchyme, and the AER antagonizes Wnt3a and Wnt7a signals that maintain FGF8 and Shh expression17. Thus, a complex 4-dimensional interplay between Wnt, BMP, FGF, Hedgehog, Dkk1 and other paracrine signals controls the growth, lineage specification, and differentiation of osteochondroprogrenitors in limb bud mesenchyme16. Likewise, paracrine PTHrP signals at the mineralizing growth plate control the temporospatial timing of endochondral bone formation via negative regulation of Indian Hedgehog (Ihh), and the FGF- and VEGF-dependent vascularization required for endochondral ossification18 as mentioned above. Again, in the craniofacial skeleton, membranous bones and teeth form without the requirement of a preceding calcified cartilage template13. Whether arising from endochondral or intramembranous processes, once true bone has formed paracrine BMP and endocrine PTH tone maintains osteogenic Wnt/β-catenin signals that support post-natal skeletal anabolism, mechanical integrity, and fracture repair8, 9, 19.
Sox9-positive mesenchymal condensations demarcate the skeletal morphogenetic field11, 20. Chondrogenic programming requires coordinated actions of L-Sox5, Sox621 and HIF22. Robust osteogenic programming of mesenchymal progenitors requires gene regulatory programs directed by at least six DNA binding transcription factors – Runx2 (a.k.a. Cbfa1), Msx1 and Msx2, LEF1, NFATc1, and Osx 11. β-catenin, a transcriptional co-adapter that promotes transcription directed by the TCF/LEF transcription factor family, is indispensible for osteoblast differentiation as well 23 and maintains the elaboration of osteogenic transcriptional programs (Figure 1). Runx2 11 has all the attributes of a ‘master gene’ differentiation factor for the osteoblast lineage and bone matrix gene expression in both membranous and endochondral bone. During embryonic development, Runx2 expression precedes osteoblast differentiation and is restricted to mesenchymal cells destined to become osteoblast 11(Figure 1). In addition to its critical role in osteoblast commitment and differentiation, Runx2 appears to control osteoblast activity, the rate of bone formation by differentiated osteoblasts, in concert with ATF411. In the skull, additional signals provided by the homeodomain proteins Msx1 and Msx2 are critical for intramembranous bone formation – a process that that cannot proceed even in the presence of an intact Runx2 gene when Msx expression is lacking24. Down-stream of Msx, Runx2, and β-catenin genomic programs is osterix (Osx), a zinc finger transcription factor required for matrix mineralization in concert with NFATc125 (Figure 1). Importantly, the expression of Msx1, Msx2, Runx2, and Osx during osteogenic differentiation is critically dependent upon bone morphogenetic protein (BMP) signaling11. Moreover, BMP2 signaling in mesenchyme maintains the specification of osteoprogenitors required for post-natal skeletal integrity8.
BMPs are secreted polypeptides, a subgroup of the transforming growth factor (TGF) β superfamily of growth factors. BMPs were first identified as a bioactivity in demineralized and pulverized bone powder capable of inducing ectopic endochondral bone formation in muscle26. Over the ensuing years, more than 15 distinct BMP family members have been identified (BMP system is extensively reviewed elsewhere)27, 28. The BMPs are secreted in an active form, with function modulated by extracellular antagonists27 such as noggin, chordin, gremlin, cross-veinless 2 (CV2, also referred to as BMPER), cartilage oligomeric matrix 2 (COMP2)29, and matrix Gla protein (MGP; Figure 2). BMPs elicit their effects through activation of receptor complexes composed of type I and type II Ser/Thr kinase receptors, to which the BMPs bind independently. There are seven type I receptors, termed activin receptors-like kinase (ALK1 to ALK7), which determine the specificity of the BMP signal in concert with five type II receptors including the BMP type II receptor (BMPR2) 30, 31. When the ligand binds, the type II receptors phosphorylate and activate the type I receptors, which propagate the signaling by phosphorylating transcription factors referred to as Smads (Figure 2). Receptor complexes may also contain so-called type III co-receptors 28 which further modulate signaling. The BMP type I receptors (ALK1, ALK2, ALK3 and ALK6) activate the receptor-activated (R)-Smads, Smad1, Smad5 and Smad8, whereas the activin and the TGFβ type I receptors (ALK4, ALK5 and ALK7) phosphorylate Smad2 and Smad331, 32. Activated Smads assemble into complexes together with the common mediator (co)-Smad4, which translocate to the nucleus and modulate gene expression in concert with Runx2, TCF/LEF and other factors. This process is regulated in part via inhibition by the inhibitory I-Smads, Smad6 and Smad7 32(Figure 2). BMPs can activate non-Smad signaling pathways as well. Studies by Rawadi, Baron and colleagues first identified one of these non-Smad signals to be a paracrine Wnt/β-catenin relay that promotes alkaline phosphatase expression and matrix biomineralization in conjunction with and parallel to BMP/Smad signaling 33, 34(Figure 3).
