help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Doherty, T. M.
Right arrow Articles by Rajavashisth, T. B.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Doherty, T. M.
Right arrow Articles by Rajavashisth, T. B.
Endocrine Reviews 25 (4): 629-672
Copyright © 2004 by The Endocrine Society

Molecular, Endocrine, and Genetic Mechanisms of Arterial Calcification

Terence M. Doherty, Lorraine A. Fitzpatrick, Daisuke Inoue, Jian-Hua Qiao, Michael C. Fishbein, Robert C. Detrano, Prediman K. Shah and Tripathi B. Rajavashisth

Atherosclerosis Research Center and Burns and Allen Research Institute (T.M.D., P.K.S., T.B.R.), Division of Cardiology, Department of Medicine, Cedars-Sinai Medical Center and David Geffen School of Medicine at the University of California Los Angeles (UCLA), Los Angeles, California 90048; Division of Endocrinology, Diabetes, Metabolism, and Nutrition (L.A.F.), Department of Internal Medicine, Mayo Clinic and Mayo Foundation, Rochester, Minnesota 55905; Division of Endocrinology and Metabolism (D.I.), Department of Medicine and Bioregulatory Sciences, University of Tokushima Graduate School of Medicine, Tokushima 770-8503, Japan; Department of Pathology and Laboratory Medicine (J.-H.Q., M.C.F.), David Geffen School of Medicine at UCLA, Los Angeles, California 90095; and Division of Cardiology (R.C.D.), Department of Medicine, Harbor-UCLA Research and Education Institute, Torrance, California 90512

Correspondence: Address all correspondence and requests for reprints to: Tripathi B. Rajavashisth, Ph.D., Atherosclerosis Research Center, Davis Research Building, Room 1062, Division of Cardiology, Department of Medicine, Cedars-Sinai Medical Center and David Geffen School of Medicine at University of California, Los Angeles, 8700 Beverly Boulevard, Los Angeles, California 90048-1865. E-mail: rajavashisth{at}cshs.org


    Abstract
 Top
 Abstract
 I. Introduction: Clinical...
 II. Types of Arterial...
 III. Molecular Mechanisms of...
 IV. Endocrine, Pharmacological,...
 V. Genetic Determinants of...
 VI. Conclusions
 References
 
Pathologists have recognized arterial calcification for over a century. Recent years have witnessed a strong resurgence of interest in atherosclerotic plaque calcification because it: 1) can be easily detected noninvasively; 2) closely correlates with the amount of atherosclerotic plaque; 3) serves as a surrogate measure for atherosclerosis, allowing preclinical detection of the disease; and 4) is associated with heightened risk of adverse cardiovascular events. There are two major types of calcification in arteries: calcification of the media tunica layer (sometimes called Mönckeberg’s sclerosis), and calcification within subdomains of atherosclerotic plaque within the intimal layer of the artery. There are important similarities and differences between these two entities. Of particular interest are increasing parallels between cellular and molecular features of arterial calcification and bone biology, and this has led to accelerating interest in understanding how and why bone-like mineral deposits may form in arteries. Here, we review the two major pathological types of arterial calcification, the proposed models of calcification, and endocrine and genetic determinants that affect arterial calcification. In addition, we highlight areas requiring further investigation.

I. Introduction: Clinical Perspective
II. Types of Arterial Calcification
A. Medial calcification
B. Intimal calcification

III. Molecular Mechanisms of Arterial Calcification
A. Molecular and genetic mechanisms of bone formation
B. The active model of arterial calcification
C. The potential role of OLCs in arterial calcification
D. The passive physicochemical model of arterial calcification
E. A unified model

IV. Endocrine, Pharmacological, and Lipid Influences on Arterial Calcification
A. Parathyroid hormone and vitamin D
B. Estrogen
C. Pharmaceutical agents that regulate calcification
D. Role of lipids

V. Genetic Determinants of Arterial Calcification
A. Angiotensin-converting enzyme
B. Apolipoprotein E
C. E-selectin
D. Matrix metalloproteinase-3 (MMP-3)
E. Matrix Gla protein
F. CC chemokine receptor 2
G. Estrogen receptor-{alpha}
H. Ecto-nucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1)
I. Epoxide hydrolase
J. Osteoprotegerin
K. Other human genetic disorders and arterial calcification

VI. Conclusions


    I. Introduction: Clinical Perspective
 Top
 Abstract
 I. Introduction: Clinical...
 II. Types of Arterial...
 III. Molecular Mechanisms of...
 IV. Endocrine, Pharmacological,...
 V. Genetic Determinants of...
 VI. Conclusions
 References
 
DESPITE TREMENDOUS ADVANCES in our understanding and treatment of cardiovascular disorders, coronary heart disease (CHD) now accounts for most death and disability in developed countries, exacts a financial toll approaching $500 million annually in the United States alone (1, 2), and will soon eclipse infection as the major cause of all morbidity and mortality (3, 4). Yet the presence of atherosclerotic plaque—the pathobiological substrate upon which the overwhelming majority of CHD events occurs—does not of itself lead irrevocably to poor outcomes (5). Atherosclerosis is characterized by chronic arterial inflammation instigated and exacerbated by disordered lipid metabolism and other well-characterized risk factors (6, 7, 8, 9). Large postmortem studies have revealed that atherosclerosis begins surprisingly early and is ubiquitous in middle-aged and older adults (10, 11) (Fig. 1Go), yet epidemiological studies have paradoxically shown that despite the very high prevalence of atherosclerosis in adults, most will never suffer a clinical event from the disease and will ultimately die of other causes (5, 12, 13). The resolution of this paradox came from seminal investigations that elegantly established that the majority of devastating CHD events such as myocardial infarction are directly precipitated by rupture or erosion of unstable plaque, and it is the composition rather than the amount or size of plaque that critically determines instability (14, 15, 16). These startling findings lead to the disquieting realization that individuals with fairly extensive atherosclerosis may nonetheless lead normal event-free lives if their disease remains structurally quiescent, but those with relatively little atherosclerosis can and do suffer devastating CHD events if they have plaque that becomes vulnerable to rupture. Accordingly, the focus of vascular biologists and clinical scientists has now shifted toward understanding how plaque composition is determined and why and how it becomes vulnerable.



View larger version (99K):
[in this window]
[in a new window]
 
FIG. 1. The spectrum of atherosclerosis from the cradle to the grave. Fatty streaks, neointimal thickening, and raised dots or plaques on the arterial surface can appear during infancy. A, Scanning electron micrograph of an intact endothelial layer over a raised intimal lesion. B, Photograph of raised fatty dots on the surface of the aorta from a 5-yr-old child obtained at autopsy. Plaques grow from within the artery, but because the artery may undergo progressive compensatory dilatation (C), blood flow may remain unimpeded unless the majority of the cross-sectional area becomes diseased and the artery can no longer dilate further. The lesion shown in panel C (stained with oil red O, lipid appears red) appeared normal on angiography despite significant plaque. Often without warning, a plaque surface may tear or become eroded, exposing thrombogenic arterial components to flowing blood. D, A large thrombus adhering to the luminal arterial surface that was due to endothelial erosion or tearing of the plaque cap. Note that the plaque is intact and the degree of stenosis is only mild. Such plaques are not severe enough to compromise blood flow and are therefore typically asymptomatic, but they can nonetheless be lethal. Small blood vessels within the plaque may also hemorrhage into an otherwise structurally intact plaque. If intraplaque bleeding is extensive (E), the luminal diameter of the artery may be compromised, causing ischemia. F, Complete thrombotic occlusion has occurred and is associated with a large, lipid-filled lesion. G, A similar plaque. Again, total thrombotic occlusion of the artery has occurred (green arrow). Note that there is also a small intraplaque hemorrhage (blue arrow). The plaque cap appears to be intact, suggesting that thrombotic occlusion of this artery occurred by a mechanism that did not involve outright structural failure of the plaque cap. In such types of thrombosis, platelet-rich thrombi may form on damaged mural surfaces and then either resolve spontaneously or develop into much larger clots that occlude the lumen of the artery. H, A plaque whose overlying cap has obviously ruptured (white arrows). When arterial blood comes into contact with subendothelial tissues in the artery wall, thrombosis very rapidly forms, and it may fill the plaque and continue to propagate into the lumen of the artery, causing an acute coronary event such as unstable angina, myocardial infarction, or sudden death. Note that this plaque also is associated with only mild obstruction of the arterial lumen. Together, these images underscore a quintet of critical concepts: 1) atherosclerosis emerges early in life; 2) its progression is closely linked to the dynamics of plaque structure; 3) its natural history tends to be episodic and nonlinear with respect to time; 4) outcomes are not closely related to the severity of the disease, as reflected by plaque size or degree of stenotic obstruction; and 5) rapid worsening and precipitation of devastating events may suddenly occur without warning, even when the amount of disease is minimal and undetectable by anatomic (e.g., coronary angiography) or physiological (e.g., exercise stress testing) diagnostic tests. [Panels A–F, H reproduced with permission from M. J. Davies (713 ); panel G reproduced with permission from H. C. Stary (11 ).]

