Fetuin-A Regulation of Calcified Matrix Metabolism

Fetuin-A Biosynthesis

Fetuins are highly expressed liver-derived plasma proteins bearing posttranslational modifications proteolytic processing,^6,7 complex glycosylation,^8,–,10 phosphorylation (Ser and Thr),^6,11,–,15 and sulfation.¹⁶ Human fetuin-A/α₂-HS glycoprotein is processed from a single chain precursor to the mature circulating two-chain form.⁶ Human fetuin-A is susceptible to further proteolytic cleavage in septicemia,⁷ and bovine fetuin-A is processed by matrix metalloproteinases.¹⁵ Figure 2 illustrates secondary modifications and allelic variants identified in human fetuin-A/α₂-HS glycoprotein. These modifications may regulate protein expression levels, stability, and biological activity. Phosphorylation is indispensible for fetuin-A interaction with the insulin receptor,^17,18 whereas phosphorylation seems not to be required for mineral interaction.¹⁹ Desialylation will result in immediate hepatic clearing through the asialoglycoprotein receptor.^20,–,22

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cartoon of human fetuin-A/α₂-HS glycoprotein showing cystatin-like domains 1 and 2 and a third unrelated domain in green, yellow, and blue shading, respectively. The cotranslational and posttranslational modifications disulfide bridges (C-C in yellow) Ser-phosphorylation sites (S in red), protease-sensitive sites (R-K, dibasic tryptic cleavage site; L-L, chymotryptic cleavage site; R-T, furin-sensitive cleavage site; all in green), allelic variants (T230/T238; M230/S238) depicted as orange asterisks, and Asn-linked complex N-glycosylation sites, and Ser/Thr O-glycosylation sites depicted as blue symbols, respectively. Data taken from.(^6,7,13,96

Binding Properties of Fetuin-A

The type 3 members of the cystatin superfamily are glycoproteins produced mainly in the liver, which circulate in plasma at high concentrations. Fetal calf serum contains more fetuin-A than albumin. Apart from the vasculature, fetuins are present throughout the extracellular spaces and the extracellular matrix. Given the high expression levels and the many possible ways of molecular interaction with multiple ligands, it is fair to assume that fetuins primarily exercise carrier and scavenger functions like albumin. This poses an important complication for ex vivo experimentation, because crude fetuin preparations contain impurities. Like pure albumin preparations, pure fetuin preparations are difficult to make. High-abundance plasma protein preparations are notoriously contaminated with lower-abundance biological molecules that copurify either because of natural association or because of mere coincidence. Various growth factors or growth-promoting substances associate with crude fetuin preparations and form the basis of the “enigmatic growth promoting properties” of fetuin in cell culture.²³

Whether associations of fetuin-A with certain ligands are physiologically relevant is a matter of controversy and is almost impossible to decide unless genetic models are developed to test such interactions in vivo. We remind readers to heed the old wisdom of Racker²⁴ that also became part of the “10 commandments” of Kornberg,²⁵ “Do not waste clean thinking on dirty enzymes.” Most commercial preparations of fetuin-A on the market today do not go much beyond the quality of “fetuin” from 1944,¹ which is better viewed as a “protein concept” like “globulin” or “albumin” rather than a clean product. We distinguish between fetuin and fetuin-A whenever possible. Also, before the year 2000 when fetuin-B was cloned in silico²⁶ and the year 2003 when it was finally shown to be expressed as a plasma protein,¹⁰ both fetuin-A and fetuin-B were collectively addressed as “fetuin,” and most fetuin sold today is not screened for the presence of fetuin-B. Studies of human α₂-HS-glycoprotein/fetuin (α₂-HS glycoprotein) will always signify fetuin-A; publications studying “fetuin” protein are probably dealing with fetuin-A as well, but it is impossible to exclude that fetuin-B was also studied because of the physicochemical similitude of both proteins.

Fetuin-A binding is prodigious and reaches from small molecules to entire organisms. Plasmodium sporozoites use fetuin-A binding to “hitch-hike” their way into liver cells,²⁷ suggesting that fetuin-A function is indispensable and despite negative selection pressure has been maintained long enough for this docking mechanism to evolve. Soon after its discovery, fetuin has been shown to inhibit trypsin.²⁸ Both positive and negative fetuin interactions with proteases are documented, including matrix metalloproteinase-9,²⁹ matrix metalloproteinase-3,³⁰ meprin metalloproteinases,³¹ the cysteine proteases m-calpain,³² cathepsin L in rat bone,³³ and inhibition of recombinant human cathepsin V. Proteinase interactions are thought to be involved in the regulation of tumorigenesis and tumor progression.^34,35

Fetuin markedly accelerates incorporation of exogenous fatty acids into cellular triglycerides.^36,–,39 Thus, fetuin may share a function with fatty acid binding proteins, a family of abundantly expressed 14-kDa to 15-kDa proteins. Like fetuin, fatty acid binding proteins reversibly bind hydrophobic ligands, including saturated and unsaturated long-chain fatty acids, and other lipids with high affinity.⁴⁰

Because of their rich complex glycosylation pattern, fetuins serve as model substances for lectin and glycoprotein research. Lectin binding should always be seriously considered when fetuin binding to cells and to the extracellular matrix is studied. On a practical note, the strong binding of pertussis toxin to terminal sialic acid residues in fetuin form the basis of a Food and Drug Administration-approved pertussis toxin test.^41,42 Fetuin-A sequestration of lectins proved a major complication in experimental cytotoxic therapy using cancer cell–specific antibodies coupled to the Ricinus communis agglutinin ricin.⁴³ The immunotoxins were rapidly cleared by the asialoglycoprotein receptor and caused liver toxicity.⁴⁴