The Wnts are a family of secreted, lipid-modified polypeptide ligands35 that classically signal via activation of heterodimeric receptor complexes containing LRP5, LRP6, or LRP4 and a GPCR co-receptor of the Frizzled (Fzd) gene family36 (Figure 3). LRP co-receptor complexes with ROR2 are clearly also important, and reviewed in detail elsewhere37. In bone formation and skeletal homeostasis, Wnt10b, Wnt7a, Wnt7b, Wnt3a, and Wnt1 have emerged as important, although other ligands are certain to contribute34. With canonical or classical Wnt signaling mechanisms, nuclear accumulation of β-catenin ensues38. β-catenin is a transcriptional co-adaptor indispensible for Runx2- and LEF1- mediated programming of osteogenic differentiation23. Collaborative interactions with BMP- and TGFβ- regulated Smad complexes also occur39, 40. Even in the presence of skeletal osteochondrogenic transcription fractor expression (Runx2, Msx1/2, LEF1, Osx), robust elaboration of osteogenic programs is impossible in the absence of activated Wnt/β-catenin signaling23. The canonical pathway sequentially utilized a casein kinase 1 / dishevelled signaling cascade to phosphorylate GSK3 and inhibit the β-catenin destruction complex, thus increasing cellular β-catenin levels and nuclear accumulation41. Additionally, a “non-canonical” pathway also requiring dishevelled but proceeding via protein kinase C – delta activation has been identified to support Osx expression and bone formation42. In a manner analogous to the negative regulation of BMP signaling, secreted Wnt binding proteins known as secreted frizzled related proteins (sFRPs) inhibit Wnt ligand binding and activation of LRP5/6-Fzd cell surface receptors43. Two classes of non-signaling “ligands” -- members of the Dkk family44 and SOST/sclerostin45–47 – bind to the osteogenic LRPs and inhibit signaling (Figure 3). In addition, intracellular inhibitors of Wnt/β-catenin signaling such as ICAT (inhibitor of β-catenin and TCF/LEF) antagonize protein-protein interactions between β-catenin and nuclear transcription factor complexes bound to DNA48 (Figure 3). Thus, the sequential and parallel interactions between the BMP and Wnt signaling cascades control osteochondral mineralization 34, with intracellular and extracellular fine-tuning of signal duration and strength.
When BMP2, BMP4, and BMP6, were detected in calcified areas of atherosclerotic lesions 6, 49, 50 it was therefore presumed that they enhanced vascular calcification - even more so when it became evident that vascular calcification is largely driven by osteogenesis in the vascular media51, 52. Indeed, a causal relationship between BMP activity and vascular calcification has now been established53–55. However, BMP signaling is not only driving ectopic calcification, but is essential for cardiovascular development, with critical roles in the establishment of endothelial tubes during vasculogenesis, the recruitment and differentiation of vascular smooth muscle cell (VSMC) precursor cells, and vascular patterning56, 57. Mutations in the genes coding for the ALK1 receptor and endoglin, an ALK1 co-receptor, cause hemorrhagic hereditary telangiectasia 58, which is characterized by abnormal angiogenesis and arteriovenous malformations. Furthermore, mutations in the BMPR2 and ALK1 receptors have been linked to pulmonary arterial hypertension 59, 60, characterized by a dysregulation of vascular cell growth and differentiation.
BMP activity is important for the regulation of phenotypic plasticity, proliferation and differentiation in VSMC52. BMP2 in particular has an inhibitory effect on VSMC proliferation and differentiation, whereas BMP7 promotes the VSMC phenotype in a manner reminiscent of TGF-β 52. Furthermore, BMP inhibition -- potentially in later steps -- appears to be a key factor in maintaining VSMC differentiation. Loss of MGP, an inhibitor of BMP2 and BMP4 61–63, causes extensive calcification of elastic and muscular arteries 64. The loss of MGP results in a significant increase in aortic BMP activity and leads to osteochondrogenic trans-differentiation of VSMC with subsequent mineralization 65, supporting the notion that regulation of BMP activity is essential for maintaining a normal media. The potential anti-calcific roles of other BMP antagonists, such as noggin, CV2, gremlin, chordin-like 1, and cartilage oligomeric proteins 2 27 are not yet known although gremlin is over-expressed and may have anti-osteogenic effects in the calcified media in uremia66.