 
Clues to key features of vulnerable plaque have emerged from autopsy data, examination of coronary atherectomy specimens, intravascular ultrasound studies, and other approaches (14, 15, 16) (Fig. 1Go). The proven efficacy of therapeutic agents such as statins is now thought to be largely a result of their ability to favorably transform the architecture of plaque and render it less likely to rupture (17, 18, 19). The ability to understand plaque composition and to favorably alter its composition is likely to become increasingly important in developing novel therapeutic approaches. But on a molecular and cellular level, our understanding of the determinants of plaque composition, the natural history of specific plaque components, and how these contribute to vulnerability to rupture is poor.

One highly prevalent component of atherosclerotic plaque is calcium mineral deposits. Arterial calcification has attracted intense interest because it can be readily quantified noninvasively with radiographic imaging techniques (Fig. 2Go), and clinical studies have shown that high amounts of coronary artery calcification predict a heightened risk of myocardial infarction and sudden coronary death (20, 21, 22, 23, 24). At least in the coronary arteries, calcification seems to invariably indicate the presence of plaque, but if calcification is not present, it cannot be reliably concluded that plaque is absent (25, 26). Like atherosclerosis itself, arterial calcification begins quite early in life and increases with age at a rate that is roughly commensurate with the rate that atherosclerosis develops. By the seventh or eighth decade, the prevalence of arterial calcification, like atherosclerosis, is virtually universal, but may be present in amounts that vary widely among individuals. Calcification of plaques may precede clinical manifestations of plaque instability by many years, leading to the notion that detection of "subclinical" atherosclerosis by the use of a surrogate measure—arterial calcification—might allow accurate prospective identification of those destined to suffer clinical syndromes caused by atherosclerosis, while at the same time delineating those who are very unlikely to suffer an event. The clinical issues and pertinent evidence are controversial, far from straightforward, and beyond the scope of this review. For more detailed discussion, the reader is referred to recent expert consensus guidelines issued jointly by the American Heart Association and the American College of Cardiology (21), metaanalyses of clinical data (22, 24), and a number of commentaries (27, 28, 29, 30, 31). Here, we will limit our focus to potential molecular, endocrine, and genetic mechanisms mediating calcification in atherosclerotic plaque.



View larger version (100K):
[in this window]
[in a new window]
 
FIG. 2. Arterial calcification as detected by various imaging modalities. A–E, Intravascular ultrasound (IVUS) images of human coronary arteries. A normal artery (A) and an atherosclerotic artery with no calcification (B) are shown. C, Moderate calcification (arrows) in association with a large atherosclerotic plaque (dashed red region). D and E, Heavily calcified plaque, which almost completely encircles the artery in E (arrows). F, A radiograph of a heart taken at autopsy. The proximal sections of all three major coronary arteries are outlined by calcification. G–J, More modern radiographic imaging methods include CT. I and J, Three-dimensional reconstruction of images is possible. K–P, In vitro radiographic analytic imaging methods recently developed include high resolution synchrotron radiation x-ray microcomputed tomography. This method also allows three-dimensional image reconstruction and rendering. Calcification in a plaque from a human artery (K) and a saphenous vein bypass graft (L) are shown. K, Arrows indicate the arterial wall (top arrow), calcium deposits (middle arrow), and the sample holder (bottom arrow). M–P, Variations in volumetric three-dimensional rendering methods with synchrotron tomographic images. Arrows indicate the vessel lumen in M and O and calcium deposits in N and P. [Panels A–D courtesy of Dr. Steven L. Goldberg, University of Washington, Seattle, WA; panel E reproduced with permission from S. E. Nissen et al. (714 ); panel F reproduced with permission from M. J. Davies (713 ); panels G–J courtesy of Dr. Matthew J. Budoff, Harbor-UCLA Medical Center; panels K and L reproduced with permission from H. Jin et al.: Phys Med Biol 47:4345–4356, 2002 (715 ); panels M and N courtesy of Dr. Richard L. Kurtz, Louisiana State University.]

 

    II. Types of Arterial Calcification
 Top
 Abstract
 I. Introduction: Clinical...
 II. Types of Arterial...
 III. Molecular Mechanisms of...
 IV. Endocrine, Pharmacological,...
 V. Genetic Determinants of...
 VI. Conclusions
 References
 
Considerable confusion has arisen in the literature from failure to properly distinguish two distinct types of arterial calcification: medial arterial calcification, and calcification associated with atherosclerotic plaque (Figs. 3–5GoGoGo). Attempts to extrapolate results of studies involving medial arterial calcification to make general conclusions about atherosclerotic calcification are fraught with difficulties and not justified by our current understanding of the underlying pathologies. There are several basic lines of evidence consistent with the conclusion that medial calcification is not the same entity as plaque calcification, including histomorphological, clinical, epidemiological, genetic, and animal studies. For this reason, much of the data reviewed in subsequent sections must be interpreted in light of whether or not the model involves atherosclerosis. On the other hand, there may be some common elements mediating medial and atherosclerotic calcification; the two are not necessarily mutually exclusive. Nevertheless, conclusions from studies that have produced or inhibited calcification in models that do not involve atherosclerosis cannot always be assumed to apply to the pathogenesis of intimal plaque calcification. As will be seen below, there are important reasons to suspect that such studies may result in very different conclusions when they are replicated in the setting of atherosclerosis. However, it should also be emphasized that in most cases, such studies have not yet been reported and are urgently needed to further clarify the nature(s) and mechanism(s) of arterial calcification, both in the context of atherosclerotic plaque and in arterial tissues that are devoid of such disease.



View larger version (87K):
[in this window]
[in a new window]
 
FIG. 3. Histological examples of human arterial calcification. A, Arteries may exhibit concentric medial arterial calcification, which can occur in arteries free from atherosclerotic plaque. In this specimen, obtained during iliac artery surgery from a world-class athlete, there is concentric medial hypertrophy with medial calcification and little or no intima (Verhoeff-van-Gieson stain). Medial calcification may involve only elastic fibers or may sometimes develop into large calcific deposits. B, Calcification (black) in the media (red arrow) and in an intimal plaque (black arrow) in an artery from a patient with renal disease. The boxed area in panel B is magnified (x40) in panel C (I, intima; P, plaque; M, media; MacNeal’s tetrachrome staining). D and E, The human arterial specimen shown in panel D (with boxed area enlarged in E) demonstrates mineralization of medial elastic fibers and areas of large calcium deposits in the absence of significant intimal plaque. Calcification in atherosclerotic human arteries is usually localized in the deeper aspect of the intimal layer adjacent to the media. F–H, Typical well-developed calcification within intimal human plaques. Special embedding using methylmethacrylate allows sectioning without initial decalcification and reveals a fine, diffuse pattern of intimal calcification (G, arrow) in addition to large focal calcium deposits (teal color). It is not known whether diffuse calcification eventually becomes more organized into focal deposits. Goldner’s-Masson trichrome staining (F and G) reveals calcification that is not apparent on a section contiguous to panel F using hematoxylin and eosin alone (H). I and J, Circumferential calcification, largely intimal, can be seen in panel I, whereas more focal calcium deposits are seen in panel J (both human arteries stained with von Kossa; calcium is black). [Panel A reproduced with permission from C. A. Kral et al.: J Vasc Surg 36:565–570, 2002 (716 ); panels B and C reproduced with permission from S. M. Moe et al.: Kidney Int 63:1003–1011, 2003 (224 ).]