In a search for natural transforming growth factor-beta (TGF-β) receptor antagonists, a sequence homology was found between TGF-β receptor type II and fetuin-A.⁴⁵ The TGF-β receptor II homology 1 domain from fetuin bound preferentially to bone morphogenetic protein (BMP)-2. Full-length fetuin-A bound directly to TGF-β1 and TGF-β2 and with greater affinity to the TGF-β–related BMP-2, BMP-4, and BMP-6. Finally, and likely impinging on fetuin's role in mineralization biology, fetuin or neutralizing anti-TGF-β antibodies blocked osteogenesis and deposition of calcium-containing matrix in mineralizing cell cultures.⁴⁶ An altered bone phenotype in fetuin-A–deficient mice (Ahsg^−/−) was explained accordingly in terms of failure to block TGF-β–dependent signaling in osteoblastic cells.⁴⁷ Tumorigenesis experiments using Ahsg^−/− mice further supported the hypothesis that fetuin-A is an antagonist of TGF-β in vivo, in that it inhibited intestinal tumor progression.⁴⁸

Antiinflammatory Role of Fetuin-A

Fetuin-A, one of the most abundant fetal plasma proteins, was found to be essential for the inhibition of the proinflammatory cytokine tumor necrosis factor production by spermine and its synthetic analogues.^49,50 Accordingly, the strong fetal expression of fetuin and spermine have been associated with the tolerance of the fetus, “nature's transplant,” by mothers.⁵¹ The strong antiinflammatory effects of fetuin were verified in vivo using several models of inflammation, including lipopolysaccharide-induced miscarriage in rats,⁵² carrageenan injection,⁵³ cerebral ischemic injury in rodents,⁵⁴ and cecal ligation and puncture in mice.⁵⁵ In all cases fetuin-A was associated with reduced inflammatory response and increased survival, and administering additional fetuin generally improved outcome.

Thus fetuin-A generally may be regarded as antiinflammatory.⁵⁶ The antiinflammatory property of serum α₂-HS glycoprotein/human fetuin-A was further supported by the demonstration that fetuin-A is a potent and specific crystal-bound inhibitor of neutrophil stimulation by hydroxyapatite crystals.⁵⁷ Calcium phosphate crystals induce proinflammatory cytokine secretion through the NLRP3 inflammasome in monocytes/macrophages,^58,–,60 cell death in human vascular smooth muscle cells,⁶¹ and cell activation in chondrocytes.^62,–,64 Antiapoptotic activity of fetuin-A has been observed in smooth muscle cells⁶⁵ and dampening of the cell-specific responses would generally be expected to alleviate the detrimental consequences of local inflammation, cell death, and cartilage degradation. The proven protective function of fetuin-A in many animal models of inflammation,^52,–,55 the inhibition of proinflammatory compounds,^{50,51,66,–,68} and the inhibition of crystal-induced neutrophil activation⁵⁷ collectively suggest that fetuin-A may generally protect during pathological mineralization as well.⁶⁹

Fetuin-A in Metabolic Syndrome

Several clinical studies proposed that high-serum fetuin-A levels are associated with metabolic syndrome (MetS) and that fetuin-A therefore may present a risk factor for MetS.^70,–,73 The association of fetuin-A, insulin signaling, diabetes type 2, and MetS dates back to a publication in 1989 stating that pp63, a liver-secreted phosphoprotein, inhibited insulin receptor and downstream substrate phosphorylation signaling.¹⁷ The cDNA sequence of pp63 was identified as rat fetuin-A,⁷⁴ and it was disputed whether the activity tested in the original article was actually rat fetuin-A or a copurified protein.^75,76 We tested authentic human fetuin-A and found minor, if any, inhibitory activity at the insulin receptor and downstream substrate level.⁷⁴ We analyzed phosphotyrosine modification of cellular proteins in rat fibroblasts expressing the human insulin receptor and could not detect any robust and reproducible inhibition of insulin signaling by human or bovine fetuin including phosphofetuin-A immunoaffinity purified from HepG2 cells. Testing of fetuin-A in relevant cells showed no classical insulin receptor inhibition, but showed unexplained downregulation of insulin-stimulated Elk1-phosphorylation.⁷⁷

In another series of experiments, commercial bovine fetuin and supernatants of baculovirus-infected insect cells expressing human fetuin-A did inhibit insulin receptor signaling. Furthermore, fetuin-A was shown by immunoprecipitation to directly interact with the insulin receptor.^18,78 These researchers also showed that fetuin-A–deficient mice maintained on a mixed genetic background of C57BL/6N and 129Sv had improved insulin response, resistance to weight gain, protection against obesity, and protection against insulin resistance associated with aging.⁷⁹ We back-crossed these mice that were originally generated in our laboratory⁸⁰ onto pure C57BL/6N as well as DBA/2 genetic background, and we could not detect evidence of diabetes-associated symptoms in the genetically more homogeneous mice. The genetic background of mice and even substrains of mice differ with respect to insulin sensitivity. A deletion variant of nicotinamide nucleotide transhydrogenase that has spontaneously occurred in C57BL/6J, but not in C57BL/6N strains, is thought to influence insulin signaling.^81,82 Therefore, sibling mice from the same colony should be ideally used when comparing the influence of single gene deletions on insulin sensitivity. Mice with homogeneous defined genetic background were not yet available at the time of the first study in the case of fetuin-A–deficient mice.⁸³