In addition to the aorta, MGP is also expressed in organs such as the lungs where it plays a role in the formation of the pulmonary arterial tree by inhibiting BMP4 67. BMP4 acts as an angiogenic factor and induces expression of ALK168, which is pivotal in endothelial maturation and recruitment of VSMC precursors. ALK1 activation in turn induces MGP, which provides negative feedback on BMP468. The interplay between BMP2, BMP4, ALK1, and MGP may affect morphogenetic processes that in some cases are linked to vascular calcification. BMP2 has been proposed to be an activation morphogen for calcifying vascular cells, promoting formation of large nodules and mineralization in vitro69, 70. The cellular patterns resulting from these interactions can be predicted using a reaction-diffusion model for the two types of morphogens69. In vivo, at least three types of patterns or morphogenetic processes might be affected by the balance between BMP and MGP or other inhibitors. The first process is the formation of the layered media, which includes normal VSMC differentiation. A disturbance in this process is exemplified by the abnormal and highly calcified media in the MGP null mouse64, 65. The two other processes, vascular branching and arteriovenous malformation (AVM) formation, have probably less to do with calcification but may present problems if the BMPs are targeted in anti-calcific therapies. Vascular branching is disturbed in the MGP transgenic mice, which exhibit loss of side branching and stunted growth of pulmonary arteries67. AVMs, on the other hand, are characteristic of ALK1 deficiency71 and can be described as a short-circuited and disrupted network in the capillary bed. Because ALK1 is regulated by BMP4 and in turn regulates MGP, the balance between BMP4 and MGP is likely to affect the capillary network. As highlighted by Rajamannan et al, osteogenesis and angiogenesis are coupled processes during vascular calcification72. Thus, the BMPs are true to their name in shaping the vasculature itself as well as inducing ectopic calcification.
An activated endothelium may be a significant source of BMP and a potential calcific stimulus. BMP expression is easily activated in the endothelium in response to a number of different pathogenic stimuli, many of which are well known to stimulate atherosclerosis and vascular calcification. Sorescu, Jo and colleagues were the first to demonstrate that BMP4 is induced in aortic endothelial cells (ECs) by oscillatory shear stress, reactive oxygen species (ROS), and inflammatory cytokines 53, 73, 74. Subsequently, Csiszar, Ungvari, and colleagues showed that BMP2 is similarly induced by inflammatory cytokines and ROS and functions as an inflammatory mediator75, 76. Both BMP2 and BMP4 are also induced by high glucose levels, whereas BMP4 is more stimulated by high fat diets than BMP253, 55. Even though BMP2 and BMP4 are highly homologous on a protein level, initial results point to a difference in function when secreted from glucose-treated ECs. BMP4 promotes angiogenesis and adipogenic differentiation whereas BMP2 promotes mineralization53, 77, thus suggesting that BMP2 has a more direct link to calcification and osteogenic differentiation.
Interestingly, BMP inhibitors are induced in ECs in response to similar stimuli as BMP2 and BMP4, or by an increase in vascular BMP activity. In vitro, follistatin, noggin, and MGP and CV2 are induced by oscillatory shear stress78 or high glucose levels 53. Most of these inhibitors are also induced in the aortic wall of fat-fed apoE null mice or diabetic mice and rats. 53, 55. Enhanced BMP activity causes induction of some inhibitors such as MGP and CV2, but not others such as noggin or chordin55, pointing to a finely tuned mechanisms for regulation of BMP activity. Interestingly, HDL, a vasculo-protective factor, promotes higher expression of MGP in ECs and aortas of apoAI transgenic mice79, thereby pushing the balance towards BMP inhibition. In addition, CV2 is induced by statins in ECs80, and also promotes BMP inhibition. Thus, the balance between endothelial BMPs and their inhibitors may determine the health of the endothelium, and an efficient limitation of endothelial BMP activity may confer protection against atherosclerosis. This also emphasizes the consideration of “BMP activity” or “BMP tone” as a context-specific working concept arising from the expression of specific BMP ligands, receptors or antagonists.