 


View larger version (107K):
[in this window]
[in a new window]
 
FIG. 4. Microscopic features of human atherosclerotic arterial calcification. A, Calcium in a lesion from a coronary artery of a 35-yr-old female. Small calcium mineral aggregates (blue arrows) are scattered throughout the lipid core, and macrophage foam cells (yellow arrow) cluster in an area of the core facing the lumen. B, A detail of this section at higher magnification. Calcium particles intermingle with cholesterol crystals and other lipids comprising the core of the lesion. C, Calcification (yellow arrows) and fibrosis (pink region denoted by asterisk) at the base of the lipid core of a lesion in the abdominal aorta. D, When lipid cores regress, residual areas of calcification may persist. The well-defined elastic lamina (arrows) separates a thin media and the adventitia (lower part of field) from the lesion. The subregion of the lesion adjacent to the elastic lamina consists of aggregates of calcium particles (denoted by asterisk) seen mostly as irregular empty spaces because of decalcification before sectioning. Residual extracellular lipid and cholesterol crystals can be seen in the midregion of the intimal plaque, with overlying SMCs and collagen. E, Focal calcification (basophilic blue-purple aggregates) associated with a macrophage-rich lipid core. Individual SMCs may become calcified as well. F and G, SMC containing variable amounts of calcium aggregates, some of which may represent calcified intracellular organelles. Calcium aggregates can also be seen in the intercellular space in panel G. Calcification in human plaques may form in focal intracellular and intercellular regions (as in F and G) but may also appear as a "wavefront" that produces plate-like circumferential intimal calcification. H, Mild mineralization of medial elastic lamellae (Mönckeberg’s sclerosis) in a human artery. von Kossa staining (brown-black) localizes only to elastic fibers and the internal elastic lamina (arrows). No staining is seen in the intima (red "In"). Media is denoted by a red "m." I, Advanced medial calcification (red arrow), with concomitant intimal calcification (green arrow). J, Artery exhibiting amorphous medial calcification (blue arrows), intimal (In) calcification (black arrows), and a region of bone (red arrow). Osteocytes can be seen (small nonstaining holes) within bone. K, Calcification (von Kossa, dark brown-black) within {alpha}-SMC actin-stained media (small arrowheads) (m, media; Ad, adventitia; Ca, calcification; arrows indicate mineralized internal elastic lamina). L, Staining by the macrophage marker CD68 is seen in the adventitia but not in the media or in the vicinity of areas of calcification. A–D, F, and G, Toluidine blue and basic fuchsin stain; E, hematoxylin and eosin; H–L, von Kossa stain. [Panels A–D, F and G reproduced with permission from H. C. Stary (11 ); panel E reproduced with permission from M. J. Davies (713 ); panels H–L reproduced with permission from C. M. Shanahan et al.: Circulation 100:2168–2176, 1999 (126 ).]

 


View larger version (79K):
[in this window]
[in a new window]
 
FIG. 5. Bone formation in human arteries. A–D, Examples of bone formation with mineralized trabeculae in arteries. B, Mature lamellar bone (white B) in a carotid artery plaque (red P), with a focus of calcification (red C). Osteocytes in lacunae (arrowheads) and an osteoclast in a resorption space (red arrow) can be seen. Mast cells and mononuclear cells (arrow) are seen in association with bone (C). D, Calcification (dark-colored areas within the intensely red-stained regions) integrated within mineralized bone tissue (intense red staining; hemotoxylin and eosin staining). E, Staining with von Kossa (black) and counterstaining with Safranin O reveals osteocytes and adjacent cells. F–N, High-magnification details of arterial bone tissue. F, Marrow-like tissue containing lymphocytes, mast cells, monocytes, and blood vessels within a mineralized region. Macrophages (brown, stained with HAM56 in panel G; and CD68, black arrows in panels I and N), SMCs (H), giant cells (K and N, arrows), multinucleated TRAP-positive osteoclasts (I, green arrow; J, black arrow) are shown in close association with arterial bone tissue. CD68-positive macrophages and giant cells (L and M, arrows) are seen adjacent to mineralized regions (L and M, asterisks), colocalized with BMP-6 expression (M). O–R, Atherosclerotic arteries demonstrating calcification(O, green arrows), cartilaginous metaplasia (O, black arrows), chondrocytes and chondroid tissue (Q and R, dark blue/purple regions), osteoblasts (P, black arrows) on the surface of an osseous region (P, asterisk), and multinuclear OLCs (P, green arrows). Thrombus (T) can be seen in the lumen of the artery shown in panel O. [Panels A, C–M reproduced with permission from M. Jeziorska et al.: Virchows Arch 433:559–565, 1998 (238 ); panel B reproduced with permission from J. L. Hunt et al.: Stroke 33:1214–1219, 2002 (237 ); panel N reproduced with permission from M. Jeziorska et al.: J Pathol 185:10–17, 1998 (259 ).]

 
A. Medial calcification
Mineral deposition in arteries can occur either as a component of atheroma or in the absence of atherosclerotic plaque. There are important distinctions between these two major forms of arterial calcification, in terms of clinical significance, morphological features, and pathophysiology. In atherosclerotic plaque, calcification typically forms in the intima, most often near the base of the plaque adjacent to the medial layer of the artery. Calcifications near the luminal surface may result in erosions and have been implicated in plaque disruption and thrombosis. In contrast, calcification of the medial layer of arteries occurs independently of atherosclerosis (32) (Figs. 3Go and 4Go). This type of calcification, initially referred to as Mönckeberg’s sclerosis (33), has been frequently observed radiographically with an appearance that has been likened to railroad tracks (34, 35, 36). Medial calcification can occur in otherwise normal young patients with no overt metabolic disease (37, 38), increases linearly with age (39, 40, 41, 42, 43, 44, 45), and is common in elderly patients (34, 35, 36, 39, 40, 41, 42, 43, 44, 45). Medial calcification is observed with particularly high frequency and severity in disorders characterized by generalized metabolic, electrolyte, or pH derangements, such as hypervitaminosis D (46), end-stage renal disease (ESRD), and diabetes mellitus (described below), but it also occurs in animals as diverse as elephants (47), dogs (48), and cows affected by enteque seco (enteque ossificans; secondary to ingestion of the vitamin D-containing plant Solanum glaucophyllum) (49). Mönckeberg’s sclerosis typically affects arteries that are less prone to develop atherosclerosis, such as the abdominal visceral arteries, arteries to the thyroid and breast, and the arteries supplying the extremities, but it also is frequently encountered in the aorta. Notably, medial calcification has not been reported to occur in the coronary arteries, allowing one to conclude that when calcification is observed in the coronary arteries, it is almost certainly associated with intimal plaque.

Medial calcification can be produced experimentally by a variety of treatments, including administration of vitamin D in rats (50, 51, 52, 53, 54, 55), pigs (56), and rabbits (57); vitamin D plus nicotine with or without supplemental cholesterol in rats (58, 59, 60, 61) and rabbits (62, 63); vitamin D with magnesium deficiency in pigs (64) and rabbits with chronic renal failure induced by renal cauterization and contralateral nephrectomy (65); 1-{alpha} hydroxycholecalciferol with or without parathyroidectomy in rats rendered uremic by subtotal kidney resection (66); warfarin with (51, 67, 68, 69, 70, 71) or without (70) vitamin K supplementation in rats; and vitamin D, warfarin, and vitamin K in rats (71, 72) (Fig. 6Go).



