Genetics in Arterial Calcification

Abstract

Artery calcification reflects an admixture of factors such as ectopic osteochondral differentiation with primary host pathological conditions. We review how genetic factors, as identified by human genome-wide association studies, and incomplete correlations with various mouse studies, including knockout and strain analyses, fit into “pieces of the puzzle” in intimal calcification in human atherosclerosis, and artery tunica media calcification in aging, diabetes mellitus, and chronic kidney disease. We also describe in sharp contrast how ENPP1, CD73, and ABCC6 serve as “cogs in a wheel” of arterial calcification. Specifically, each is a minor component in the function of a much larger network of factors that exert balanced effects to promote and suppress arterial calcification. For the network to normally suppress spontaneous arterial calcification, the “cogs” ENPP1, CD73, and ABCC6 must be present and in working order. Monogenic ENPP1, CD73, and ABCC6 deficiencies each drive a molecular pathophysiology of closely related but phenotypically different diseases (generalized arterial calcification of infancy (GACI), pseudoxan-thoma elasticum (PXE) and arterial calcification caused by CD73 deficiency (ACDC)), in which premature onset arterial calcification is a prominent but not the sole feature.

A. How Heritable Is Human Artery Calcification?

In complex disorders, such as arterial calcification, multiple genes predispose to the phenotype, but each gene typically has small effects on the phenotype. Genetic factors, including some that mediate osteochondral differentiation, vascular cell growth, and inflammation, fit into “pieces of the puzzle” in intimal calcification in atherosclerosis and artery tunica media calcification in chronic kidney disease (CKD), diabetes, and aging (Figure 1). Among the limitations of studies on genetics in arterial calcification are methodologies in assaying vascular calcification, especially distinguishing intimal from medial calcification using current techniques, such as electron beam computed tomography. Also, because coronary artery calcification (CAC) is also a marker for plaque burden, genetic associations might also represent effects of genes underlying atherosclerotic disease independent of calcification.

Figure 1. Pieces of the “puzzle”: Host, environmental, and genetic factors contributing to human arterial calcification.

Although the inhibitory role of some of the specific factors with regard to arterial calcification has been proven, others may have promoting or direct or indirect “regulatory” effects. Some of the factors, including COMP¹⁷¹ and TGM2,¹⁷² are not explicitly discussed in the text.

In the Multiethnic Study of Atherosclerosis (MESA), traditional cardiovascular risk factors were associated with risk of incident coronary calcium deposition and increases in existing calcification,¹ but family CAD history was associated with CAC prevalence and amount in all ethnic groups in this cohort.² Accordingly, CAC prevalence and amount were heavily influenced by ethnicity in this cohort.^3,4 Forty-nine percent (P<0.001) of overall variance in abdominal aortic calcification was due to heritability in 1119 extended pedigrees from the original Framingham Heart Study.⁵ For CAC quantified by electron beam CT in 698 asymptomatic white adults, 43.5% of variation in CAC quantity was attributed to genetic factors.⁶ In coronary angiograms of 882 siblings with CAD from 401 German families, heritability of CAC was 0.51 (SEM, 0.17; P=0.001), comparable to heritability of proximal stenoses in left main CAD (h²=0.49; SEM, 0.12; P=0.001).⁷ Similarly, in 478 asymptomatic Amish adults, heritability for CAC quantity, adjusted for age and sex, was 0.42 (P=0.0001),⁸ and CAC was linked with lipid levels. It must be emphasized that heritability has been examined only for abdominal aortic⁵ and coronary atherosclerotic calcification.^6-8 Moreover, the imaging methods implemented cannot discriminate between atherosclerotic and media calcification. Thus, heritability for calcification of the artery intima and media may differ.

B. Specific Methods to Identify Genetic Factors Contributing to Arterial Calcification

1. Genome-Wide Association Studies

Four independent consortia identified a common locus on chromosome 9p21.3 as a major risk allele for coronary artery disease (CAD).^9-11 One of these consortia⁹ used DNA samples from the Dallas Heart Study (DHS), in which electron-beam CT-measured CAC was quantified¹² for validation of their findings, and the risk allele was associated with both CAC and premature atherosclerosis.⁹ In the more recent multi-ethnic Atherosclerotic Disease, Vascular Function and Genetic Epidemiology (ADVANCE) study, high-risk alleles were associated with increased CAC in all non-African race/ethnicity groups, with magnitude of these associations by racial/ethnic group closely mirroring magnitude for clinical CAD.¹³ The identified 58kb region on chromosome 9p21.3 adjacent to the tumor suppressor genes CDKN2A and CDKN2B does not contain any annotated gene or microRNAs,⁹ but a conserved sequence within the 9p21.3 locus has functional enhancer activity in smooth muscle cells (SMCs). RNA expression of the short variants of ANRIL, an antisense noncoding RNA in the INK4 locus, was increased by 2.2-fold, whereas expression of the long ANRIL variant was decreased by 1.2-fold in healthy subjects homozygous for the risk allele. Expression of these variants correlated with that of cyclin-dependent kinase inhibitor CDKN2B (p15) and TDGF1 (Cripto), respectively. Hence, the risk allele may promote atherosclerosis (and CAC) by modulating cell proliferation.¹⁴ In this respect, ANRIL regulates gene silencing by functionally coordinating with chromatin-associated factors that modify and interact with histone H3 lysine 27 methylation.¹⁵

CAC was also linked with chromosomal regions 6p21.3 and 10q21.3.¹⁶ LOD scores (log₁₀ odds in favor of linkage) were 2.22 (P=0.00070) for 6p21.3 (map position 76.4cM between markers D6S1053 and D10S2327) and 3.24 (P=0.000057) for 10q21.3 (map position 91.8 cM between markers D10S1423 and D10S2327). The region on chromosome 6 close to the marker DS1053 contains a group of collagen genes, BMP5, and IL17. The marker D10S189 at map position 19.0 cM on chromosome 10p14 gave the strongest signal (P=0.0012); candidate genes within this region include CALML5 (calmodulin-like 5) and CALML3 (calmodulin-like 3), which encode calcium-binding proteins.¹⁷