It should also be kept in mind that the adventitial microvasculature is also exposed to oxidative stress and inflammatory mediators similar to the aortic endothelium and likely responds with higher BMP2/4 secretion. In addition, microvascular pericytes may contribute multipotent cells to the calcific process 81–83. Thus, a calcific process could also be triggered from capillaries in the periphery of the vessel.
Vascular BMP activity is stimulated in cells other than ECs. BMP2 production is upregulated in SMC treated by high phosphate in the form of nanocrystals84, and in pericytes and mesangial fibroblasts by high glucose85. In addition, medial expression of BMPs, BMP receptors and inhibitors is increased in diabetic mice 86–88. A subset of myofibroblasts also expresses BMP2 in response to diabetes and hyperlipidemia 51, even though myofibroblasts have been reported to down-regulate BMP6 as a potential step in vascular remodeling89.
Endothelial BMP2 and BMP4 cause an induction of endothelial adhesion molecules in the ECs, in several instances using non-canonical signaling pathways 73, 75, 76. This provides a direct link between BMP activity and the initial stages of atherosclerosis, which sets the stage for atherosclerotic lesion development and ultimately lesion calcification. Once lesion formation gets under way, there may be local reactivation of developmental conditions where enhanced BMP activity could trigger osteogenesis in susceptible cells. Indeed, MGP, noggin and chordin are down-regulated in de-differentiated VSMCs54 which would make them more susceptible to BMP action. A number of BMP antagonists are later detected at increased levels in calcified vessels, and may in that context help limit the progression of the ectopic calcification.
As already discussed, the endothelium may have a role in triggering vascular calcification, and several populations of multipotent cells that respond to BMPs have been identified in the vasculature. These include the calcifying vascular cells (CVC) isolated from the aortic media 6, 90, myofibroblasts present in the adventitia of diabetic and hyperlipidemic LDLR−/− mice 51, and microvascular pericytes well known to have osteogenic potential81 (Figure 1). Circulating stem cells may also contribute to osteogenic changes in the artery wall 91, and VSMC “trans-differentiation” certainly contributes to the vascular endochondral ossification mechanisms in MGP−/− mice as elegantly demonstrated by Giachelli and colleagues65. However, the relative contribution of the respective cell population to local osteogenesis and vascular calcification in different arteriosclerotic disease states remains to be established.
The anti-calcific role of MGP may depend on normal EC differentiation. Recently, MGP transgenic mice were crossed with ApoE−/− mice, a model of atherosclerosis and lesion calcification, to study the effect of BMP inhibition on lesions. The increase in MGP resulted in diminished SMAD1/5/8 signaling, inflammation, lesion formation and calcification after fat-feeding 55. The majority of the MGP was found in proximity to the endothelium, supporting the importance of the endothelium in preventing lesions and lesion calcification. MGP null mouse on the same ApoE−/− background had enhanced SMAD1/5/8 signaling 92, and extensive medial calcification as previously reported 64, 65. However, the MGP null mice lacked lesion formation and the expected induction of endothelial adhesion molecules and inflammatory markers, suggesting a truly abnormal endothelium. Interestingly, ALK2 was induced in the MGP-deficient aortas, and endothelial differentiation has recently been linked to ALK2 activity. Constitutively active ALK2 causes endothelial-to-mesenchymal transition and acquisition of a stem-cell-like phenotype in ECs, which may be triggered to undergo osteogenic differentiation91. In patients with fibrodysplasia ossificans progressiva, characterized by heterotopic calcification and caused by mutations in the ALK2 gene91, chondrocytes and osteoblasts expressed EC markers. The implication for vascular calcification is that the endothelium might be an additional cell source of osteoblastic cells via endothelial – mesenchymal transitioning (EMT) 93 -- dependent upon Smads and β-catenin94. Strong ALK2 activation by exogenous factors might therefore interfere with endothelial differentiation and protective function. Capillary ingrowth in atherosclerotic lesions may also promote calcification analogous to normal bone mineralization 77. This may be an indirect way for BMP4, via its angiogenic properties, to enhance the calcification process.