View larger version (26K):
[in this window]
[in a new window]
 
FIG. 6. Outline of the major genetic, molecular, cellular, and endocrine mediators of bone remodeling. In developing embryos, embryonic stem cells beget more restricted pluripotent progenitors that play important roles in tissue repair and remodeling processes in the adult. Mesenchymal and hematopoietic stem cells give rise to increasingly specialized cells that mediate bone formation and degradation, respectively. Mesenchymal stem cells can differentiate further into cells of the adipocyte, myoblast, and chondrocyte lineages, depending on genetic and morphogenetic programs and cues that are not well understood. Endochondral bone formation involves a cartilage intermediate. Mesenchymal progenitors become progressively committed to the chondrocytic lineage, form increasingly ordered condensations, proliferate, differentiate, and begin elaborating matrix elements consisting largely of collagens and aggrecan. At around this time, chondrocytes begin to hypertrophy and enter apoptosis. Simultaneously, neovascularization into the mesenchymal tissues develops, and preosteoblast cells that have differentiated from mesenchymal progenitors migrate into the region occupied by the now dying chondrocytes. There, they develop into terminally differentiated osteoblasts and begin mineralization of the matrix template left behind by the chondrocytes. Meanwhile, hematopoietic progenitors give rise to mononuclear lineages destined to become monocytes, macrophages, and osteoclasts. Cells capable of producing each of these exist in bone marrow, and can also be found in the circulating mononuclear fraction in blood. Proper development and function of osteoclasts critically depend upon the cytokines CSF-1 and RANKL, which signal via their cognate cell-surface receptors (c-fms and RANK, respectively). These are produced by osteoblasts/stromal cells, as well as by various circulating cells. In bone, development of osteoclasts appears to interact closely with nearby osteoblasts/stromal cells. Giant multinucleated cells appear to arise by fusion of mononuclear cells. Bone degradation involves the formation of a sealed zone beneath the osteoclast via adhesion molecules, particularly {alpha}vß3 integrin. Mineral degradation requires secretion of powerful proteases, such as cat K, and a marked decrease in pH under the osteoclast, accomplished primarily by proton-pumping enzymes such as ATP6i. The expression of TRAP, CA-II, cat K, CTR, ATP6i, and {alpha}vß3 integrin in giant multinucleated cells defines mature, fully differentiated osteoclasts fully capable of mineral resorption. Bone remodeling also involves a coordination or coupling of the activities of osteoblasts and osteoclasts such that there is normally little or no net change in bone mass, but precisely how this is accomplished is not well understood. The major transcription factors, hormones, and cytokines implicated in osteoclastogenesis, chondrogenesis, and osteoblastogenesis are shown. Proteases including members of the MMP family play important roles in remodeling tissues to facilitate endochondral bone formation. Intramembraneous bone formation occurs mainly in certain bones of the skull and does not require a cartilage intermediate to form before mineralization by osteoblasts. A number of the cell types, transcription factors, signaling molecules, growth factors, and hormones shown have been found in calcified areas of atherosclerotic plaque. Their potential role in mediating plaque formation is currently under investigation.

 
Coronary artery disease and heart failure are the leading causes of death in patients with ESRD (73, 74, 75), probably a result of the severe metabolic abnormalities associated with the disease (76). ESRD patients also have a very high prevalence and extent of arterial calcification, even young patients (77). This has often been attributed to secondary hyperparathyroidism elevating serum calcium and phosphorus levels to supersaturation levels, but studies of patients with ESRD on hemodialysis have failed to find a relationship between PTH levels and coronary artery calcification (78) or skin arteriolar calcification (79). However, there does appear to be an association between arterial calcification and increased serum phosphorus and increased [Ca2+ x PO4–3] ion product (78, 79). Although the precise mechanism is not understood, it seems reasonable to conclude that patients with ESRD manifest increased arterial calcification secondary to metabolic derangements (80). Moe et al. (81) recently reported that circumferential medial calcification of the epigastric artery was found in about one third (12 of 39) of ESRD patients undergoing renal transplant, and only one (of the 12) had evidence of atherosclerotic plaque. In addition, medial calcification colocalized with expression of osteopontin (OPN), bone sialoprotein (BSP), alkaline phosphatase, and collagen type I in the media; and increased expression of these bone matrix proteins roughly corresponded to the extent of calcification.

Numerous reports have documented the high prevalence and extent of medial calcification commonly seen in patients with diabetes mellitus (82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94). Medial calcification appears to be an indicator of the severity and/or duration of diabetes, because it is closely associated with many of the complications and sequelae of diabetes, particularly autonomic neuropathy (90, 92, 95, 96, 97). However, there also may be an independent association of medial calcification with nondiabetic neuropathies. For example, 6–8 yr after bilateral lumbar sympathectomy, Goebel and Fuessl (98) found that 93% of patients exhibited medial arterial calcification in both feet, and most of these patients were not diabetic. Twenty patients had no evidence of medial calcification before surgery, but seven who underwent bilateral sympathectomy exhibited medial calcification in both feet, and 11 who received unilateral sympathectomy developed medial calcification only on the side of the operation. In general, after unilateral sympathectomy, the incidence of calcified arteries on the same side as the operation was higher than that on the contralateral side (88 vs. 18%; P < 0.01). Diabetic patients appeared to have more extensive calcification than nondiabetic subjects, but the apparent differences between groups were not significant. Thus, sympathetic denervation can cause Mönckeberg’s sclerosis regardless of whether diabetes mellitus is also present or not (98). Young et al. (90) found that nonneuropathic diabetic patients and age-matched nondiabetic patients had a similar prevalence of medial arterial calcification, but on the other hand, they also found a moderate but significant correlation (r = 0.32) between medial calcification and duration of diabetes. Logistic regression showed that serum creatinine, vibration perception, and duration of diabetes predicted the probability of vascular calcification. However, several studies have shown that in patients with diabetes mellitus and/or renal dysfunction, quantification of coronary artery calcification is not an accurate marker of the severity of coronary atherosclerosis determined by angiography (99) and poorly predicts clinical outcomes (100). One potential explanation is that in such patients, increased amounts of medial calcification that is not associated with intimal plaque tend to weaken the relationship between calcification and atherosclerosis, and thereby significantly dilute predictive power. However, although it is clear that assessment of calcification in such patients is of little value, much remains to be learned about the specific and perhaps unique features of atherosclerosis in the setting of diabetes and renal dysfunction (80).

Genetic analyses on a large group of Pima Indians in the southwest United States also supports the notion that medial calcification may have determinants that are independent of diabetes. These Indians have been extensively studied (101, 102, 103, 104, 105, 106, 107, 108, 109, 110) because they have one of the world’s highest reported rates of obesity, insulin resistance, and type 2 diabetes (109, 111, 112, 113) but nevertheless exhibit relatively low risk for atherosclerosis and the cardiovascular complications that are highly prevalent in other diabetic populations (114, 115, 116). As might be anticipated, Pima Indians have a very high rate of medial arterial calcification. However, Narayan et al. (117) showed that offspring of parents with medial calcification were significantly more likely to have medial calcification, even after adjusting for the effects of age, sex, diabetes, serum cholesterol, and blood pressure. Offspring of one parent with medial calcification had an odds ratio (OR) of 3.3 for having calcification [95% confidence interval (CI), 1.5 to 7.6], whereas if both parents manifested medial calcification, the OR increased to 8.1 (95% CI, 3.4 to 18.8). Taken together, these studies are most consistent with the interpretation that medial artery calcification is strongly associated with diabetes, but in addition there is a genetic component that is independent of the determinants of diabetes and also of atherosclerotic disease.

In patients with diabetes, medial arterial calcification may be associated with increased risk of cardiovascular complications. For example, Lehto et al. (89) found that medial calcification was a strong independent predictor of total (risk factor-adjusted OR, 1.6; and 95% CI, 1.2 to 2.2), cardiovascular (OR, 1.6; 95% CI, 1.1 to 2.2), and CHD (OR, 1.5; 95% CI, 1.0 to 2.2) mortality. Others have similarly reported that medial calcification is associated with increased risk of future cardiovascular events in patients with diabetes (118). On the other hand, Maser et al. (119) studied 657 patients with insulin-dependent diabetes in the Pittsburgh Epidemiology of Diabetes Complications Study and did not find that medial arterial calcification was an independent predictor of elevated cardiovascular risk. The reason(s) for these discordant results are unclear. Edmonds (82) has suggested that the deleterious sequelae of medial calcification in patients with diabetes may be largely secondary to stiffening of arterial tone, increased systolic blood pressure, and impairment of endothelium-dependent relaxation (55), all of which would be expected to cause abnormal flow characteristics that may facilitate or synergize common complications of diabetes such as atherogenesis. How this might occur is unclear at this point, particularly in view of evidence that suggests that in patients without diabetes mellitus, medial arterial calcification may not predict increased cardiovascular risk. For example, Everhart et al. (120) found that nondiabetic subjects with medial arterial calcification did not have higher mortality rates than subjects without medial arterial calcification. A recent report by London et al. (121) assessed intimal carotid arterial calcification with carotid B-mode ultrasonography and peripheral medial calcification with soft-tissue x-rays of the pelvis and thigh in 202 patients on stable hemodialysis. Not surprisingly, intimal carotid calcification was more often seen in older patients with a prior history of cardiovascular disease, whereas peripheral medial calcification was typically seen in young and middle-aged patients without increased cardiovascular risk. Medial calcification was linked to duration of hemodialysis and calcium-phosphate electrolyte disorders. Both types of calcification were associated with all-cause and cardiovascular mortality upon follow-up, independent of risk factors. Patients with medial calcification survived longer. These findings must be interpreted with considerable caution, because both enrollment of subjects and interpretation of outcomes were susceptible to a number of sources of selection bias (122, 123, 124). Probably the most clearly interpretable result from this study was the association of metabolic disorders of calcium and phosphate metabolism with medial calcification. Other evidence also supports the conclusion that metabolic disorders can contribute importantly to medial calcification (125).