Genome-Wide Association Studies (GWAS) limitations for CAC are evident, as in other chronic diseases. Pinpointing genetic loci that confer susceptibility to artery calcification probably will require large-scale sample sizes and subdivision of precise phenotypes (ie, intimal and media calcification) to lessen heterogeneity. The next generation of human genetics studies for arterial calcification should include systems biology approaches. Genetic analyses of micro-RNAs probably will be a fruitful approach. Epigenetic approaches to arterial calcification are also ripe for development. The concept that there are racial differences between African Americans and other ethnic groups in gene expression profiles contributing to differential risks of artery calcification, merits further study.¹⁸

2. Candidate Gene-Based Association Studies

Candidate gene studies screening for small numbers of SNPs in cases and control subjects have been informative, though how some SNPs affect gene expression or protein function to modulate artery calcification remains unclear. Specific candidate genes have been selected due to influence on endothelial function and leukocyte adhesion and activation (CC chemokine receptor 2 CCR2, angiotensin-converting enzyme ACE, E-selectin ELAM1, epoxide hydrolase EPHX2), redox stress modulation (glutathione peroxidase-1 GPx-1), arterial extracellular matrix remodeling (matrix metalloproteinase 3 MMP3), bone metabolism (bone morphogenetic protein 7 BMP-7), and vitamin K metabolism that modulates γ-carboxylation of certain proteins in osteochondral differentiation such as matrix Gla protein (MGP) and osteocalcin (calumenin CALU, vitamin K epoxide reductase complex subunit 1 VKORC1).

Chemokine (C-C Motif) Receptor 2

The most studied chemokine (C-C motif) receptor 2 (CCR2) SNP is V64I, but there is both evidence of V64I effect on CCR function¹⁹ and lack of effect on CCR2 expression.²⁰ The SNP was linked with reduced CAC in one study²¹ but not in others.^22,23

Angiotensin-Converting Enzyme

One study on patients with documented CAD and coronary calcification suggests a role for the ACE insertion/deletion (I/D) polymorphism in intron 16 of the ACE gene in arterial calcification.²⁴ Patients with the DD genotype had significantly more calcified lesions. Another study on the ACE I/D polymorphism and coronary calcification was negative.²⁵

Epoxide Hydrolase 2

Epoxide hydrolases catalyze degradation of vasoactive epoxyeicosatrienoic acids into their corresponding dihydroxyeicosatrienoic acids.²⁶ Epoxyeicosatrienoic acids and dihy-droxyeicosatrienoic acids promote vasorelaxation of small arteries and display anti-inflammatory properties through the inhibition of nuclear factor-κB activation. Polymorphisms in EPHX2, the gene encoding soluble epoxide hydrolase, were associated with CAC in some^27,28 but not all studies.²⁹ Because of the low allele frequency for most of the SNPs²⁹ false-positive association is likely. Overall, effect size probably is small.

E-Selectin

The S128R E-selectin (ELAM1) polymorphism in the EGF-like domain of the protein can functionally alter leukocyte-endothelial interactions.³⁰ Association between E-selectin S128R polymorphism and CAC was reported in women 50 years of age or younger.³¹ However, no relationship was observed for men or older women.

Matrix Metalloproteinase 3

Matrix metalloproteinases (MMP) that degrade components of the extracellular matrix (ECM) including collagen and elastin, can modulate artery calcification, probably not only by calcification-promoting effects on ECM, but also, partly by regulating migration, growth and apoptosis of vascular cells. A MMP3 genotype was recently linked with higher promoter activity and CAC.³²

Vitamin D 24-Hydroxylase

Circulating 1,25 (OH)₂D is biologically inactivated through a series of reactions beginning with 24-hydroxylation.³³ The vitamin D 24-hydroxylase is encoded by vitamin D 24-hydroxylase (CYP24A1), and CYP24A1 is pivotal in vitamin D homeostasis. Deletion of Cyp24a1 in mice causes 1,25(OH)₂D excess and hypercalcemia with severe bone mineralization defects and ectopic calcification (renal calcium deposition) after chronic treatment with 1,25(OH)₂D.³⁴ On the other hand, transgenic rats that constitutively express Cyp24a1 develop atherosclerotic lesions in the aorta, which greatly progress with high-fat and high-cholesterol feeding.³⁵ A recent study found a common variant in the CYP24A1 gene associated with CAC quantity in 3 independent populations.³⁶ Interplay of transcriptional intermediary factor 1α (Tif1α) with vitamin D receptor target genes in mice, an essential native inhibitory system for mouse artery calcification,³⁷ is discussed below.

Bone Morphogenetic Protein 7

Mice unable to express the inhibitor of bone morphogenetic protein (BMP)signaling Smad6 develop arterial calcification.³⁸ BMP-7 appears anabolic for bone and protective against arterial calcification.^39,40 In vitro studies suggest that BMP7 promotes the vascular SMC phenotype⁴¹ and that receptors for the BMPs are expressed in vascular SMCs.^42,43 In type II diabetes, SNPs in BMP-7 were associated with arterial calcification.⁴⁴ Interestingly, this association was inversely correlated with bone mineral density. Defective BMP-7 functionality could lead to reduced osteoblastic differentiation in bone and perturbation of the bone-vascular axis.⁴⁴

Calumenin

The functions of calumenin (CALU), a protein encoded by the gene CALU, are connected with Ca²⁺-dependent processes. Calumenin endogenously regulates the activity of the γ-carboxylation system, consisting of vitamin K1 2,3-epoxide reductase (VKOR), and γ-glutamyl carboxylase.⁴⁵ Calumenin may have a role in atherogenesis, since it was localized in atherosclerotic plaques.⁴⁶ The CALU A29809G polymorphism was linked with CAC and prognosis in non-ST-elevation acute coronary syndrome.⁴⁷ The CALU 29809G allele was suggested to be protective for arterial calcification, contributing to better prognosis of carriers.