Thus, it is unclear if fetuin-A is a cause or consequence of high-caloric feeding in mice or whether the association of high-serum fetuin-A and MetS is a bystander phenomenon. First, fetuin-A is a highly expressed constitutively secreted hepatic serum glycoprotein. Using nuclear run-on assays, LeCam et al⁸⁴ determined that the rat fetuin-A/pp63 promoter strength was approximately three-times to four-times that of the strong liver-specific albumin promoter. Second, fetuin-A is traditionally regarded as one of the few negative acute phase proteins^55,85,86 and strong associations between low-serum fetuin-A levels and inflammatory markers like C-reactive protein have been published.^87,–,89 Thus, downregulation rather than upregulation of fetuin-A during “fat inflammation” in obesity-related insulin resistance would be expected.⁹⁰ Therefore, it seems counterintuitive that fetuin-A levels would be increased in MetS. Recent work showed that fetuin-A induced inflammatory cytokines,⁹¹ and that one of the major mediators of inflammatory cytokine action, NFκB, further upregulated hepatic fetuin-A synthesis.³⁹ The conflicting data on fetuin-A regulation clearly need to be reconciled in terms of timing of events and causal relationships vs mere association. The association of fetuin-A serum levels with MetS and type 2 diabetes in humans rests on single measurements, and most studies did not correct for a major confounder, total liver protein synthesis. Therefore, we suggest that association studies should at least be normalized to serum albumin as an indicator of total liver protein synthesis, unless serum albumin should also be called a risk factor of MetS.

Role of Fetuin-A in Mineralization Biology

We started working on fetuin-A as part of an ongoing research project studying the structure and function of type 3 members of the cystatin superfamily.^6,92,–,96 At that time, fetuin-A/α₂-HS glycoprotein was a “structure in search of a function.” Intrigued by the description of rat fetuin as a natural inhibitor of insulin receptor kinase inhibitor,¹⁷ and by the report on the identity of fetuin of hemonectin, a protein present in rabbit bone marrow extracellular matrix and a suspected homing factor for myeologenic cells in the bone marrow,⁹⁷ we studied fetuin-A binding to cell membranes and extracellular targets. Binding to mineralized bone matrix as described by Triffitt et al^98,–,100 turned out to be the most robust binding phenomenon by far. The high affinity of fetuin-A for bone mineral mediates selective fetuin-A accumulation from plasma into bone.^98,99,101 Fetuin-A in bone accounts for 25% of the noncollagenous proteins and thus is one of the two most highly abundant noncollagenous proteins.¹⁰² We showed that fetuin-A also has a particularly high affinity to nascent apatite mineral and is an inhibitor of de novo apatite formation from supersaturated mineral solutions.¹⁹ The inhibitory effect is mediated by acidic amino acids clustering in cystatin-like domain D1. This fundamental observation formed the base of two decades of fetuin-A mineralization research and has been successfully repeated by many independent laboratories. In full accord with this biochemical finding, genetic work using fetuin-A–deficient mice^80,103 likewise suggested that fetuin-A is a mineral chaperone¹⁰⁴ mediating the transport of mineral from the extracellular space and the general circulation.^103,105,106

Fetuin-A–Mineral Complexes, Calciprotein Particles, Calcifying Nanoparticles: Fetuin–Mineral Complexes, Calciprotein Monomers, Calciprotein Particles

Fetuin-A binds calcium phosphate and calcium carbonate with high affinity, and it binds magnesium phosphate less well.¹⁹ Importantly, fetuin-A only inhibits the de novo formation of calcium phosphate and does not dissolve preformed mineral. Fetuin-A is an inhibitor of mineralization in solution and of cell-mediated mineralization in that it regulates the process of matrix mineralization in rat calvaria osteoblastic cells.¹⁹ On addition of β-glycerophosphate and ascorbate, these cells formed mineralized bone nodules in DMEM culture medium containing 10% fetal calf serum or bovine fetuin-A or human α₂-HS glycoprotein. However, when the cells were maintained serum-free and serum albumin was the sole bulk protein present in the medium, the cells catastrophically calcified and died.¹⁹ Fetuin-A inhibition of lethal calcification was dose-dependent in that higher concentrations of fetuin more consistently inhibited calcification than lower concentrations. Fetuin-A similarly inhibited vascular smooth muscle cell calcification induced by elevated concentrations of extracellular mineral ions.⁶⁵ This was achieved in part through inhibition of apoptosis and caspase cleavage. Fetuin-A was internalized by vascular smooth muscle cells and concentrated in intracellular vesicles. Subsequently, fetuin-A was secreted via vesicle release from apoptotic and viable vascular smooth muscle cells. The presence of fetuin-A in vesicles abrogated their ability to nucleate calcium phosphate precipitation. In addition, fetuin-A enhanced phagocytosis of vesicles by vascular smooth muscle cells. These observations showed that the uptake of fetuin-A by cells and the vesicular recycling of fetuin–mineral complexes reduced both apoptosis and calcification in live cells subjected to elevated extracellular concentrations of mineral ions.¹⁰⁷ Therefore, the stabilization of mineral solutions mediated by fetuin-A in test tube assays¹⁹ was also effective inside living cells, thus constituting a strong defense mechanism against pathological mineralization.