Studies of medial calcification typically have relied upon radiographic detection methods (such as plain x-ray films) that are very specific but have low sensitivity. Additional studies in both diabetic and nondiabetic cohorts using imaging modalities with higher sensitivity may help to better elucidate the prognostic significance of medial calcification. At this point, based on available data, it seems reasonable to tentatively conclude that medial calcification probably imparts higher cardiovascular risk in diabetic patients, but not in nondiabetic patients.

Several studies have reported differential expression of bone matrix proteins in medial calcification compared with normal arteries or to intimal calcification. For example, Shanahan et al. (126) compared mRNA and protein expression of bone matrix proteins in human peripheral arteries with and without medial calcification. Normal arteries expressed matrix Gla protein (MGP) and osteonectin in the media. In the medial layer of the arteries with medial calcification, there were decreased levels of expression of MGP and osteonectin, but increased expression of alkaline phosphatase, BSP, bone Gla protein (BGP, or osteocalcin), and collagen II compared with normal vessels. Medial calcification was noted to invariably occur directly adjacent to medial smooth muscle cells (SMCs) expressing typical SMC markers such as SMC{alpha} actin, but was never seen in the vicinity of macrophages or lipid deposits (126).

Other studies have documented expression of MGP in the media closely associated with SMCs and elastic lamellae. Extensive medial calcification is seen in MGP–/– mice (127, 128), which has led to the suggestion that MGP may have a homeostatic role in chelating or sequestering Ca2+ ions, thereby inhibiting mineralization. In addition to MGP, other proteins can bind or chelate Ca2+ ions. However, genetic ablation of several of these, including OPN (129, 130), osteonectin (131, 132), and BGP (133), does not produce arterial calcification in mice. Interestingly, double knockout (OPN–/–; MGP–/–) mice exhibited accelerated arterial calcification compared with single knockout (OPN+/+; MGP–/–) mice (134). Recent studies by Bostrom and colleagues (135, 136, 137) suggest that MGP may inhibit chondrocyte development by acting as an extracellular inhibitor of bone morphogenetic proteins (BMPs). This may explain why numerous chondrocytes are observed in arteries of MGP knockout mice (127, 128).

B. Intimal calcification
The remainder of this review will focus on arterial calcification observed in the setting of atherosclerotic plaque. This type of calcification occurs almost exclusively in the intimal layer of the artery (25). Normal arteries demonstrate little or no intima (9, 138, 139). However, at sites of hemodynamic stress, fibromuscular proliferation, referred to as adaptive intimal thickening, does occur and is reported to be the site at which clinically significant atheroma develops in the coronary arteries (6, 7, 9, 138, 139) (Fig. 1Go). Calcification in the intima occurs in at least two distinct patterns (Figs. 3Go and 4Go). First, there are clumps of mineralized areas that have been described as punctate; these may enlarge, and most often are noted in the basal regions of the intima adjacent to the media. Sometimes these may appear organized, and foci of calcification may undergo osseous metaplasia, which includes bone marrow as well as bone (Fig. 5Go). Hematopoietic marrow, osteoblast-like cells, chondrocyte-like cells, multinucleated osteoclast-like cells (OLCs), and perhaps even osteocytes may be present in association with proteins normally associated with bone metabolism and typically not expressed in normal arteries (140, 141).

Pathologists routinely process arterial specimens by an initial decalcification step, which is intended to remove enough calcium deposits to enable tissue sectioning with standard microtomes. Although this facilitates rapid tissue sample processing, it is likely that the amount of calcification observed will be underestimated. Innovative histochemical techniques performed by Fitzpatrick and colleagues (142, 143) that do not require a decalcification step have now established that there is a fine, diffuse pattern of calcification that commonly occurs in atherosclerotic plaque in addition to the punctate pattern (Fig. 4Go). This diffuse pattern has several important characteristics and implications. First, it may be observed in virtually all areas of the intima. Second, it is unlikely to be detected by radiographic or other imaging modalities, because the overall tissue density is nearly the same as that of adjacent tissue that does not exhibit any calcification at all. Third, the natural history of the diffuse pattern and its relationship to the punctate pattern is not known. It is possible that they are independent of one another, but it is also conceivable that the diffuse pattern may, at least in some cases, be an early stage of what eventually becomes the punctate pattern. Resolving these questions will be an important goal of future studies.


    III. Molecular Mechanisms of Arterial Calcification
 Top
 Abstract
 I. Introduction: Clinical...
 II. Types of Arterial...
 III. Molecular Mechanisms of...
 IV. Endocrine, Pharmacological,...
 V. Genetic Determinants of...
 VI. Conclusions
 References
 
A number of studies have shown intriguing parallels between arterial calcification in atherosclerotic plaque and osteogenesis. Over a century ago, pathologists observed that fully formed bone tissue, including marrow, can form in atherosclerotic arteries (144, 145) (Fig. 5Go). More recently, cells, proteins, and cytokines known to be involved in new bone formation have been reported to be present in arteries, and there is increased expression of a number of bone-related proteins in atherosclerotic plaque, particularly in subdomains of plaque where calcification occurs (23, 141, 146, 147, 148). Because of these similarities, it has been proposed that the mechanism by which arterial calcification occurs is similar to the mechanism of new bone formation (141, 148, 149, 150, 151). Although the parallels are intriguing and suggestive, these data are either observational expression analyses of bone-related proteins in arteries or extrapolations based upon in vitro cell culture studies. Direct tests of possible mechanisms using in vivo models have not yet been reported.

Nevertheless, several models postulating mechanisms for the formation and/or inhibition of calcification have now been proposed (141). These are: 1) the active model; 2) the passive physicochemical model; and 3) the arterial OLC model. Each model will be described below, and all are uniquely predicated on specific principles of bone formation and/or resorption. These models are not necessarily mutually exclusive alternatives, but may be complementary. Widely accepted principles of both vascular and bone biology form the foundation for these models and will therefore be briefly reviewed.

A. Molecular and genetic mechanisms of bone formation
In vertebrates, the three-dimensional pattern of the endoskeleton is specified by complex genetic programs derived from, and related to, phylogenetically ancient signaling systems that direct embryogenesis and morphology (152, 153, 154). Osteoblasts are responsible for bone formation (155, 156). Osteoclasts degrade bone, and thus counterbalance the actions of osteoblasts on overall bone mass under homeostatic conditions (157, 158) (Fig. 6Go). These cell types are derived from mesenchymal and hematopoietic (mononuclear phagocyte) precursors, respectively. Bone is formed in one of two distinct ways. The major skeletal elements develop by endochondral ossification that involves a cartilage intermediate, but some bones, notably the craniofacial bones, are formed by intramembraneous ossification in which bone arises directly from mesenchymal cell condensates without a cartilaginous intermediate (159, 160). Chondrocytes are the third major cell type involved in bone formation, are derived from mesenchymal cells, function to generate an initial cartilage template upon which endochondral bone formation including mineral deposition may occur, and develop in response to specific genetic programs (161, 162). Chondrocytes and osteoblasts appear to be derived from a common progenitor cell (159). Chondrocytes deposit a cartilagespecific extracellular matrix, proliferate, become hypertrophic during maturation, and undergo apoptosis. Surrounding mesenchymal cells differentiate into osteoblasts and invade the zones occupied by hypertrophic and dying chondrocytes. Simultaneously, neovascularization and osteoclastogenesis occur in the same region. The transition is synchronized, at least in part, by soluble morphogens expressed by chondrocytes (163). Osteoclasts degrade the cartilaginous matrix to make way for further bone development, and osteoblasts synthesize a bone-specific matrix, using the degraded cartilaginous matrix as a scaffold.