Vitamin K Epoxide Reductase Complex Subunit 1

Vitamin K epoxide reductase (VKOR) mediates recycling of vitamin K epoxide to vitamin K hydroquinone, an essential cosubstrate for γ-carboxylation of vitamin K-dependent proteins such as MGP and osteocalcin.⁴⁸ The inhibition of VKOR by warfarin results in undercarboxylation of MGP and subsequent medial calcification of the arterial vessel wall.⁴⁹ Numerous polymorphisms were identified in the vitamin K epoxide reductase complex subunit 1 (VKORC1) gene,⁵⁰ with one (1173C>T) associated with aortic calcification.⁵¹

Glutathione Peroxidase 1

Glutathione peroxidase-1 (GPx-1) protects against reactive-oxygen-species (ROS)-induced oxidative stress and to a lesser extent to reactive nitrogen species (RNS) in vivo.⁵² Alterations of GPx-1 have been linked to the etiology of cardiovascular disease and diabetes.⁵³ One study reported the common variant Pro197Leu of GPx-1 to be associated with 40% decreased activity.⁵⁴ This polymorphism is associated with CAC in type II diabetes.⁵⁵

3. Genetics of Murine Arterial Calcification

Mice are increasingly used to model arterial calcification and study risk factors. The degree of calcification differs among different strains, ranging from no aortic calcification in MRL substrains to 50% in DBA/2J strains on a chow diet. Even on atherogenic diet, incidence of aortic calcification in MRL mice (0%) was significantly lower than in C57BL/6J mice (24%). Hence, size and properties of murine atherosclerotic lesions, and susceptibility to calcification, are genetically determined.⁵⁶ The presence of aortic cartilaginous metaplasia within atherosclerotic lesions in the “susceptible” C57BL/6J strain but not the C3H/HeJ strain was informative because analysis of a genetic cross between these strains revealed a recessive inheritance pattern for this abnormality.⁵⁷

Age-related spontaneous dystrophic cardiac calcinosis (DCC) occurs in several susceptible inbred strains of mice, including BALB/c, DBA/2, and C3H.⁵⁸ Susceptibility to DCC was also observed after a standardized myocardial freeze-thaw injury, suggesting a common genetic basis.⁵⁹ By quantitative trait analysis, the gene determining DCC, designated dyscalc, was mapped to mouse chromosome 7.⁵⁸ The causal gene for dystrophic calcification was identified to be Abcc6,^60,61 whose implication in PXE is discussed below. Using an F2 mouse cross with 2 inbred strains of mice, C3/HeJ and C57/BL6, both null for the apoE allele as the breeders for the F1 generation, quantitative clinical trait loci (cQTL) causal for medial disruption and medial calcification were mapped to chromosome 7.⁶² The most likely candidate gene on this locus was Abcc6, subsequently confirmed to be causative for the phenotype by transgenic complementation.⁶² Also in this study, the hepatic expression quantitative trait locus (eQTL) for annexin A4 correlated with the cQTL for both medial calcification and medial disruption. Thus, annexin A4, a calcium and acidic phospholipid binding protein, which is a prominent component of mineralizing matrix vesicles from bone or vascular smooth muscle cells may also play a role in artery media calcification.⁶³

4. Identification of Candidate Genes for Human Arterial Calcification by Analyzing Natural Occurring Mouse Mutants or Targeted Mouse Mutants: Mixed Lessons for Human Artery Calcification

Characterization of targeted or naturally occurring mouse mutations of a variety of cartilage and bone proteins has led to the identification of 13 inhibitors of arterial calcification in vivo (Table). These studies have proven the concept that arterial calcification is normally inhibited by physiological function of resident arterial cells. Deficient expression of even one inhibitor of arterial calcification is sufficient to trigger the calcification process.^64-66 It is noteworthy that polymorphisms in the corresponding human gene of some of these inhibitors have been used as candidates for association studies in various populations, mostly with inconclusive results.

Table. List of Genes, Targeted Mutations of Which Lead to Artery Calcification in Mice With Respective Association Studies in Humans.

Gene Symbol, Human Protein Name, [OMIM Entry No.], Human Gene Map Locus	Mouse Model	Major Phenotypic Features	SNP-Based Association Studies in Humans	References
MGP, matrix Gla protein, [154870], 12p13.1-p12.3	mgp−/−	Arterial and cartilage calcification, tracheobronchial stenosis. Human correlate: Keutel syndrome	Increased CAC in men	73, 121, 122, 170
ENPP1, Ectonucleototide pyrophosphatase/ phosphodiesterase1, [173335], 6q22-q23	ttw/ttw enpp1−/−	Articular cartilage calcification, hyperostosis, spine and peripheral joint fusion, arterial calcification. Human correlate: Generalized arterial calcification of infancy	Increased aortic arch calcification in patients with type II diabetes Higher coronary calcification scores in ESRD patients	92, 93, 153, 154
ANKH, Progressive ankylosis [605145], 5p15.2-p14.1	ank/ank	Articular cartilage calcification, hyperostosis, spine and peripheral joint fusion, arterial calcification. Human correlate: Craniometaphysial dysplasia, Chondrocalcinosis-2		97
OPG, Osteoprotegerin, [602643], 8q24	opg−/−	Osteoporosis, vascular calcification	Increased risk for coronary artery disease	83-86
Smad6/Madh6, SMA- and MAD-related protein 6 [602931], 15q21-q22	madh6-mutant	Endocardial cushion defects, aortic ossification		38, 102
FBN1, Fibrillin1 [134797], 15q21.1	mgΔ/mgΔ, mgR/mgR	Aortic aneurysm, long bone overgrowth, artery media calcification. Human correlate: Marfan syndrome		103
CA2, Carbonic anhydrase2 [259730], 8q22	car-2−/−	Osteopetrosis, renal tubular acidosis, calcification of media of small arteries. Human correlate: Carbonic anhydrase II deficiency		99
KL, Klotho [604824], 13q12	klotho−/−	Vascular calcification, accelerated aging	Early onset of coronary artery disease	81, 82
AHSG, α2-HS-glycoprotein/fetuin, [138680], 3q27	ahsg−/−	Widespread but mild vascular calcification	Coronary artery calcified plaques in type II diabetes patients	79, 171
SPP1, Osteopontin, [166490], 4q21-q25	opn−/−	Enhanced calcification of cardiac valve implants		72, 113
ABCC6, ATP-binding cassette transporter subtype 6 [264800], 16p13.1	abcc6−/−	Myocardial necrosis and calcification, arterial calcification after freeze/thaw injury. Human correlate: Pseudoxanthoma elasticum		58, 60, 61
TIF1α, transcriptional intermediary factor 1 [603406], 7q32-q34	Tif1α−/−	Calcification of arterioles, arteries, vibrissae and lungs		37
ADIPOQ, Adiponectin [605441], 3q27	adiponectin−/−	Calcification of the aortic media in 30-week-old mice on chow diet		108, 110