Fetuin-A stabilizes supersaturated mineral solutions by forming soluble colloidal nanospheres.⁵ Similar nanoparticles were detected in the ascites fluid of a pertioneal dialysis patient experiencing calcifying sclerosing peritonitis.^108,109 Figure 3 shows electron micrographs of nanoscopic protein–mineral particles generated in the laboratory (Figure 3A, B) or taken from the patient's ascites fluid (Figure 3C). These colloids are variously termed calciprotein particles in analogy to lipoprotein particles^{5,19,109,–,113} or fetuin–mineral complexes^114,–,118 or, most recently, calcifying nanoparticles¹¹⁹ or nanons.^120,121 Aspartic acid and glutamic acid residues located in the amino-terminal cystatin-like domain D1 are required for the inhibitory activity of fetuin-A.^5,19 Using a wide array of high-resolution spectroscopic techniques including small-angle neutron scattering, small-angle X-ray scattering, dynamic light scattering, transmission electron microscopy, and conventional photometry, it was determined that the formation of calciprotein particles is a multistep process.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Electron microscopic pictures of synthetic and patient-derived calciprotein particles (CPP). A, Primary calciprotein particles have a diameter of 30 nm to 150 nm and contain amorphous mineral.⁵ B, Subsequently, the primary CPP were subjected to a major structural rearrangement. The resulting secondary CPP consist of a core of crystalline basic calcium phosphate, which is covered by a layer of fetuin-A (A and B: 50 μmol/L bovine fetuin-A, 10 mmol/L CaCl₂, 6 mmol/L Na₂HPO₄, pH 7.4).¹¹³ These particles are stable for days at 37°C. C, Similar particles have been detected in ascites of patients with sclerosing calcifying peritonitis.^108,109 Bars represent 100 nm.

Figure 4 depicts schematically the sequence of calciprotein particles/fetuin–mineral complexes formation, starting with the spontaneous formation of prenucleation clusters from metastable solutions, their sequestration to acidic aspartic acid, and glutamic acid side chains in the cystatin-like domain 1 of fetuin-A–forming mineral-laden fetuin-A monomer or calciprotein monomer. Calciprotein monomer was discovered when quantitative small-angle neutron scattering data analysis revealed that even at a fetuin-A concentration close to the stability limit of supersaturated salt solutions, only approximately one-half of the mineral ions and only 5% of the fetuin-A were contained in the calciprotein particles. The remaining supersaturated mineral ion fraction and 95% of noncalciprotein particles of fetuin-A were associated with a mineral-laden fetuin-A monomer fraction that could be separated from mature calciprotein particles by ultrafiltration through a 300-KDa cut-off membrane. Small-angle neutron scattering analysis showed that the fetuin-A monomer in this fraction was closely associated with coalesced subnanometer-size mineral ions clusters reminiscent of Ca₉(PO₄)₆ Posner clusters (Figure 4A).¹¹¹

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Formation of primary and secondary calciprotein particles (CPP) from fetuin-A monomer and amorphous mineral precursors. A, Hydroxyapatite precursor viewed along the [001] axis. The dotted lines confine the hexagonal unit cell. The circle represents a Ca₉(PO₄)₆ Posner cluster as a building block of the apatite structure. B, A three-dimensional view of five Posner clusters juxtaposed to the acidic β-sheet of the computer-modeled aminoterminal fetuin-A domain D1. The domain surface is plotted semitransparent with negative charge indicated in red and positive charge indicated in blue. C, A computer-generated homology model structure illustrating the fetuin-A–mineral interface. The fetuin-A protein structure was modeled after published structures for cystatin-like domains 1 (green) and 2 (yellow).¹⁷⁶ Domain 3 (blue) was modeled after the Escherichia coli protein malonyl-CoA:acyl carrier protein transacylase (1MLA).¹⁷⁷ Fetuin-A does not influence the formation of mineral nuclei. However, fetuin-A does prevent the rapid growth and aggregation of nuclei and, thus, mineral precipitation.¹¹² D, Primary CPP are spherical and rather unstructured agglomerates of mineral (clusters) and fetuin-A, whereas (E) secondary CPP consist of a crystalline mineral core covered by fetuin-A. In conclusion, fetuin-A effectively shields the mineral phase, leading to stable CPP that can be transported and cleared, as shown in Figure 5. The computer graphics were generated using the software packages VESTA,¹⁷⁸ VMD,¹⁷⁹ and APBS.¹⁸⁰

Fetuin-A binds to apatitic mineral surfaces, as shown in Figure 4C. Fetuin-A does not influence the formation of mineral nuclei. However, fetuin-A prevents the growth and aggregation of nuclei to larger entities and ultimately mineral precipitation.¹¹² Thus, fetuin-A effectively shields spontaneously formed mineral nuclei, leading to transiently stable calciprotein particles.