Calcification is extremely common in diabetic vasculopathy, often in the form of medial calcification (also referred to as Mönckeberg’s media sclerosis or elastocalcinosis). The media calcification occurs along the elastic lamellae95 and frequently co-exists with atherosclerotic lesion calcification. The first link between BMP activity and diabetic vasculopathy was found in adventitial myofibroblasts in the aorta of fat-fed diabetic LDLR−/− mice 51. BMP2 was instrumental in augmenting the BMP-2/Msx2-Wnt pathway leading to an osteogenic phenotype in a subset of the myofibroblasts. It was later discovered that diabetes in mice and rats activated BMP signaling throughout the vascular wall 53. Interestingly, the location of BMP2 and BMP4 differed in diabetic aortas in that BMP-4 was found in the endothelium and BMP2 all the way through the vascular wall 53. The high BMP4 was associated with high MGP, ALK1, ALK2 and MGP, which were all regulated as a group in ECs. When a MGP transgene was introduced to limit the BMP activity, MGP expression increased predominantly in the endothelium, even though BMP activity and calcification diminished in the entire media53, suggesting crosstalk between the endothelium and the media.
Diabetes is a common cause of chronic kidney disease (CKD), another powerful stimulator of vascular calcification96. The calcification in CKD has been primarily attributed to increased phosphate levels, which strongly promotes SMC mineralization in vitro and in vivo97. High phosphate levels stimulate secretion of BMP2 in SMC84 and BMP2 in turn upregulates the type III sodium-dependent phosphate co-transporter PiT1/SLC20A1, which mediates high phosphate-induced mineralization98–100. Treatment with BMP7 counteracts vascular calcification in CKD, in part due to reduced phosphate levels and direct effects on SMC differentiation101. Therefore, it is important to understand all aspects of vascular BMP signaling in order to correctly target BMPs for anti-calcific effects.
A particularly severe and rapidly deadly form of vascular calcification – calcific uremic arteriolopathy (CUA, a.k.a. calciphylaxis) -- arises in a subset of patients with CKD treated with warfarin for anticoagulation102–104, and rarely in patients with severe chronic liver disease. Widespread calcium phosphate deposition in pulmonary, intestinal, and dermal soft tissues -- with predilection for the lower extremity, thighs and gluteal regions, and the abdominal pannus -- is characteristic for CUA. Histopathologically, arteriolar (< 0.6 mm diameter) medial calcification occurs -with concomitant fibroproliferative intimal vessel occlusion, dermal fat infarction, and skin necrosis. Immunohistochemistry demonstrates the expression of BMP4 in the periarteriolar adventitia of dermal CUA105. Warfarin inhibits the MGP gamma-carboxylation that is necessary for inhibition of BMP2/4 signaling62 as well as fetuin-dependent matrix vesicle clearance106, 107. Thus, it is highly probable that the warfarin-induced compromise in dermal arteriole MGP function contributes to the pathobiology of CUA. However, the precise reasons for the histoanatomic disease predilection remain to be explained – and other gamma-carboxylated proteins such as Gas6108 may also contribute to disease via Axl tyrosine kinase regulation of VSMC osteochondrocytic differentiation.
Calcific aortic valve disease (CAVD) afflicts ca. 2% of our population over the age of 60109. Progressive fibrosis and valve calcium accumulation impairs leaflet compliance and coaptation, impeding outflow, increasing myocardial workload, and permitting variable regurgitation. Once symptoms occur – e.g., chest pain, syncope, congestive heart failure – progressive aortic valve stenosis conveys a two-year mortality that approaches 50% in the absence of surgical intervention110. Using echocardiographic criteria, Rosenhek and colleagues demonstrated that the presence of aortic valve calcium in patients with asymptomatic mild or moderate aortic stenosis was the single most significant predictor of clinical progression111, 112.
Numerous epidemiologic studies identified risk factors for CAVD development, which are similar to those of vascular atherosclerosis, including smoking, male gender, body mass index, hypertension, elevated lipid and inflammatory markers, bicuspid aortic valve, type II diabetes mellitus (T2DM) and/or metabolic syndrome and renal failure113. Although aortic stenosis may occur in individuals with otherwise anatomically normal tricuspid aortic valves, congenital valve abnormalities markedly increase the risk. Importantly, nearly half of older individuals with aortic stenosis also have a bicuspid aortic valve (BAV), an aortic valve that develops with two functional leaflets instead of the normal three109, 114, 115. BAV occurs in about 2–3% of the population and is the most common congenital cardiac malformation. Although the causes for the development of BAV are unclear, genetic factors have been identified in some cases. BAV disease tends to progress more rapidly for reasons that are poorly understood. Genetic mutations associated with BAV that cause cellular dysfunction may also predispose an individual to other congenital heart defects or to dilation and dissection of the ascending aorta114.