Matrix Gla Protein

Biochemical and genetic studies have established matrix Gla protein (MGP) as an inhibitor of pathological calcification in cartilage and blood vessels,⁶⁷ as discussed elsewhere in this series of reviews. In mice, targeted deletion of the Mgp gene causes extensive ectopic chondrogenesis and lethal calcification of arteries.⁶⁷ In humans, in vitro studies suggest that polymorphisms in MGP are associated with altered promoter activity.^68,69 In addition, there is some evidence that MGP SNPs are associated with arterial calcification,^70,71 though these results are not consistent.^68,72 In one study, significant associations of polymorphisms in MGP and CAC were found in men only.⁷³

Fetuin-A

Fetuin-A (AHSG), a major systemic inhibitor of calcification, is discussed in the context of CKD in this series of reviews. Distinct AHSG1 and AHSG2 alleles are linked to different serum fetuin-A levels.⁷⁴ For example, in healthy Japanese volunteers, the different AHSG alleles were associated with decreased fetuin-A and increased free P_i levels.⁷⁵ Some polymorphisms in AHSG have been associated with decreased serum fetuin-A levels⁷⁶ and with different factors affecting arterial calcification, such as type II diabetes,⁷⁷ obesity,⁷⁸ and dyslipidemia.⁷⁷ Moreover, 4 different AHSG polymorphisms were associated with extent of coronary artery calcified plaque in type II diabetes.⁷⁹ Patients with end-stage renal disease (ESRD) carrying AHSG polymorphism T256S had lower fetuin-A levels compared with patients without this variant. Additionally, inflammation had an inhibitory effect on fetuin-A levels on patients carrying the T256S polymorphism. Thus, patients with ESRD and a genetic propensity for lower fetuin-A levels, for example, patients carrying the T256S polymorphism, have been hypothesized to have higher risk of cardiovascular premature death if subjected to chronic inflammation.⁸⁰

Klotho

Klotho (KL) is a β-glucuronidase that has a variety of effects, including regulation of insulin sensitivity and endothelial NO production. Klotho circulating levels decline in aging and in renal impairment. Defects in kl gene expression in mice result in a syndrome that is similar to human aging, including arteriosclerosis, osteoporosis, infertility, emphysema, and ectopic calcification,⁸¹ associated with increased production of vitamin D. Indeed, dietary vitamin D restriction inhibits development of arterial disease and premature aging in these mice. Studies in humans have identified functional variants of KL with CAD, HDL levels, systolic blood pressure, risk of stroke, and bone density, but no association was found in a study of arterial calcification.⁸²

Osteoprotegerin

Osteoprotegerin (OPG) is discussed elsewhere in this review series, and Opg knockout mice show severe osteoporosis with calcification in the aorta and renal arteries.^83-85 OPG gene polymorphisms were associated with increased risk for CAD and osteoporotic fracture,⁸⁶ but another study found no association with arterial calcification and CAD.⁸⁷ In contrast, polymorphisms in the OPG ligand RANKL were found to be associated with arterial calcification of the aorta in Korean women.⁸⁸

Transcriptional Intermediary Factor 1α and the Calcium-Sensing Receptor

Mice carrying an activating mutation of calcium-sensing receptor (Casr) demonstrate arterial calcification involving the media in arterioles and medium-sized arteries.⁸⁹ Transcriptional intermediary factor 1α (Tif1α) knockout mice develop similarly distributed spontaneous vascular calcifications³⁷ and Casr expression and demonstrate heightened renal expression of other vitamin D receptor (VDR) direct target genes (Car2, Cyp24a1, Trpv5, Trpv6, Calb1, S100g, Pthlh, and Spp1).³⁷ Therefore, TIF1α and the renal CASR and VDR pathway form a major physiological regulatory system that normally suppresses arterial calcification. CASR gene polymorphisms (A990G, C1011G) were not linked with cardiac valve calcification in one study.⁹⁰

ENPP1 (ttw/ttw) and ANK (ank/ank) Phenotypes

The tiptoe-walking mouse (ttw/ttw) had, for many years, been a model to study pathological calcification of perispinal ligaments. Okawa et al were able to identify a naturally occurring truncation mutation in Enpp1 (npps) encoding for ecto-nucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1)⁹¹ as the cause of the phenotype. The discovery of ENPP1 mutations in the human disease GACI by our group⁹² led to reappraisal of arterial media calcifications in this mouse model.⁹³ Enpp1 knockout mice recapitulate the phenotype of ttw/ttw mice and also share features of human GACI, including spontaneous periarticular soft tissue calcifications and artery media calcification adjacent to the internal elastic lamina in early life. However, the combination of perispinal and periarticular soft tissue ankylosis is lethal in Enpp1 knockout and ttw/ttw mice, whereas GACI is variably associated with scattered periarticular calcifications. Also, GACI is associated with frequently lethal myointimal proliferation with associated calcifications in large and mediumsized arteries, as opposed to nonlethal artery calcification as a less dominant feature of the phenotype in Enpp1-deficient mice.^94,95 ENPP1 is a major physiological generator of extracellular PP_i, a potent inhibitor of hydroxyapatite crystal formation and growth, and a molecule that exerts a variety of other, unexpected effects on cell differentiation and function.⁹⁶ A phenotype remarkably similar to the ttw/ttw mouse is found in the ank/ank mouse, carrying a C-terminal truncation mutation in the ANK protein, a transporter of PP_i, predominantly from the intracellular to the extracellular space.⁹⁷ Both the ttw/ttw and ank/ank mouse models have spurred insight into the role of PP_i metabolism in the context of arterial calcification (see below).