A hierarchical role of fetuin-A was established in that fetuin-A was critically required during the early steps of calciprotein particles formation and stabilization and that other acidic plasma proteins including serum albumin further stabilized the initial colloids and could substitute for fetuin-A at later stages of calciprotein particles formation.^109,122 These later stages were studied in detail using small-angle neutron scattering. Figure 4D illustrates the aggregation of many calciprotein monomers and other mineral nuclei into transiently stable primary calciprotein particles of initially <100 nanometers in diameter. After a lag period, these particles grow into elongated and more crystalline particles of approximately twice the initial size, termed secondary or mature calciprotein particles (Figure 4E). Thus, formation and maturation are two separate and successive processes following the principles of Ostwald ripening.^{5,110,–,112} Similar particles may be generated with other acidic macromolecules as well. However, fetuin-A is exceptionally potent regarding activity and specificity of inhibition. Increased fetuin-A concentration leads to smaller particles. An increased temperature, mineral ion concentration, and a reduced fetuin-A concentration, respectively, all accelerate the particle ripening process, demonstrating that calciprotein particle maturation follows the Arrhenius law. Secondary calciprotein particles are stabilized by a compact outer fetuin-A monolayer against further growth (Figure 4E) for up to 30 hours at body temperature, which is ample time for clearing of calciprotein particles from circulation.

The formation and maturation of calciprotein particles can be followed by optical monitoring¹¹⁰ of mineralizing solutions. Supplemental Figure I (available online at http://www.circresaha.org) shows dynamic light scatter diagrams illustrating that the transformation from primary to secondary calciprotein particles is fetuin-A concentration-dependent. Higher fetuin-A concentrations cause prolonged stability of both calciprotein monomers and calciprotein particles, and thus a right-shift in the transformation curves. A clinical test could use the kinetic parameters of the precipitation reaction, thus measuring the overall calcification risk in biological fluids, eg, in the blood of dialysis patients.

Protein–mineral complexes are soluble precursors of physiological mineralization of bones and teeth as well. Recent research has shown that biomineralization starts with amorphous mineral–protein complexes containing fetuin-A.^123,–,126 Price et al^127,–,129 have shown that fetuin-A is critically required to direct mineralization to the interior of synthetic matrices that have size exclusion characteristics similar to those of collagen. Fetuin-A does so by selectively inhibiting mineral growth outside of these matrices. This mineralization by inhibitor exclusion is likely redundant, because fetuin-A–deficient mouse bone is mineralized perfectly well. The mice nevertheless show a bone phenotype of stunted femur length, indicating premature growth plate mineralization and disturbed osteogenic signaling.⁴⁷

Nanobacteria: A Red Herring From Mars

Nanobacteria initially described nanoscopic life forms detected by electron microscopy in rock sediments.¹³⁰ Similar entities were discovered in meterorites from Mars¹³¹ and, finally, in cell culture.¹³² Nanobacteria attracted a lot of scientific and economical attention as a causal agent of major diseases, including vascular calcification, kidney stones, and cancer.^132,–,136 In the year 2000, Cisar et al¹³⁷ determined that simple mixtures of phospholipids and calcium phosphate crystals closely resembled nanobacteria in that they showed life-like growth and replication. Nucleic acid sequences previously thought to be diagnostic markers of nanobacteria were in fact diagnostic of common laboratory contaminants.¹³⁷ Two groups of researchers finally solved the riddle of pathological mineralizing nanobacteria.^120,138 A series of experiments on the origin of putative nanobacteria showed that calcium phosphate together with proteins and further nonmineral compounds formed nanoparticles that resembled nanobacteria in shape and behavior.¹³⁹ Minerals containing nanoparticles had a high binding capacity for charged molecules, including ions, carbohydrates, lipids, and nucleic acids. Depending on the exact composition, mixtures of mineral and nonmineral compounds sustained either crystallization of hydroxyapatite mineral or the formation of complex protein–mineral complexes.^122,140 It was found that the main protein component of the nanobacteria was fetuin-A.¹²⁰ Besides fetuin-A, serum albumin and apolipoproteins were also identified;¹²¹ hence, the term nanobacteria was exchanged for the more apt term calcifying nanoparticles, virtually identical with the protein–mineral complexes, calciprotein particles, or fetuin–mineral complexes described before.

Calciprotein Particles Metabolism and Clearing

Figure 5 illustrates the putative metabolism of calciprotein particles, soluble protein–mineral complexes, which are now regarded as a physiological byproduct of mineral metabolism. The scheme is partly hypothetical and requires experimental verification. Soluble protein–mineral complexes including fetuin-A have been detected in serum from etidronate overdosed rats,¹¹⁸ in adenine-treated rats,¹⁴¹ in peritoneal dialysis patients with sclerosing calcifying peritonitis,^108,109 and in dialysis patients.¹⁴² The scheme is modeled after lipoprotein particle metabolism, a well-known transport system for otherwise insoluble lipids like cholesterol esters.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Hypothetical pathway of calciprotein particle (CPP) circulation and clearing by endothelial cells and tissues resident phagocytes. Fetuin-A stabilizes CPP in the circulation and mediates their efficient uptake. In healthy individuals, CPP may form spontaneously and be present in small numbers throughout all tissues, or may occur substantially where bone resorption by osteoclasts (OC) is releasing high levels of calcium and phosphate by acid-mediated mineral dissolution. CPP may be transported in the blood or retrieved locally after nearby osteoclastic resorption, and are used locally by osteoblasts (OB) during bone formation. When mineral homeostasis is severely disturbed, eg, in dialysis patients or in severely etidronate-overdosed rodents, bouts of hypercalcemia and hyperphosphatemia will result in the formation of high numbers of CPP and fetuin-A will be consumed in the process as part of the CPP complex. CPP in the interstitial spaces will be cleared by monocytes (MC) or macrophages (MP) or by any other endocytosing cell types. Overly abundant CPP may overwhelm the clearing capacity of the reticuloendothelial system phagocytes and may result in apoptosis of endothelial cells (EC), MP, and smooth muscle cells (SMC), and in deposition of calcified apoptotic cell remnants. This pathway is hypothetical and needs to be experimentally verified. It may also be of significance that many inhibitors of calcification activate monocytes/macrophages and stimulate phagocytosis; therefore, this mode of calcified remnant clearance may be of similar importance. Reproduced from Jahnen-Dechent W, Schäfer C, Ketteler M, et al. Mineral chaperones: a role for fetuin-A and osteopontin in the inhibition and regression of pathologic calcification. J Mol Med. 2008;86:379–389.