The coordinated development of osteoblasts and osteoclasts involves cell-cell contact between precursors of both lineages and soluble cytokines and is interconnected with apoptosis of chondrocytes. The entire process is elegantly orchestrated by genetic programs that regulate expression of numerous specific molecules in an ordered spatiotemporal manner (164, 165, 166). Among the myriad morphogens, cytokines, and signaling factors involved in bone formation (155, 156, 157, 158, 164, 165, 166, 167, 168, 169) (Table 1Go), two pivotal upstream molecules are BMP-2 and the transcription factor core-binding factor {alpha}1 [Cbfa1; also called runt-related transcription factor 2, osteoclast-stimulating factor 2, and acute myeloblastic leukemia 3].


View this table:
[in this window]
[in a new window]
 
TABLE 1. Selected bone-related genes and their human chromosomal loci

 
The BMP signaling pathway is critical to proper bone formation. BMP-2 and BMP-4 are also expressed in calcified regions of atherosclerotic arteries (146, 170, 171). For this reason, it has been suggested that BMP signaling is an important component of the mechanism of arterial calcification. BMPs are members of the TGF-ß superfamily, control fundamental aspects of limb patterning, and are involved in chondrocyte and osteoblast differentiation and proliferation. The late Marshall Urist (172) was the first to discover that BMPs can cause transdifferentiation of mesenchymal cells into osteoblasts, which in turn can mediate ectopic bone formation. This finding has been replicated by numerous other investigators in diverse models (173, 174). BMPs are among the most potent osteogenic agents known, are phylogenetically conserved, and are ubiquitous in all phyla in the animal kingdom from insects to humans. Across vertebrate and invertebrate phyla, BMPs are highly conserved; when mammalian BMP-4 is expressed in Drosophila with a mutation in the BMP homolog decapentaplegic (dpp), a normal Drosophila phenotype is rescued (175), and conversely, dpp can functionally substitute for BMPs in vertebrates (176). Attesting to their critical developmental and homeotic roles, most genetic mutations in BMPs or their receptors are embryonic lethal (177, 178).

BMP signaling utilizes a heteromeric complex of type I and type II transmembrane serine/threonine kinase receptors. Binding of TGF-ß superfamily ligands (TGF-ß, activin, and BMPs) to the type II receptor causes recruitment of the type I receptor into a heterotetrameric structure. There are seven distinct type I receptors identified in vertebrates that can interact with one of five different type II receptors, presenting many combinatorial possibilities for assembly of the receptor complex (173, 177, 178, 179, 180). Because BMP signaling results in diverse biological effects, one might imagine that regulation of downstream gene targets would be a function of the specific heteromeric type I and type II receptor complexes formed. Surprisingly, biological output appears to be exclusively determined by the type I receptor (174, 181). Furthermore, the type I receptor funnels the BMP signal into a single cytoplasmic pathway targeting specific nuclear genes. A series of transphosphorylations allow transmission of the signal to cytoplasmic proteins called Smads (mammalian homologs of mothers against dpp), which then can carry the signal to nuclear effectors. Once nuclear translocation has occurred, highly complex and poorly understood interactions occur between Smad signal transducers and genetic components that result in stimulation or repression of specific target genes to produce biological responses that are variable. Smad heteromeric complexes bind directly to DNA response elements, but high-affinity binding requires cooperative interactions with coactivators and repressors that modulate DNA binding affinities (174, 180, 181, 182, 183, 184). Thus, the BMP signaling pathway resembles an hourglass, with the strategic bottleneck localized at the type I cell surface receptor, and is subject to tight regulation at both the extracellular (185) and intracellular (181, 186, 187, 188, 189, 190, 191, 192, 193, 194) levels. The complexities and dynamic interrelationships of BMP signaling with other signaling pathways have become increasingly apparent (174, 179, 180, 181, 184, 187, 188, 189, 190, 191, 192, 193). Studies examining how BMP signaling participates in arterial calcification are currently in progress.

Cbfa1 is one of the transcription factors that regulates osteoblastic differentiation and bone formation (195, 196, 197, 198, 199). It is clear that Cbfa1 is essential for osteoblast and perhaps also chondrocyte development. Cbfa1 is expressed early in development in cells that have the potential to differentiate into either osteoblasts or chondrocytes and precedes the appearance of bone. Mice with targeted deletion of the gene encoding Cbfa1 manifest a complete lack of osteoblasts and die soon after birth (195, 196). BMPs (including BMP-2) induce the expression of Cbfa1 mRNA (195, 196, 199, 200, 201, 202). BMP signaling and Cbfa1 interact in complex ways that are not fully understood but may be either dependent or independent of interactions with Smads (203, 204, 205, 206).

Although Cbfa1 has been characterized as the "master" upstream osteoblast transcription factor, recent evidence indicates that other mechanisms critically regulate osteoblast formation and function. For example, mice deficient in the low-density lipoprotein (LDL) receptor-related protein 5 demonstrate an osteopenic phenotype due to decreased numbers and activity of osteoblasts (207). Another member of the runt-related transcription factor family, CBFß (core binding factor ß; also called polyoma enhancer binding protein 2ß) was recently shown to have a critical role in bone development (208, 209). The CBFß–/– mutation is embryonic lethal in mice, due to hemorrhage and lack of hematopoiesis (210, 211). CBFß is required for myeloid and lymphoid differentiation but does not play a critical role in erythroid differentiation (212). CBFß is expressed in developing bone and interacts closely with Cbfa1. Knock-in mice partially expressing CBFß demonstrate a phenotype similar to that of Cbfa1–/– mice, i.e., delayed chondrocyte differentiation and endochondral and intramembraneous ossification, but the phenotype is less severe than that exhibited by Cbfa1–/– mice.

Pluripotent mesenchymal tissues are capable of differentiation into adipocytes, myoblasts, chondrocytes, or osteoblasts. Lineage commitment depends upon coordinated expression of sets of specific transcription factors and concomitant suppression of other genetic programs that specify alternative cellular fates. PPAR and C/EBP genes are required for adipogenesis (213, 214), MyoD and myogenin specify myoblast cell fate (215), and Sox5, Sox6, and particularly Sox9 genes are essential for development and maturation of chondrocytes (160, 216).

BMPs are powerful inducers of ectopic bone formation when injected sc at nonbone sites (172, 217). The mechanism involves transformation of mesenchymal cells into osteoblasts in a manner reminiscent of endochondral bone formation. These observations are physiologically relevant. In developing organisms, BMP signaling regulates early commitment and differentiation of pluripotent mesenchymal progenitors by induction of complex genetic programs that are not yet completely understood (159). Among the many gene targets of BMP signaling identified to date, three transcription factors—Distal-less5 (Dlx5), Cbfa1, and osterix (Osx)—are now considered to be the master genes essential for differentiation of mesenchymal progenitors into terminally differentiated osteoblasts. Determining the precise function and regulation of these genes in the adult organism has been limited by the fact that targeted deletion of these genes is either embryonic lethal (Dlx5) or produces mice that are not viable and die shortly after birth (Cbfa1 and Osx). In mice deficient in Cbfa1 (195) or Osx (218), mesenchymal cells are unable to differentiate into functional osteoblasts and do not deposit bone matrix. Cbfa1 and Osx (219) expression are both induced by BMPs. Cbfa1 expression in Osx–/– mice is normal (218), suggesting that Cbfa1 lies upstream of Osx. Recent data indicate that Dlx5 is a direct gene target of BMP-2 signaling and that Dlx5 is a key upstream regulator of both Cbfa1 and Osx (220). One proposed model suggests that Dlx5 induces first Cbfa1 and then Osx in a temporally controlled manner that is not yet understood. The primary role of Cbfa1 appears to be to provide a pool of osteoblast progenitors, and Osx in turn is largely responsible for terminal differentiation of these preosteoblasts into fully functional osteoblasts. However, Cbfa1 also participates in functional activities of osteoblasts (155, 196, 221). In addition, mice overexpressing Cbfa1 manifest osteopenia resulting from increased osteoblast formation but also increased osteoclastogenesis by a mechanism that appears to involve enhanced expression of receptor activator of nuclear factor {kappa}B (RANK) ligand (RANKL), macrophage colony-stimulating factor (M-CSF), and matrix metalloproteinase-13 (MMP-13) (collagenase 3) and decreased synthesis of osteoprotegerin (OPG) by osteoblasts (222).