Carbonic Anhydrase-2

Carbonic anhydrase-2 (CA2) provides protons and bicarbonate ions to the local microenvironment and allows for robust acid secretion that leads to degradation of the organic matrix in bones.⁹⁸ Mice deficient in Car2 develop age-dependent medial calcification in the arterioles and smaller arteries in numerous organs. In male car2^−/− mice, the genital tract revealed most extensive arterial calcifications, and male mice were affected more than females, suggesting an effect of sex on vascular calcification in this mouse model.⁹⁹ In human beings, loss of function mutations in CA2 encoding carbonic anydrase-2 lead to osteopetrosis, renal tubular acidosis, and cerebral calcification.^100,101 However, arterial calcifications have not been linked with CA2 mutations in human beings.

SMA-Related and MAD-Related Protein 6

SMA-related and MAD-related protein 6 (SMAD6) plays a pivotal role in negative regulation of transforming growth factor-β (TGFβ) family signaling as a feedback molecule as well as a mediator of cross-talk with other pathways. Mice lacking Madh6, the mouse homologue of SMAD6, develop cartilaginous metaplasia and trabeculated bone within the arterial wall, recapitulating the complete process of endochondral ossification.³⁸ In a recent association study, the SMAD6 SNPs rs16950046, rs16950516 and rs266336 were significantly associated with sudden cardiac death.¹⁰²

Fibrillin-1

Fibrillin-1 (FBN1) appears to be essential for homeostasis of established elastic fibers and for cell adhesion involved in remodeling the artery ECM,¹⁰³ as shown in fibrillin-1 knockout mice. These mice reveal a predictable sequence of abnormalities including elastic fiber degeneration, excessive deposition of extracellular matrix proteins, elastolysis, intimal hyperplasia, and calcification.¹⁰³ Interestingly, vascular SMCs from fibrillin-1-deficient mice show an abnormal synthetic repertoire in early vascular lesions including elastin, among other matrix elements, and MMP-9, a known mediator of elastinolysis.¹⁰⁴ In this respect, elastin degradation is related to onset and progression of elastin calcification,¹⁰⁵ and this process is mediated by MMPs that include MMP-2 and MMP-9.¹⁰⁶

Fibrillin-1 mutations result in the pleiotropic manifestations of Marfan syndrome in humans, which is associated with joint hypermobility, aortic aneurysm formation and calcification, and long bone overgrowth.¹⁰⁷ To our knowledge, associations of FBN1 polymorphisms with other human arterial calcification phenotypes have not been defined.

Adiponectin

Adiponectin (ADIPOQ)¹⁰⁸ is an adipocyte derived hormone with antidiabetic, anti-inflammatory, and protective effects for the vasculature.¹⁰⁹ Adiponectin-deficient mice develop mild aortic calcification after being fed with a chow diet for 30 weeks.¹⁰⁸ The inhibitory effect of adiponectin on osteoblastic differentiation of vascular SMCs is mediated through the adiponectin receptor 1/p38 signaling pathway.¹⁰⁸ In a most recent study in 2847 participants of the MESA study cohort,¹ African Americans with ADIPOQ genotypes AG/GG of rs2241767 had 36% greater (95% confidence interval [CI], 16% to 59%, P=0.0001) CAC prevalence. Also, in African Americans, genotypes AG/AA of rs1063537 were associated with a 35% (95% CI, 14% to 59%; P=0.0005) greater CAC prevalence.¹¹⁰

Osteopontin (SPP1)

Osteopontin (OPN) is a major noncollagenous matrix protein of bone and a constitutive component of normal elastic fibers in the skin and aorta regulating chronic inflammation and vascular calcification.¹¹¹ OPN, like PP_i, potently inhibits hydroxyapatite crystal deposition and calcification by SMCs in vitro.¹¹² Although unchallenged Opn-knockout mice are grossly normal, enhanced calcification of implanted glutaral-dehyde-fixed aortic valve leaflets has been demonstrated.¹¹³ Likewise, increased arterial medial calcification was observed in Opn^−/−Mgp^−/− double knockout mice, supporting the concept that OPN is an inducible inhibitor of arterial media calcification.

Furthermore, in apoE^−/−/Opn-deficient mice vascular calcification was increased 2.5-fold in comparison to apoE^−/− mice.¹¹⁵ However, a study of the common T443C and Asp94Asp single nucleotide polymorphisms of SPP1 in human patients with coronary artery calcification failed to detect any statistically significant association.⁷²

5. Knockout Mouse Models Modulating Arterial Calcification Through Primary Deficiency of a Native Artery Calcification Regulator

The propensity of certain genes to promote arterial calcification has come to light also by studying certain mouse models, with knockout of different genes, which manifest artery calcification. Such genes include vanin-1 (VNN1), discoidin domain receptor-1 (DDR1), and RAGE.

Vanin-1

Vanin-1 (VNN1) is a glycosylphosphatidylinositol-anchored plasma membrane ecto-enzyme involved in cysteine and glutathione (GSH) metabolism. Vanin-1 promotes ectopic chondrogenesis in vitro, mediated by decreased cellular GSH stores.¹¹⁶ Increased chondrogenesis of ank/ank bone marrow stromal cells and increased chondrogenic transdifferentiation and calcification by ank/ank aortic smooth muscle cells and explants were corrected by Vanin-1 knockout in ank/ank Vnn1^−/− cross-breeding studies.