The basic tenet of calciprotein particles metabolism holds that fetuin-A stabilizes mineral complexes and at the same time mediates their transport and clearing. Fetuin-A should be regarded as an opsonizing serum protein with a high affinity for mineral complexes and debris. The opsonizing properties of fetuin-A have been determined several decades ago.^143,–,145 Fetuin-A affects microparticle phagocytosis by dendritic cells,¹⁴⁶ phagocytosis of apoptotic cells by macrophages,¹⁴⁷ and opsonizes phospholipid particles.¹⁴⁸

Fetuin-A is, however, not generally sticky like the “big 12” plasma proteins, which adhere to most materials in blood contact,¹⁴⁹ and fetuin-A was also not listed among specific proteins interacting with Sepharose beads in a study that used quantitative mass spectrometry to determine bead proteomes.¹⁵⁰ Therefore, binding may be restricted to a narrow range of cationic, lipidic, or mineral ligands, and a few more interacting TGF-β–related cytokines mentioned earlier in this text. Given the high abundance of fetuin-A in plasma, any clearing mechanism would have to rely on conformational or structural changes in fetuin-A or on multivalent binding that turns low-affinity binding into high-avidity binding.

Uptake of fetuin-A by cultured human vascular smooth muscle cells has been demonstrated,^65,107,151 but the exact form of fetuin-A was not determined. Clustering of fetuin-A molecules on the surface of secondary calciprotein particles as demonstrated¹¹¹ ideally fulfills the ligand clustering required to increase binding strength. Our preliminary results of fetuin-A monomer and fetuin-A containing calciprotein particle clearing in vivo show that calciprotein particles are cleared vastly more efficiently than fetuin-A monomer. Nevertheless, specific receptors for calciprotein particle clearing discriminating against fetuin-A monomer remain to be determined.

Fetuin-A Knockout Phenotype

The use of animal models has uncovered major inhibitors of pathological calcification and their mode of action.¹⁵² Fetuin-A is one of the major systemic inhibitors of pathological mineralization, but it is not the only one. This is vividly illustrated by the history of fetuin-A–deficient mice (Ahsg^−/−). While establishing that fetuin-A is a regulator of mineralization,¹⁹ we also generated Ahsg^−/− mice to test this hypothesis in an animal model. For technical reasons, the first Ahsg^−/− mice had a mixed 129Sv × C57BL/6 genetic background designated 129,B6Ahsg^tm1mbl.⁸⁰ Out of an initial colony of approximately 200 mice, only approximately 10 female ex-breeders showed a spontaneous calcification phenotype.⁸⁰ All other mice were apparently normal. Force-feeding supraphysiological doses of active vitamin D (calcitriol) and a high-mineral diet (phosphate-rich) resulted in nephrocalcinosis and pulmonary and myocardial calcification of these mice.¹⁰³

Unsatisfied by this highly variable yet low-penetrating calcification phenotype, we decided to combine the fetuin-A deficiency with the calcification-prone DBA/2 genetic background.^153,–,155 For comparison, we also backcrossed the mice onto the widely used calcification-resistant genetic background C57BL/6 and thus generated two more Ahsg^−/− mouse strains, D2-Ahsg^tm1wja and B6-Ahsg^tm1wja. The latter mice behaved like the mixed genetic background mice in that they scarcely had any spontaneous calcification. They did, however, have calcification when challenged with heminephrectomy and a high-mineral diet.^105,106 In stark contrast, virtually all D2-Ahsg^tm1wja mice show spontaneous massive soft tissue calcification throughout their bodies.¹⁰³ This strain is arguably the most calcifying mouse strain in existence in that the myocardium of D2-Ahsg^−/− mice contains almost one-tenth the calcium content of normal bone.¹⁵⁶ Kidney, lung, skin, brown fat, pancreas, and reproductive organs are equally strongly calcified.

Supplemental Figure II (available online at http://www.circresaha.org) shows random examples of 11-month-old wild-type or Ahsg^−/− mice maintained against the genetic background C57BL/6 or DBA/2. Even from the low-resolution radiology photographs, the uniformity of genetic penetrance and the extent of the calcification in D2-Ahsg^−/− mice can be readily appreciated. Kidney and skin calcification are easily visible. Details may be glanced at higher magnifications of the electronic pictures provided with this article. We point out that the pictures were taken of live anesthetized mice. Therefore, myocardium, lung, and pancreas calcification are invisible or blurred because of breathing motion of the chest region. Surprisingly, the soft tissue calcification is not lethal like the arterial calcification of matrix GLA protein–deficient mice.¹⁵⁷ Calcification does, however, greatly compromise reproduction in mice aged 6 months and older. The 1-year survival rate of D2-Ahsg^−/− mice in our specified pathogen-free colony is 60% to 70% of all born pups, compared to 90% to 95% in DBA/2 wild-type mice.