Detailed clinical histopathology reveals valve inflammation as a common theme whether CAVD arises with BAV, hypertension, dyslipidemia, or any of the other risk settings highlighted above. Furthermore, the pathological lesions of calcified aortic valves demonstrate the complexities and heterogeneity of soft tissue mineralization. Calcium accrual in areas of fibrosis, necrosis, atherosis/lipid accumulation, hypertrophic cartilage and ectopic bone can all be observed116, 117. Otto and colleagues described the fibro-fatty expansion and inflammation of the lamina fibrosa, displaced and/or split elastic lamina, and intracellular and extracellular lipid accumulation with stippled calcification in the earliest of valve lesions118. With progression, woven bone formation – complete with marrow elements – and nodular amorphous calcium phosphate accumulates, the latter presumably via epitaxial mineral deposition on virtually acellular concretions. Nevertheless, the molecular “fingerprints” of active osteochondrogenic mineralization are present in all calcifying human aortic valve specimens -- even if overt ossification (ectopic bone replete with marrow) is not observed116, 119–121. Features of both membranous and endochondral osteogenic programs are elaborated by calcifying valve interstitial cells (VICs)121, with osteogenic potential highly dependent upon the stiffness of the extracellular matrix environment122. Additional osteoprogenitors may derive from chondrocytes originating at the base of the semilunar valve leaflet insertions123 and from valve ECs via EMT (see below). The frequent (~15%) ectopic bone formation and virtually omnipresent elaboration of osteochondrocytic programs in CAVD prompted evaluation for the potential contributions of Wnt/β-catenin signaling9.
The first evidence that activation of Wnt/β-catenin signaling participated in the calcification of human aortic valves came from quantitative western blot analysis comparing calcifying human tricuspid aortic valves vs. normal aortic valves and mitral valves undergoing myxomatous degeneration. Accumulation of β-catenin, the prototypic mediator and marker of canonical Wnt signaling necessary for osteoblast formation23, was upregulated 3.5- to 4-fold in calcifying tricuspid aortic valves as compared to normal aortic valves or myxomatous mitral valves119. Expression of the osteochondrogenic transcription factor Runx2/Cbfa1 was also increased along with the chondrocytic transcription factor Sox9 and the bone-specific matrix proteins bone sialoprotein and osteocalcin. Additionally, expression of the Wnt co-receptor LRP5 and the ligand Wnt3 were identified as being increased in calcified aortic valves as assessed by immunohistochemistry119. Subsequent studies by Miller, Heistad and colleagues confirmed the upregulation of Runx2 and Msx2 in calcifying valves, with overlapping yet distinct patters of expression. Analyses of preclinical models have provided mechanistic insights into the role of Wnt/β-catenin signaling in diabetes- and dyslipidemia- induced arteriosclerotic calcification121. Transgenic augmentation of aortic valve and vascular Msx2 expression upregulates multiple arterial Wnt ligand expression and increases in arterial calcium deposition. Implementing the TOPGAL mouse that possesses a LacZ-based reporter for endogenous activation of β-catenin dependent-transcription, concomitant upregulation of Wnt/β-catenin was observed in calcifying aortic valves as well as arterial tunica media124. Consistent with this, the capacity of VICs to undergo myofibroblastic conversion ex vivo is also dependent upon β-catenin signaling that synergizes with TGFβ40. The specific Wnt ligands and LRP receptors participating in valve calcification have yet to be characterized, although LRP5 and Wnt3 are potential candidates based upon evaluation of calcifying human valve specimens.