Muscle segment homeobox 2 (Msx2) is another homeodomain transcription factor that appears to be an important downstream target of BMP signaling modulating osteoblast development from mesenchymal pluripotent progenitors (223). In vitro, Msx2 was strongly up-regulated by BMP-2, and it was demonstrated that Msx2 increased primary aortic myofibroblast expression of Osx 10-fold compared with control conditions. Cbfa1 expression was not changed by Msx2, but alkaline phosphatase was increased 50-fold, whereas genes associated with adipogenesis were suppressed (223). These results are most consistent with the interpretation that Dlx5 and Msx2 are two important gene targets of BMP signaling in osteoblast development, but that Dlx5 acts earlier than Msx in differentiation and perhaps function of osteoblasts. Cbfa1 (224, 225) and Sox9 (225) are expressed in arterial tissues, and increased expression of Cbfa1 is associated with calcification of atherosclerotic plaque (224, 225, 226, 227). However, expression of Osx, Dlx5, and Msx2 in arterial tissues has been reported.

Bone is constantly being degraded throughout life as a part of normal remodeling and to maintain mineral and electrolyte concentrations within physiologically acceptable limits. Bone formation is normally closely linked to bone resorption; however, there is net bone formation during the first two decades of life. Total bone mass peaks around the third decade, and thereafter bone degradation exceeds formation and there is a gradual loss of bone mass for the remainder of the life span. Differentiation and maturation of osteoclasts are critically regulated by multiple secreted molecules and transcription factors, including c-fos, nuclear factor-{kappa}B, PU.1 (also called Spi-1; the protein encoded by the purine-rich box-1 gene), and microphtalmia transcription factor, and involve cell-cell interactions with osteoblasts or bone marrow stromal cells (155, 156, 157, 158, 228, 229). Mature osteoclasts stain positively for tartrate-resistant acid phosphatase (TRAP) and express a number of characteristic molecules, including calcitonin receptor (CTR), cathepsin K (cat K), carbonic anhydrase II (CA-II), the proton pump ATP6i, and the {alpha}vß3 integrin. Osteoclasts are traditionally considered to be present only in bone; however, some osteoclast progenitors circulate in the mononuclear fraction (230, 231, 232, 233, 234) and are recruited by specific interactions with vascular endothelial cells (ECs) (235). A number of genetically altered mouse models with bone disorders characterized by altered bone resorption or osteoblast activity have been developed and widely used to study bone diseases such as osteoporosis and osteopetrosis (236).

B. The active model of arterial calcification
Pathologists have long known that heterotopic calcification such as that seen in atherosclerotic arteries can evolve into mature bone tissue histomorphologically indistinguishable from skeletal bone (25, 144, 145, 237, 238) (Fig. 5Go). The active model of arterial calcification evolved partly from these observations, stimulated in large part by studies of Bostrom et al. (170), who reported the existence of pluripotent arterial cells that they named calcifying vascular cells (CVCs), which were immunologically distinct from other arterial cells. CVCs colocalized with expression of bonerelated proteins including BMP-2, a potent osteogenic agent, and also could form mineralized nodules in cell culture under certain conditions (170, 239, 240). Subsequently, several groups independently confirmed these in vitro results but also demonstrated that, besides CVCs, arterial SMCs could also form mineralized nodules and express matrix proteins typically found in bone (226, 241, 242, 243, 244, 245). Moreover, expression of bone matrix-associated proteins appears to colocalize with a number of cell types in arterial tissue sections (224, 225, 227, 246, 247, 248, 249). For these reasons, the precise identification of the cell type(s) participating in arterial calcification in vivo remains uncertain.

Recently, Engelse et al. (227) found that expression of Cbfa1 was largely restricted to monocyte-macrophages and, to a lesser extent, SMCs in human atherosclerotic lesions. It is unlikely that cells derived from the mononuclear phagocytic lineage give rise to osteoblast-like cells within the artery. However, this same group recently reported that in bone, osteoclasts (which are derived from mononuclear phagocytic precursors) demonstrate immunopositivity for Cbfa1 in the absence of any detectable Cbfa1 gene expression (i.e., mRNA transcripts) (250). What at first appeared to be Cbfa1 protein expression by osteoclasts was actually the result of phagocytosis and incomplete digestion of cells derived from the osteogenic/chondrogenic lineages, which are well known to express Cbfa1. It is possible that a similar process accounts for apparent Cbfa1 protein expression in arterial monocyte-macrophages. However, because Cbfa1 transcripts were detected in the MM6 monocytic cell line and in a subset of macrophages in a limited number of human arterial sections (227) but not in bone osteoclasts (250), more definitive resolution of this issue will require further study. Cbfa1 is also expressed in both the intima and media in the arteries of patients with ESRD and both medial and intimal (atherosclerotic) calcification (224). Numerous other reports have consistently confirmed that in atherosclerotic arteries, and particularly areas of plaque where calcification occurs, increased expression of proteins normally restricted to developing bone is observed (142, 225, 246, 247, 248, 249). Nevertheless, the validity of the active model has not been tested in vivo, and alternative models, as described below, have been proposed.

Studies of both human arteries and animal models of atherosclerosis have revealed the existence of cells that are phenotypically similar to all major cell types involved in endochondral bone formation—chondrocytes, osteoblasts, and osteoclasts (144, 145, 237, 238, 249, 250, 251, 252, 253, 254, 255, 256, 257, 258, 259) (Figs. 4–7GoGoGoGo). Furthermore, these cells possess the required signaling pathways and express proteins normally associated with their analogs in bone (140, 141). Mineralization occurs in bone via matrix vesicles, and these have also been described in arterial tissues (260, 261, 262, 263) (Fig. 7Go). The mineral composition of bone (hydroxyapatite) is chemically very similar to that observed in calcific deposits in atherosclerotic arteries (170, 264, 265, 266, 267). Collectively, these findings strongly support a model wherein arterial calcification occurs in plaque microenvironments in a manner that recapitulates osteogenesis.



View larger version (112K):
[in this window]
[in a new window]
 