These results showed that fetuin-A is a systemic and soluble inhibitor of pathological mineralization that is backed-up by other genetic factors rendering mouse strains prone to or resistant to dystrophic calcification. Integrative genomics of the so-called Dyscalc locus led to the identification of Abcc6 as the major gene determining dystrophic cardiac calcification.¹⁵⁸ Abcc6-deficient mice have soft tissue calcifications develop, like D2-Ahsg^−/− mice, albeit they are less extensive in both localization and extent. Interestingly, Abcc6 deficiency is associated with reduced plasma fetuin-A levels and calcifications can be partially corrected by overexpressing fetuin-A,¹⁵⁹ suggesting that fetuin-A acts downstream or in concert with Abcc6. Despite heavy early-onset and life-long progressing dystrophic calcification in Abcc6^−/− and especially in D2- Ahsg^−/− mice, true osteogenesis involving clearly identifiable osteochondrocytic cells has never been reported in these mouse models of calcification. This casts doubt on the now popular hypothesis that pathological calcification is always a form of ectopic osteogenesis. More likely, both genes seem to be involved in systemic mineral homeostasis and transport. Therefore, we have termed fetuin-A a mineral chaperone mediating the solubilization, transport, and elimination from circulation of otherwise insoluble minerals, much like apolipoproteins help in lipid transport and metabolism.¹⁰⁴

Because fetuin-A is highly expressed in bone and accounts for 25% of all noncollagen proteins of bone (noncollagenous proteins),¹⁰² we examined bone growth and remodeling phenotypes in mixed background 129,B6-Ahsg^−/− mice.⁴⁷ The skeletal structure of these mice appeared normal at birth, but abnormalities were observed in adult 129,B6-Ahsg^−/− mice. Maturation of growth plate chondrocytes was impaired, and femurs lengthened more slowly between 3 and 18 months of age. Previously, it had been found that fetuin-A is a soluble TGF-β/BMP-binding protein controlling cytokine access to membrane signaling receptors.^45,46,160 Hence, the altered bone phenotype was explained in terms of failure to block TGF-β–dependent signaling in osteoblastic cells. Mice lacking fetuin-A displayed growth plate defects, increased bone formation with age, and enhanced cytokine-dependent osteogenesis.⁴⁷

In view of the mineralization regulation effected by fetuin-A, we revisited the bone phenotype of pure-bred B6-Ahsg^−/− mice. These mice are unaffected by the secondary hyperparathyroidism and osteoporosis typical of D2-Ahsg^−/− mice.¹⁰³ We essentially reproduced all major findings of the previous study performed in 129,B6-Ahsg^−/− mice.⁴⁷ The mice displayed normal trabecular bone mass in the spine but increased cortical thickness in the femur. Bone composition and mineral and collagen characteristics of cortical bone were unaffected by the absence of fetuin-A. The long bones, especially femora, were severely stunted in B6-Ahsg^−/− mice compared to wild-type littermates, resulting in increased biomechanical stability. In addition, we determined increased mineral content in the growth plates of B6-Ahsg^−/− long bones, corroborating its physiological role as an inhibitor of excessive mineralization in the growth plate cartilage matrix, a site of vigorous physiological mineralization. We thus demonstrated that growth plate chondrocytes are prone to “pathological calcification” in the absence of fetuin-A, and that active mineralization inhibition is a necessity for proper long bone growth, at least in mice.

The fact that three of the most potent anticalcification compounds, matrix GLA protein,¹⁵⁷ pyrophosphate,¹⁶¹ and fetuin-A, all operate in and around growth plate chondrocytes suggests that this cell type is physiologically extremely prone to premature calcification. The pathomechanism of osteogenic vascular calcification likewise critically requires chondrogenic, but not necessarily osteoblastic, differentiation of vascular cells.^162,–,164 It will be interesting to revisit the influence of fetuin-A alone and in combination on chondrocyte mineralization in a way similar to that previously performed with primary osteoblasts¹⁹ and smooth muscle cells⁶⁵ as models of physiological and pathological mineralization, respectively.

Clinical Epidemiology of Serum Fetuin-A Levels and Polymorphisms

Approaching the clinical situation in humans and based on the outlined experiments, it is strongly suggested to generally view serum fetuin-A levels in combination with serum albumin as an indicator of nutritional state, as another negative acute phase protein, and as an indicator of overall hepatic protein synthesis, especially in dialysis patients. Patients in advanced stages of chronic kidney disease (CKD) clinically have the most serious cardiovascular and soft tissue calcifications develop, more than any other population, and it is now well-understood that the individual magnitude of calcification significantly corresponds with impaired survival in CKD.^165,–,167 Furthermore, CKD patients tend to be in a state of malnutrition and microinflammation or macroinflammation, or both, in which downregulation of proteins such as fetuin-A may be expected.¹⁶⁸