At this, point, the important role of the endothelium –including both the aortic and ventricular surfaces of valve endothelium -- – should be re-emphasized. The biology of the aortic valve is regulated by valve ECs adjacent to valve interstitial myofibroblast cells125. These ECs maintain the health of the valve and mediate valve disease in the presence of cardiovascular risk factors and/or genetic signals. In the skeleton, angiogenesis and osteogenesis are tightly coupled processes77. In keeping with this relationship, Rajamannan et al have provided evidence of osteogenic-angiogenic “coupling” in ossifying human rheumatic valves analyzed following removal for valve replacement surgery72. The full spectrum of paracrine signals participating in valve cell-cell interaction – and the signals that trigger activation and differentiation of VICs along the osteogenic lineage -- have not been fully established. However, BMP-Wnt signaling in endothelial integration of valve matrix metabolism and remodeling will continue to emerge as being important. As mentioned above, the inflamed endothelium elaborates BMP2 and BMP4. Triggers include hemodynamic shear stress, abnormal endothelial nitric oxide synthase function, inflammatory cytokines and growth factors, and the intracellular metabolic environment arising from dyslipidemia, hyperglycemia, and uremia. Moreover, as D’Amore and colleagues first noted, endothelial cells express LRP5 and LRP6, in addition to multiple Wnt ligands126, 127 – and platelet –derived LRP5/6 antagonist induces an inflammatory EC phenotype128. Intriguingly, valve endothelial cells have the capacity to undergo mesenchymal transitioning (EnMT or EMT),129 a process whereby cadherin-mediated cell-cell interactions are re-organized, endothelial differentiation programs are down-regulated, and myofibroblastic gene regulatory programs are activated. Upon acquiring the myofibroblast phenotype, valve ECs post-EMT have the capacity to undergo osteogenic differentiation93 – resembling in this fashion the emerging role of the circulating endothelial progenitor cell (ePC) as a vascular osteoprogenitor91, 130. During development, members of the TGFβ superfamily drive and coordinate valve EMT in concert with Msx gene family members and β-catenin dependent signals94, 131. Finally, ECs of the valve ventricular surface elaborate gene regulatory programs distinct from those of the aortic surface132. The molecular regulators control post-natal valve EC fates and phenotypes have yet to be fully characterized. Of note, both Smad2 and β-catenin signaling have been implicated in the epithelial-mesenchymal transitioning that contributes to myofibroblast load in idiopathic pulmonary fibrosis133. Lessons learned from these studies may help guide productive experimental approaches to valve sclerosis that occurs either in the presence or absence of calcium accrual.
When fed high fat diets (HFD) characteristic of westernized societies, male LDLR-deficient mice develop diet-induced obesity, dyslipidemia, and hyper-insulinemic diabetes characteristic of T2DM and the metabolic syndrome134, 135. Recent studies from MESA identify that these metabolic syndrome parameters convey risk for aortic and arterial calcification136. Consistent with this, male LDLR−/− mice on HFD develop progressively severe arteriosclerosis and fibrosis, with medial and atherosclerotic vascular mineralization134, 135, 137, 138. Early on, aortic upregulation of BMP2 and Msx2 is observed with subsequent Runx2 expression. Msx2-expressing adventitial myofibroblasts elaborate osteogenic Wnt3a, Wnt7a, and Wnt7b expression with concomitant paracrine activation of type I collagen and osterix gene expression, alkaline phosphatase activity, and osteogenic mineralization124, 137, 139, 140. Treatment with Dkk1, an inhibitor of LRP4/5/6 specifically down-regulated by Msx2, inhibits osteogenic mineralization by aortic myofibroblasts and other mesenchymal cell types124, 141.
As first predicted by Demer, Tintut, and colleagues142, inflammatory cytokine and oxidative stress cues appear to play an important role in the initiation of diabetic arteriosclerosis55, 137, 143. The prototypic inflammatory cytokine TNF, largely derived from adipose tissue macrophages, enhances oxylipid-dependent mineralization of calcifying vascular cells and adventitial myofibroblasts on rigid matrices such as tissue culture plastic. In addition to upregulating BMP2 production in endothelial cells, TNF directly activates arteriosclerotic Msx2 and Wnt/β-catenin signaling, the latter localized again to the tunica media with the TOPGAL reporter mouse137. Pharmacologic inhibition of endogenous oxidative stress signals generated by mitochondrial and Nox – family flavoenzymes inhibits TNF activation of myofibroblast Msx2 expression138. Recent data by others have confirmed that oxylipids such as oxLDL promote aortic VSMC mineralization in part via upregulation of osteogenic Msx2 signaling144–146, although the Wnt/Dkk and β-catenin responses were not specifically evaluated. Moreover, oxLDL upregulates BMP2 expression in vascular myofibroblasts via tolloid receptors TLR2 and TLR4147. Thus, in the vasculature osteogenic BMP-Wnt signaling is entrained to innate immunity and inflammatory redox cues that initiate and propagate tissue fibrosis with osteogenic differentiation of local progenitors. Moreover, ROS signaling enhances Runx2-directed trans-activation that drives and reinforces the osteochondrogenic phenotype of trans-differentiating VSMCs148 (Figure 1). It is tempting to speculate that the “diabesity” associated inflammatory redox signals that drive BMP-Wnt signaling may promote macrovascular EMT -- and thus further contribute to fibrosis, calcification, and mural thickening in diabetic arteriosclerosis. Novel strategies that combine inhibition of inflammatory cytokine signaling with enhanced egress of inflammatory oxylipids149 may serve to inhibit arteriosclerotic disease initiation and progression via vascular down-regulation of these osteogenic morphogens.