Discoidin Receptor 1

Discoidin receptor 1 (DDR1) is a receptor tyrosine kinase, which is activated by several subtypes of triple helical collagens. Through DDR1 signaling, type I collagen regulates proliferation, migration, and differentiation of many cell types. DDR1 serves as a positive regulator of neointimal formation and plays a pivotal role in regulating vascular inflammation and matrix accumulation, mediating the responses of macrophages and vascular SMCs. Mice deficient in Ddr1 and LDL receptor (Ldlr) demonstrate significantly reduced calcium content in the aortic arch, and attenuated SMC-mediated calcification of the extracellular matrix; SMCs derived from Ddr^−/− mice showed increased ENPP1 expression.¹¹⁷

Receptor for Advanced Glycation End Products RAGE (AGER)

RAGE serves as a cognate receptor not only for the advanced glycation end-products that accumulate in diabetic and aging tissues, and also for multiple S100 calgranulins known to be increased in diseased arteries. RAGE promotes atherosclerosis and diabetic vascular and neurological complications, and mediates chondrocyte maturation and osteoclast function, as well as calcification by SMCs. Aortic explants from Enpp1^−/− mice demonstrated decreased inorganic phosphate (P_i)-stimulated release of the RAGE inhibitor sRAGE, as well as increased calcification and intra-arterial chondrogenic differentiation, which was suppressed by Rage knockout in Enpp1^−/−Rage^−/− mice.¹¹⁸ In the apoE knockout mouse model of atherosclerosis and artery calcification, selective SMC expression of the RAGE ligand S100A12 increased medial calcification by 2- to 6-fold in the proximal aorta and innominate arteries, respectively.¹¹⁹ The pro-osteogenic propensity of the S100A12 transgene required conditioned media from lipid-challenged apoE null macrophages.¹²⁰ In this model, the actions of S100A12 were dependent on RAGE and oxidative stress signaling, since both recombinant sRAGE and NADPH oxidase inhibition reduced osteogenic programming and calcification.¹¹⁹ Evidence suggests a role for the human paracrine S100/RAGE axis in enhancing vascular calcification, and given that RAGE conveys arterial osteo-chondrogenic signals activated by PP_i deficiency¹¹⁸ and by oxidative stress,¹¹⁹ RAGE probably is involved in an amplification loop of arterial osteochondral differentiation in calcification.^118,120

6. Fitting Human Puzzle Pieces Together From Mouse Models of Arterial Calcification

Effects of knockout of specific mouse genes are not fully comparable to sequence variants of the same human genes. This is exemplified by the severe phenotype observed in Mgp-knockout mice, which show a generalized ossification of the whole arterial tree,¹²¹ whereas in the human correlate, Keutel syndrome, arterial calcifications are almost absent.¹²² One factor that probably plays a major role in regulatory differences between mouse and human arterial calcification is the relative hyperphosphatemia of normal mice compared with normal humans, for example, serum P_i levels of ≈8 mg/dL in mice, as compared with 2.5 to 4.5 mg/dL in normal humans.¹²³ Some of these mouse genes have been used as candidates for SNP association studies in populations enriched for arterial calcification (see above). Most of these studies so far revealed conflicting results as exemplified by the K173Q polymorphism in ENPP1, which has been studied in multiple ethnic groups and was demonstrated to be associated with coronary artery calcification only indirectly (see below). However, from the data presented above, it became obvious that some of the mouse candidate genes have proven to be valuable for association studies with respect to CAC and cardiovascular risk prediction in human beings: For example, presence of MGP polymorphisms affect CAC risk, especially in males.⁷³

7. Rare Human Monogenic Causes of Premature Onset Arterial Calcification

Positional cloning approaches and homozygosity mapping identified mutations in ABCC6 in pseudoxanthoma elasticum (PXE),^124-129 mutations in ENPP1 in GACI,⁹² and, most recently, mutations in NT5E in another rare disease phenotype consisting of peripheral artery calcification with distal joint calcification.¹³⁰

PXE and ABCC6 Mutations

PXE; OMIM 264800, 177850 is a rare autosomal recessive disorder characterized by dystrophic calcification of soft connective tissues including the dermis of the skin, the Bruch membrane of the eye, and the arterial media.^131,132 Positional cloning approaches provided strong evidence for linkage to the short arm of chromosome 16, and the critical interval was eventually refined by the use of informative recombinants to consist of approximately 500 kb.^125,126 Systematic sequencing of the candidate genes in this region resulted in identification of the ABCC6 gene as the one harboring mutations in PXE.^124,127-129 Human PXE exhibits selective medial calcification and degeneration without atherosclerotic involvement.¹³³ Accordingly, the Lusis group¹³⁴ recently demonstrated that in mice on a apoE null background the Abcc6 genotype is a significant contributor specific to media calcification and media disruption but not to atherosclerotic calcification or atherosclerotic lesion size. These studies provide genetic evidence of the pathobiological differences between artery media and atherosclerotic intima vascular calcification.

ABCC6, which is primarily expressed in the liver encodes for MRP6, a putative transmembrane protein, which belongs to the family of ATP-binding cassette (ABC) transport proteins.¹³⁵ Several studies were set out to identify the mechanisms how ABCC6 mutations lead to aberrant tissue mineralization. Based on previous studies demonstrating that activation of MGP by γ-carboxylation of glutamyl residues is critical for prevention of unwanted mineralization,¹³⁶ Li et al found reduced γ-glutamyl carboxylation of MGP in the serum, in the liver and in various calcified tissues of Abcc6^−/− mice.¹³⁷ Accordingly, patients with PXE carrying biallelic mutations in ABCC6, characteristically show uncarboxylated MGP in PXE skin lesions.¹³⁸ Whereas reduced γ-carboxylation of MGP is considered a major pathogenic factor contributing to the increased arterial mineralization in PXE, the physiological substrate of the MRP6 transport protein is still unknown. Ilias et al have shown that ABCC6 expressed in Sf9 insect cells was capable of transporting glutathione conjugates, including leukotriene C₄ and N-ethylmaleimide S-glutathione (NEM-GS), and organic anions were postulated to be the primary substrates.¹³⁹ However, studies in humans so far have not substantiated this hypothesis.