When the biochemical properties of fetuin-A became more and more apparent, it was a straightforward approach to observe the relationship between serum fetuin-A levels and outcomes in large dialysis populations. In a cohort of >300 hemodialysis patients, the lowest tertile of serum fetuin-A levels was associated with significantly increased all-cause and cardiovascular mortality.⁸⁹ Fetuin-A was also found to be inversely correlated with C-reactive protein levels, emphasizing its nature as a negative acute phase reactant. These data were subsequently confirmed, demonstrating mortality risk prediction by fetuin-A deficiency in nearly 300 incident dialysis patients (including patients using peritoneal dialysis).¹⁶⁹ In this study, a specific fetuin-A gene polymorphism (Thr256Ser) was shown to predict particularly low fetuin-A levels and to be associated with an adverse prognosis compared to patients carrying alternative polymorphisms. Hypoalbuminemia was strongly correlated with fetuin-A deficiency, suggesting the expected involvement in the malnutrition-inflammation-atherosclerosis syndrome in clinical fetuin-A deficiency. In a pure cohort of prevalent peritoneal dialysis patients, fetuin-A deficiency was also shown to be linked to features of the malnutrition-inflammation-atherosclerosis syndrome (low albumin, elevated C-reactive protein) and to cardiovascular events and mortality, respectively.⁸⁸ Moreover, this study demonstrated an association between low fetuin-A levels and the magnitude of valvular calcification, emphasizing the hypothesized link between progressive calcification and impaired outcomes. In a smaller cohort of hemodialyis patients, a significant association between coronary calcification and fetuin-A deficiency was shown.¹⁷⁰ Taken together, these observations indicate that fetuin-A deficiency may be an important inflammation-related link between cardiovascular calcification and mortality in patients using dialysis.

In patients with normal renal function as well as in predialysis CKD patients, the available information on the relationship between fetuin-A levels, the degree of calcification, and mortality is less clear. Recently, we have shown in the first prospective longitudinal study that low baseline serum levels of fetuin-A are associated with the increase of aortic valve calcification in 77 patients.¹⁷¹ However, in nearly 1000 patients from the Heart and Soul study focusing on patients at cardiovascular risk, mostly without renal dysfunction, high fetuin-A levels were found to be strongly associated with hyperlipidemia and features of the metabolic syndrome, but not with hard outcome parameters.¹⁷² In patients with diabetes mellitus spanning CKD stages 1 to 4, the magnitude of coronary artery calcification correlated with increased rather than with decreased fetuin-A levels.¹⁷³ This finding may be specific for a pure diabetic cohort in which a different pattern of calcium-regulatory systems may be implicated and in which downregulation of fetuin-A may be partially compensated or masked by overnutrition. However, these data may imply that fetuin-A upregulation initially acts as a systemic defense mechanism (early warning system) trying to protect from or counteract against vascular calcifications in their early stages. Such an interpretation is indirectly supported by immunohistochemical findings showing strong fetuin-A deposition, but not synthesis, in areas of vascular calcification^65,170 and may be better-understood once the metabolism of fetuin-A mineral complexes (calciprotein monomers, calciprotein particles/fetuin–mineral complexes) and concomitant fetuin-A serum depletion is known. A recent study suggested that serum measurements of fetuin-A have to take into account that mineral-bound fetuin-A may compete with free fetuin-A, and that centrifuge sedimentation of serum samples may considerably influence certain assay readouts but not others.¹⁴²

When the calcification burden increases beyond a certain point in chronic long-term CKD, compensatory systems such as fetuin-A release may finally become exhausted and consecutive fetuin-A deficiency may start a vicious cycle of even more progressive extraosseous calcification and fetuin-A consumption. If this hypothesis of initial adaptive fetuin-A overproduction and later exhaustion of the system is correct, then future research should attempt to identify factors influencing fetuin-A expression and secretion.

Fetuin-A Consumption and Downregulation in Calciphylaxis/Calcific Uremic Arteriolopathy

Perhaps the most intriguing experimental example of fetuin-A consumption was published by Price et al,¹¹⁸ who demonstrated transient increases of total serum calcium up to 10 mmol/L by high-dose vitamin D treatment in rats within a few hours and a return to normocalcemia within 1 day. A concomitant loss of half the serum fetuin-A that formed a high-molecular-weight fetuin–mineral complex was observed, and it was probably cleared by the reticuloendothelial system or deposited in the bone or in extraskeletal sites.¹¹⁵ A potential clinical example of fetuin-A consumption and exhaust may be calciphylaxis (calcific uremic arteriolopathy), a rare but potentially life-threatening syndrome characterized by progressive and painful skin ulcerations associated with media calcification of medium and small cutaneous arterial vessels.^174,175 Calciphylaxis primarily affects patients using dialysis or after renal transplantation. We reported fetuin-A deficiency with very low serum levels in a case series of calciphylaxis patients.¹⁰³ To demonstrate the functional calcification inhibitory capacity of fetuin-A deficiency, we used sera from calcific uremic arteriolopathy patients in an ex vivo ^[45]CaCl₂ radioisotope assay, which enables quantification of serum-induced inhibition of a calcium phosphate precipitation. These fetuin-A–deficient calcific uremic arteriolopathy sera were significantly less effective at inhibiting calcium phosphate crystal formation than sera from healthy subjects with appropriate fetuin-A concentrations. This lack of efficacy could be reversed by the addition of purified fetuin-A to the calciphylaxis sera in quantities restoring normal serum levels. In this context, it remained unclear whether calcific uremic arteriolopathy was initially triggered by a lack of fetuin-A in the circulation, whether the system was already exhausted by a major attempt to counteract this fulminant calcification process, or whether levels were low secondary to a calciphylaxis-induced systemic inflammatory reaction.