FIG. 7. Arterial calcification in animal models. Some animals may spontaneously develop arterial calcification with time. A, A maximum intensity projection from a micro-CT scan of an aged (26 months) rat heart. Calcification is evident in all major epicardial coronary arteries. B, A cross-section of one of the coronary arteries with von Kossa staining. Calcification appears as dark regions in the arterial wall. Sections were not decalcified before methylmethacrylate embedding and Goldner’s-Masson-Trichrome (C and D, blue-green color) or von Kossa (E, black color) staining. Red regions are unmineralized matrix. Many animal models also manifest evidence of cartilage-like tissues and chondrocytes. For example, panels C–E show cartilage (arrow in C; magnified in D and E) within the vascular wall from an aged oophorectomized female rat. A diffuse mineralization pattern can be seen (C and D), with some punctate areas as well. On the other hand, some animal models exhibit calcification largely or exclusively confined to the media, typically in association with elastic fibers. Sections of abdominal aorta from rats treated with warfarin and vitamin K for 2, 3, 4, or 5 wk (F–I, respectively) and stained with von Kossa (black) shows progressive mineralization involving the media, with little or no neointima present. J and K, Aortic sections from rats treated with vitamin D alone for 96 h (J) or vitamin D plus the osteoclast inhibitor SB 242784 (K). Both sections are stained with hematoxylin and eosin, as well as with von Kossa. L, Aortic sections from rats treated for 8 wk with warfarin and vitamin K1 and stained with von Kossa. Mineralization of elastic fibers is evident, which eventually develops into larger calcium deposits, but involves only the medial arterial layer because these animals have no intimal atherosclerotic plaque. J–L, The arterial lumen. Medial calcification can, in some animal models, appear similar to that typically seen in intimal tissues. M, Hematoxylin and eosin staining of aortic section from OPG–/– mice shows calcified lesions (dark purple) in the media and also involving intimal tissue. N–R, Large, multinucleated cells can be seen in these lesions (yellow arrows) that stain positive for cat K (P) and TRAP (Q). However, other multinucleated cells in these lesions are negative for cat K, TRAP, and the macrophage marker F4/80 (R), suggesting dedifferentiated cells or pluripotent cells of undetermined origin and fate. MGP–/– mice exhibit diverse features of arterial calcification, but again this is almost completely limited to the media because little or no intima is present. S–W, Aortic sections from MGP–/– mice. Mineralization of elastic lamellae in the aorta is evident (S) (von Kossa staining for mineral appears black). T, Mineralization of internal elastic lamina (arrow) in a coronary artery. Note that calcification is not intimal but rather in the medial layer. U, A transmission electron micrograph (TEM) of aortae from MGP–/– mice showing extensive calcification of elastic fibers in the media (cEF, calcified elastic fiber; nEF, noncalcified elastic fiber). Note the calcification "front" that appears to be advancing along the length of the fiber (arrowhead). V, A TEM of MGP–/– aorta shows chondrocytes (ch) surrounded by hyaline cartilage and type II collagen fibrils (arrows), proteoglycans (arrowheads), and matrix vesicles (inset). W, Aortic cartilaginous metaplasia in an MGP–/– mouse with chondrocytes (arrows) in the midst of metachromatic cartilage matrix. Augmentation of aortic calcification is observed in MGP–/– OPN–/– mice when compared with MGP–/– mice. X, An aortic section from an MGP–/– OPN+/+ mouse stained with von Kossa. Y, A similar section from a double knockout (MGP–/– OPN–/–) mouse. Z, Fragmentation of the elastic lamellae (arrows) can be seen in aortae obtained from MGP–/– OPN–/– double knockout mice (panels X–Z: M, media; Ad, adventitia; L, lumen; asterisks, calcium deposition). AA and BB, TEM shows mineralization of intracellular (AA, arrowheads) or extracellular (BB, arrowheads) membranous compartments adjacent to sc implanted glutaraldehyde-fixed aortic valve leaflets in OPN–/– mice. Mineralization was also observed in extracellular matrix vesicles (BB, arrows). Macrophages (blue stain) and multinucleated giant cells localized to the region and stained positive for the osteoclast-related enzyme CA-II (CC, brown stain; inset shows a higher magnification of the boxed region). DD, Both multinucleated CA-II-positive cells and mononuclear phagocytes adjacent to implants. These colocalized with regions of developing mineral deposits stained with Alizarin red in OPN+/+ mice (EE). FF, A higher magnification of the boxed region. GG, There was also colocalization of OPN staining (brown) with mineral deposits (Alizarin red stain) in OPN+/+ mice. [Panels A–E reproduced with permission from L. A. Fitzpatrick et al.: Endocrinology 144:2214–2219, 2003 (253 ); panels F–I reproduced with permission from P. A. Price et al.: Arterioscler Thromb Vasc Biol 18:1400–1407, 1998 (70 ); panels J and K reproduced with permission from P. A. Price et al.: Circ Res 91:547–552, 2002 (51 ); panel L reproduced with permission from R. Essalihi et al.: Am J Hypertens 16:103–110, 2003 (72 ); panels M–R reproduced with permission from H. Min et al.: J Exp Med 192:463–474, 2000 (341 ); panels S–W adapted with permission from G. Luo et al.: Nature 386:78–81 (127 ); panels X–Z reproduced with permission from M. Y. Speer et al.: J Exp Med 196:1047–1055, 2002 (134 ); panels AA–GG reproduced with permission from S. A. Steitz et al.: Am J Pathol 161:2035–2046, 2002 (248 ).]

 
C. The potential role of OLCs in arterial calcification
How might the processes leading to plaque mineralization be limited or opposed? One possibility is that the process is self-limiting or subject to negative feedback loops. Bone remodeling is the summation of bone formation by osteoblasts and bone degradation by osteoclasts. These processes are not strictly independent, but instead affect one another in numerous ways. Overall, bone formation and degradation are tightly regulated and synchronized in such a way that there is normally little net change in overall bone mass (268, 269). Building on the similarities between bone formation and arterial calcification summarized above (Section III.B) and multiple lines of additional evidence (described later in this section), we proposed that there is a similar mechanism to oppose arterial calcification that involves putative arterial OLCs (140) (Figs. 5–7GoGoGo).

M-CSF is a key cytokine involved in survival of osteoclast progenitors and differentiation toward a mature functional osteoclast phenotype (157, 158, 229, 270). Earlier studies with mice lacking both M-CSF and apolipoprotein (apo) E suggested a potential role for M-CSF in arterial calcification (271). Apo E-deficient mice may develop atherosclerotic lesions that histologically exhibit many features of human lesions (272, 273), including calcification (254, 255, 271, 274). Mice harboring a naturally occurring structural mutation in the M-CSF gene exhibit osteopetrosis as a result of an absence of osteoclasts (275, 276). A striking finding in our study was that approximately one third of double knockout mice lacking both M-CSF and apo E developed massive arterial calcification but no atherosclerotic lesions in the vessel wall (271). This suggested the possibility that the development of calcification in atherosclerotic lesions may, in part, be the result of decreased osteoclastic activity due to lack of M-CSF. Accumulating evidence reviewed later in this section now increasingly supports this intriguing possibility.

M-CSF and RANKL signaling are both necessary and sufficient for osteoclastogenesis and function. M-CSF is a multifunctional protein that regulates the differentiation, proliferation, survival, and function of mononuclear phagocytic cells (MPCs) in vitro and in vivo by binding to cellular-feline McDonough sarcoma protooncogene (c-fms) (234, 277, 278, 279, 280). All cells of the atherosclerotic vessel wall, including ECs, SMCs, and MPCs, can express M-CSF. M-CSF stimulates the proliferation and survival of macrophages in culture and in vivo, suggesting that induction of M-CSF in atherosclerotic lesions may directly stimulate MPC proliferation. We hypothesize that M-CSF produced locally within developing atherosclerotic lesions is critical to the differentiation of a subpopulation of MPCs toward an osteoclast-like phenotype. This effect of M-CSF on MPCs may take place in an autocrine, paracrine, or juxtacrine manner. Although both M-CSF and RANKL signaling are necessary for osteoclast development, lack of M-CSF alone is sufficient to cause greatly diminished numbers of osteoclasts and osteopetrosis (234, 276, 278, 279).

M-CSF is required not only for survival of MPC precursors, but also for maturation and/or function of osteoclast precursors. This conclusion is supported by studies demonstrating that transgenic overexpression of the antiapoptotic factor B cell chronic leukemia-2 (bcl-2) in M-CSF-deficient mice results in greater than normal levels of circulating monocytes and partial rescue of the osteopetrotic phenotype (281, 282). Stanley and colleagues (283) recently reported that expression of a transgene encoding the full-length M-CSF precursor reconstitutes circulating and tissue levels of the secreted isoform of M-CSF and some of the cell-surface isoform in M-CSF–/– (op/op) mice. Expression of this transgene corrects most of the defects in op/op mice. Furthermore, preliminary studies suggest that compound mutant atherosclerosis-prone mice that are deficient in both M-CSF and apo E and also overexpress bcl-2 (i.e., apo E–/–; M-CSF–/–; bcl-2Tg) develop little or no atheroma, despite marked hypercholesterolemia and normal or greater than normal levels of circulating monocytes (282). These data highlight the central importance of M-CSF in atherogenesis, show that the antiatherogenic effects of M-CSF deficiency are not due to lack of circulating monocytes, and suggest that lack of M-CSF might contribute to arterial calcification by inhibiting development of arterial OLCs, thereby allowing unopposed activity by CVCs (140). Thus, M-CSF may have dual effects: promoting or allowing atherogenesis to develop, and inhibiting calcification.

Absence of RANK also results in osteopetrosis, but rescue of the wild-type phenotype can be achieved by adoptive bone marrow transplantation using recombination activating gene 1–/– bone marrow donors (284). Normal osteoclast precursors are found in rag1–/– mice, suggesting that RANKL signaling is predominantly required for early differentiation of precursors into mature osteoclasts. Because hematopoietic precursors from RANK–/– mice cannot be induced to form osteoclasts in the presence of RANKL and M-CSF, it appears that M-CSF cannot by itself stimulate differentiation and maturation of osteoclast precursors. Retroviral delivery of RANK cDNA into RANK-deficient mice restores normal osteoclast differentiation and function, providing an unequivocal demonstration that osteoclastogenesis is dependent on RANK-mediated signaling. However, the effect of RANK deficiency on arteries has not been fully investigated thus far, and because