One school of thought is that absence of ABCC6 activity primarily in the liver results in the deficiency of circulating factors (including partial deficiency of fetuin), which leads to loss of physiological suppression of artery calcification under normal calcium and phosphate homeostatic conditions.¹⁴⁰ In this context, overexpression of fetuin in Abcc6^−/− mice was effective in reducing soft tissue mineralization in these animals.¹⁴¹ The finding that when muzzle skin from wild-type mice was grafted onto the back of knockout mice, mineralization of the vibrissae was observed within the subsequent 2 months, also supported this hypothesis.¹⁴² Furthermore, serum from PXE patients, when added to tissue culture medium of fibroblasts, has been shown to alter the expression of elastin.¹⁴³ It was postulated that ABCC6 might participate in transmembrane transport and redistribution of vitamin KH₂, an obligatory cofactor of γ-glutamyl carboxylase,¹⁴⁴ especially when conjugated to glutathione. However, most recently it was proven that this was not the case.¹⁴⁵

GACI and ENPP1 Mutations

In GACI, OMIM No. 208000, calcification of the internal elastic lamina is associated with myointimal proliferation in medium and large-sized arteries (Figure 2).¹⁴⁶ The disease phenotype is often associated with periarticular calcification (Figure 2). Affected patients frequently die within the first weeks of life as the result of heart failure and myocardial infarction.^147,148 Based on the observation that an affected child had a systemic deficiency of PP_i,¹⁴⁶ we found decreased plasma levels of the PP_i generating enzyme ENPP1.¹⁴⁹ In a candidate gene approach, we identified homozygosity or compound heterozygosity for “loss of function” mutations in ENPP1 as the cause for this rare disorder. To date, more than 40 different causative mutations have been identified in GACI patients, accounting for more than 75% of the affected cases.¹⁴⁸ ENPP1 expression is regulated during atheromatous plaque calcification, suggesting a protective role as an inhibitor of vascular mineralization by generating extracellular PP_i also in the context of atherosclerosis.¹⁵⁰ On the other hand, as shown most recently by cross breeding mice deficient for both apoE and Enpp1, Enpp1 promotes atherosclerotic lesion progression,¹⁵¹ probably mediated in part by promotion of OPN expression.

Figure 2. Typical manifestations of generalized arterial calcification of infancy (GACI) caused by mutations in ENPP1.

A, Cardiac ultrasound (suprasternal long-axis view) of a 4-week-old boy; note increased echogenicity of the wall of the ascending aorta (arrow-heads). B, Radiograph of the left hand of a 2-week-old boy showing periarticular calcifications (arrow). C, Cross section of the left coronary artery from a boy who died of myocardial infarction at the age of 9 weeks, showing calcification at the level of the internal elastic lamina (arrows) and intimal proliferation (hematoxylin and eosin staining). Ao indicates aorta, LA; left atrium.

Hypophosphatemia can compensate for GACI, and it has been argued that this reflects a physiological compensation mechanism rather than a primary defect.¹⁴⁸ However, several mutations in the ENPP1 gene result in a phenotype with complete dissociation between GACI and autosomal-recessive hypophosphatemic rickets (ARHR),^152,153 suggesting a different pathway involved in the generation of ARHR linked to direct renal P_i-modulating functions of ENPP1.

Studies investigating an association of ENPP1 polymorphism K173Q with arterial calcification revealed conflicting results (linked with type II diabetes¹⁵⁴ and ESRD¹⁵⁵ but not in other patient cohorts¹⁵⁶). A noteworthy confounder is that ENPP1 variants differentially modulate insulin receptor signaling¹⁵⁷ and K173Q and other SNPs in ENPP1 may have an indirect effect on parameters affecting arterial calcification in certain populations.

Arterial Calcification and Distal Joint Calcification and NT5E Mutations

Recently, using a genome-wide homozygosity mapping approach, NT5E, the gene for familial peripheral arterial calcification and distal joint calcification (OMIM No. 211800), was identified in a large consanguineous pedigree.¹³⁰ The phenotype, notably, has a relatively late onset in young adulthood, and the symptomatic arterial disease, with prominent calcification, affects large lower extremity arteries, but remarkably, spares the coronary circulation.¹³⁰ The authors detected biallelic nonsense, missense, and single-nucleotide insertion-frameshift mutations in the ecto-5′-nucleotidase gene NT5E, encoding the GPI-linked plasma membrane CD73 ecto-enzyme, which generates extracellular adenosine, directly downstream of ENPP1 in the extracellular ATP-degradation pathway. In their study, reduction in extracellular adenosine levels enhanced tissue nonspecific alkaline phosphatase (TNAP) activity, and adenosine supplementation reversed the increase in TNAP activity in CD73-deficient cells. It is possible that loss of the capacity of adenosine to suppress vascular inflammation and neointima formation contributes to the arterial disease phenotype in arterial calcification and distal joint calcification (ACDC). Because CD73 deficiency contributes to arterial calcification only in specific vascular territories, namely iliac, femoropopliteal, and tibial arteries, it is tempting to speculate that the related family member CD39 and its isoforms may compensate in an imperfect, functionally redundant fashion in other vascular beds, for example, CDL39L2 in the heart.^158,159

C. “Cogs in the Wheel”: Model of Unified Molecular Pathophysiology for 3 Heritable Monogenic Deficiency States of “Cogs” That Prevent Spontaneous Arterial Calcification

Primary monogenic deficiencies of ENPP1, CD73, and ABCC6 serve as “cogs in a wheel.” Specifically, each is a minor component in the function of a much larger network of factors (Figure 3) that exert finely balanced promotion and suppression of arterial calcification. For the network to suppress spontaneous arterial calcification, the “cogs” ENPP1, CD73, and ABCC6 must be present and in working order. Monogenic ENPP1, CD73, and ABCC6 deficiencies each drive a molecular pathophysiology of closely related but phenotypically different human diseases (GACI, PXE, and arterial calcification due to CD73 deficiency [ACDC]), in which spontaneous, premature onset of arterial calcification is a prominent but not sole feature.

Figure 3. Model of an SMC functional network revealed by 3 human monogenic diseases of artery calcification (GACI due to ENPP1 deficiency, PXE due to ABCC6 deficiency, and ACDC due to CD73 deficiency).