Selective modulation of vascular BMP activity is a promising target for treatment or prevention of arterial calcification. Decreased BMP activity resulting from excess MGP limits calcification 53, 55 and BMP type I receptor inhibition by small molecules has been successfully used to limit heterotopic calcification in mice with fibrodysplasia ossificans progressiva 150. However, there are still many questions that need clarification in the targeting and design of BMP inhibitors. For example, what cells would be the best targets, endothelial, smooth muscle or adventitial cells? What would the best molecular targets, ligands, receptors or inhibitors? Similar strategies can be envisioned that would target the Wnt signaling cascade. However, since post-natal bone homeostasis and skeletal integrity is dependent upon BMP-Wnt signaling, can strategies be identified that inhibit osteogenic vascular BMP-Wnt actions without mechanism-based toxicity in the skeleton? Similar issues arise when one considers many other mechanisms for modulating osteogenic morphogens in vascular disease. For example, Notch1 sensitizes VSMC to pro-osteogenic BMP2 signaling151 including the upregulation of alkaline phosphatase via Msx2-dependent actions 152 while inhibiting bone marrow osteoprogenitor differentiation153. This suggests that antagonism of Notch1 signaling might safely help ameliorate pathogenic BMP signaling in VSMC-mediated arterial calcification. Yet, Notch1 also inhibits osteochondrogenic signaling in calcifying aortic valve cells 154, 155. Moreover, decreases in Notch1 tone predispose to calcinotic bicuspid aortic valve formation155–157 and potential vascular tumor formation with hemorrhage158. Given the emerging difference in valvular vs. vascular Notch1 actions, it remains to be determined whether modulation of the Notch1 signaling pathways offer a significantly wide therapeutic window for pharmacologic intervention.
Alternatively, strategies can be envisioned that aggressively target the pivotal signals upstream of vascular BMP-Wnt activation to prevent disease initiation and slow arteriosclerotic progression. As discussed, inflammation, an abnormal oxidative stress environment, oxylipids stimulate the vascular endothelium to upregulate the secretion of BMPs important in osteogenic cell signaling 53, 75, 76, 138. These same signals down-regulate important mineralization defense mechanisms, including the degradation of inorganic pyrophosphate by alkaline phosphatase135, 159 and fetuin160. Furthermore, mechanical stretch plays a permissive role, allowing vascular osteoprogenitors to transition to a calcifying phenotype122. Thus, as contributors to pathobiology each of this processes are potential targets for pharmacologic intervention. HMG CoA Reductase agents, angiotensin converting enzyme (ACE) inhibitors, and angiotensin receptor blockers (ARBs), provide one approach. Indeed, in the Japanese Aortic Stenosis Study JASS, the use of ARBs was associated with reduced risk of disease progression in CAVD161. Interestingly, in preclinical models pulsatile hPTH(1–34) administration simultaneously reduces vascular 139and skeletal 162, 163oxidative stress signals – while simultaneously reducing vascular calcification and enhancing bone mass accrual140. Thus, novel anti-inflammatory ApoA1 mimetics such as D-4F164, 165 may help promote the egress of inflammatory oxylipids generated in the setting of diabetes and dyslipidemia in ways that simultaneously improve vascular and bone health149. However, in any one individual, certain pathophysiological stimuli (hyperphosphatemia, hyperlipidemia, hypertension, hyperglycemia) may be more critically important for intervention than others1 – and may differ with stage of disease. This may account for the disparate responses observed with respect to the impact of statin therapy upon CAVD166. With declining renal function, hyperphosphatemia plays an increasingly important role in vascular calcification167. Finally, once a significant amount of vascular calcium has accrued – or true ectopic ossification has occurred – a “point of no return” may be reached at which medical intervention by any means is no longer possible. Thus, careful patient characterization and selection will be required when testing known or novel therapeutic interventions to prevent or retard vascular calcification.
Sources of Funding
K.I.B. is supported by NIH grants HL081397 and HL030568. N.M.R. is supported by NIH grant HL085591. D.A.T. is supported by NIH grants HL069229, HL081138, HL088651, and by the Barnes-Jewish Hospital Foundation.
Publisher's Disclaimer: 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.