This review describes how ENPP1, CD73, and ABCC6 serve as “cogs in a wheel.” Specifically, each is a minor component in the function of a much larger network of factors, illustrated here, that exert balanced effects to promote and suppress arterial calcification. For the network to normally suppress spontaneous arterial calcification, the “cogs,” ENPP1, CD73, and ABCC6, must be present and in working order. Monogenic ENPP1, CD73, and ABCC6 deficiencies each drive a molecular pathophysiology of closely related but phenotypically different diseases (GACI, ACDC, and PXE, respectively), in which premature onset arterial calcification is a prominent but not sole feature. The transmembrane ectoenzyme ENPP1 generates AMP and PP_i from ATP, which are hydrolyzed by the GPI-linked ecto-enzymes CD73 (5′exonucleotidase) and TNAP, to generate adenosine and PP_i, respectively. PP_i suppresses hydroxyapatite deposition and inhibits ectopic chondrogenesis (and modulates artery calcification by other effects). ENPP1 modulates RAGE expression, as discussed in the text, and PP_i limits promineralizing type I collagen. The PP_i transporter ANKH, as well as ENPP1, elevates extracellular PP_i, and murine deficiency of this transporter is associated with arterial calcification, driven in part by vanin-1 pantetheinase expression, which depresses cellular GSH stores and promotes ectopic chondrogenesis. Adenosine signaling suppresses TNAP expression and inhibits vascular inflammation and neointima formation. ABCC6 has been hypothesized, though not yet proven, to modulate adenosine transport. P_i signals, in part through uptake, mediated the Na⁺-dependent P_i cotransporter¹⁷³ to promote osteochondral differentiation, and P_i is a component of hydroxyapatite crystal deposition. In addition, systemic effects of hepatic ABCC6 deficiency, including circulating fetuin deficiency, appear to modulate arterial and skin features of PXE (illustration credit: Cosmocyte/Ben Smith).

Prior models for unified understanding of mechanisms of ectopic artery calcification focused on systematic regulatory influences of PP_i and P_i metabolism, mediated by ANKH, ENPP1, and TNAP on expression of OPN.¹⁶⁰ ENPP1, through PP_i generation, promotes OPN expression¹⁶¹; OPN is proatherogenic¹¹¹ and probably mediates proatherogenic effects of ENPP1.¹⁵¹ However, OPN, in part by suppressing hydroxyapatite crystal deposition, potently inhibits calcification by cultured SMCs and suppresses artery calcification in vivo.^114,162 Figure 3 schematizes the linkages between PP_i and P_i metabolism and adenosine signaling that appear to intimately link GACI due to ENPP1 deficiency, PXE due to ABCC6 deficiency, and ACDC due to CD73 deficiency. Other factors in this paradigm include secondary changes in TNAP effects,⁹⁵ because TNAP upregulation in SMCs, and consequent degradation of the artery calcification inhibitor PP_i and generation of the critical artery calcification promoter P_i, at sites of fibrillar type I collagen expression, plays a substantial role in driving a variety of models of artery calcification in vitro, and in experimental artery calcification in vivo.^123,163-166 TNAP activity reduces PP_i levels, and, in part, thereby antagonizes ENPP1 in the ectopic calcification process.⁹⁴ Regulation by ENPP1 and PP_i of OPN and type I collagen expression are central components in this functional system,^161,167 and OPN might further promote artery calcification by suppressing ENPP1 and ANKH expression.¹⁶⁸ TNAP generates procalcifying P_i by hydrolyzing both PP_i and ATP, and TNAP is a central, physiological player in osteoblast and chondrocyte maturation and normal calcification of the skeleton.⁹⁴ In this model, because the histopathologic phenotype of the arterial lesions with fragmented and broken elastic fibers in the arterial media of ACDC patients closely resembles the arterial lesion in PXE patients, it has been speculated that both PXE and ACDC share common pathogenetic pathways and that adenosine transport may be modulated by ABCC6.¹⁶⁹

Conclusion

Genetic factors clearly contribute to variation in amounts of media and intimal arterial calcification; this review did not focus on genetics in aortic valve calcification. GWAS have substantially increased our knowledge on risk stratification in CAC and atherosclerosis. However, functional significance of genes contained in the identified loci remains uncertain in most cases, especially since mutations in the respective candidate genes have not yet been detected in the affected patients. Candidate gene studies based on SNPs have more or less failed to provide reproducible results confirming genetic risk factors of arterial calcification in different human populations. We believe that systems biology approaches are the natural next step in identifying further hereditable contributing factors, along with large scale whole exome sequencing studies in at risk populations.

The analysis of naturally occurring or targeted mutant mice has substantially enhanced our knowledge on pathophysiologic mechanisms of arterial calcification, but knockout of specific mouse genes is not fully comparable to sequence variants of the homologous human genes, with functionally variable consequences for the gene products. Some of the mouse candidate genes have proven to be valuable for association studies, with respect to coronary artery calcification and cardiovascular risk prediction in human beings, which holds true for MGP, ASHG, and RANKL.

Major insight has been provided through rare monogenic human disorders associated with spontaneous, premature artery media calcification, (GACI, PXE, ACDC). The underlying disease genes, ENPP1, ABCC6, and NT5E, respectively, appear to be natural “cogs” that prevent spontaneous calcification within an arterial molecular pathophysiology network modulated by ATP metabolism, PP_i, adenosine, and P_i generation. Roles of ENPP1, ABCC6, and NT5E in the context of atherosclerosis are now emerging. Nevertheless, how genetic factors fit into the pathogenetic puzzle in most instances of artery intima and media calcification remains to be defined

Non-standard Abbreviations and Acronyms

ACDC

arterial calcification due to CD73 deficiency

ACE

angiotensin-converting enzyme

CAC

coronary artery calcification

CAD

coronary artery disease

CKD

chronic kidney disease

DCC

dystrophic cardiac calcinosis

ECM

extracellular matrix

ESRD

end-stage renal disease

GACI

generalized arterial calcification of infancy

GWAS

genome-wide association studies

P_i

inorganic phosphate

PP_i

inorganic pyrophosphate

PXE

pseudoxanthoma elasticum

SERMs

selective estrogen receptor modulators

SMC

smooth muscle cells

sRAGE

soluble RAGE

TNAP

tissue nonspecific alkaline phosphatase