Ultrastructural and biochemical aspects of matrix vesicle-mediated mineralization

2.1. Ultrastructural properties of the matrix vesicles under electron microscopy

Some of these incipient mineral crystals were initially found associated with the inner leaflet of the matrix vesicle membranes, and it seems plausible that crystal nucleation would begin at that specific site. Matrix vesicle membranes are rich in acidic phospholipids such as phosphatidylserine and phosphatidylinositol, which have high affinity for Ca²⁺. The affinity of phosphatidylserine for Ca²⁺ is particularly high and can produce a stable calcium phosphate–phospholipid complex associated with the inner leaflet of the vesicle’s membrane [7] (Fig. 2A). The possibility that such complexes may play an important role in crystal nucleation has been pointed out before [11], [12]. However, the early phases of calcium phosphate nucleation inside the matrix vesicles may be rather amorphous [13], but gradually become crystalline (hydroxyapatite) in later stages. Calcium phosphate crystals formed inside matrix vesicles can grow through the addition of Ca²⁺ and PO₄³⁻, which enter the vesicles through the action of membrane transporters and enzymes. Inside the matrix vesicle, “needle-shaped” crystalline calcium phosphates form a stellate assembly, bud off radially from the vesicle’s interior, and then, rip through the plasma membrane to form mineralized nodules, also referred to as calcifying globules [6] (Figs. 1E and 2D). While the needle-shaped crystals are identified as crystalline structures by electron diffraction, freeze-substitution and cytochemical calcium detection methods such as κ-pyroantimonate combined with energy-dispersive X-ray spectroscopy may show them as non-crystalline structures containing calcium and phosphate [14], [15], [16].

According to observations of the osteoid in bone derived from the quick frozen-freeze substitution technique with electron energy loss spectroscopy, which enables elemental mapping at the molecular level, calcium was evenly distributed in the proteoglycan-rich, peripheral region of matrix vesicles, whereas phosphate was detected predominantly in collagen fibrils [17] (Fig. 2D–F). Therefore, it seems feasible that, in non-calcified sites, the extracellular meshwork of organic substances limits the production of hydroxyapatite and inhibits precipitation of calcified crystals by controlling the spatial distribution of Ca²⁺ and PO₄³⁻, even if the extracellular fluid is supersaturated with those ions. In addition, a biological mechanism of PO₄³⁻ supplementation and subsequent transport into matrix vesicles must be in place, since PO₄³⁻ is not abundant in the periphery of matrix vesicles. Later in the text, we will discuss about the biological actions of membrane transporters and enzymes to produce PO₄³⁻ and to warrant the influx of Ca²⁺ and PO₄³⁻ into the matrix vesicles.

2.2. Mineralized nodules developed from matrix vesicles

Matrix vesicles and developing mineralized nodules can be observed in the osteoid below mature osteoblasts (Fig. 1A). It seems that the osteoid provides an adequate microenvironment for the development of mineralized nodules in bone. Even though the vesicle’s membrane is ripped during the process of crystal growth, the mineralized nodules might retain the membrane-associated enzymes and transporters, which will be mentioned later in this text.

Mineralized nodule is a globular assembly of numerous needle-shaped mineral crystals that has been exposed to extracellular environment (Fig. 1E). It seems likely that the growth of mineralized nodules is regulated by a large number of extracellular organic materials in the osteoid. Among them, osteopontin is especially suited to the task of regulating mineralization, because it effectively inhibits apatite formation and growth [18], [19]. Osteopontin is localized in the periphery of mineralized nodules, where it might act as a blocker of excessive mineralization [20]. Since other organic materials can combine with osteopontin [21], they can form the so-called “crystal ghosts” [22], [23], [24]. Among these materials, osteocalcin is known for containing γ-carboxyglutamic acid and for their ability to bind to mineral crystals [25], [26], [27]. Warfarin, which is an inhibitor for γ-carboxylation of glutamine residues, induces an embryopathy consisting of nasal hypoplasia, stippled epiphyses and distal extremity hypoplasia when given to women in the first trimester of pregnancy [25], [28]. In our observations, the administration of warfarin resulted in the dispersion of numerous fragments of needle-shaped crystal minerals throughout the osteoid [29] (Fig. 3). Recently, γ-carboxylase-deficient mice revealed the same abnormality with disassembled, scattered crystal minerals in bone [30]. Therefore, osteocalcin may play an important role in the globular assembly of needle-shaped mineral crystals, probably binding the organic components of the crystal sheath together.

3.1. Biological function of tissue nonspecific alkaline phosphatase (TNAP) in matrix vesicle-mediated mineralization

One of the most important enzymes to initiate mineralization in bone is TNAP, a glycosylphosphatidylinositol anchor enzyme associated with cell membranes (Fig. 4). TNAP can hydrolyze various phosphate esters, especially pyrophosphate (PPi), and is responsible for the production of inorganic phosphate, i.e., TNAP serve as pyrophosphatase to generate PO₄³⁻ monomer. The resultant PO₄³⁻ is transported into the matrix vesicles by means of sodium/phosphate co-transporter type III (Pit1). Therefore, many believe TNAP is a potent inducer of mineralization.

In bone, TNAP activity has been detected on osteoblasts and matrix vesicles [31], [32]. Interestingly, the distribution of TNAP on the cell membrane is not uniform in osteoblasts, which reflect the cell polarity with distinct basolateral and secretory (osteoidal) domains (Fig. 4). It has been reported that the plasma membrane Ca²⁺ transport ATPase was restricted to the osteoidal domain of the osteoblastic cell membrane, while TNAP was predominantly present on the basolateral domain [33]. Consistently, using specific antiserum to TNAP [34], we could observe relatively intense immunoreactivity and enzymatic activity for TNAP on preosteoblasts and on the basolateral aspect of mature osteoblasts’ membranes [35]. Thus, in bone, the membranes featuring intense activity of TNAP are not identical to those that serve as sites for matrix vesicle formation, for matrix vesicles derive from the secretory (osteoidal) cell membrane of osteoblasts—which show weak TNAP enzymatic activity. While the actual reason for that difference is still a matter of debate, it could be that high levels of TNAP activity in the matrix vesicles would induce mineral crystals overgrowth within the osteoid.

Mice homogyzous for Tnap gene depletion have been generated [36], [37] and mimic severe hypophosphatasia, indicating that TNAP is likely involved in mineralization. While TNAP might act as a pyrophosphatase [38], other pyrophosphatases may exist. Tnap^−/− mice are born with intact bones, but gradually develop growth retardation and other skeletal deformities. Although the absence of endogenous TNAP activity did not result in complete lack of mineral uptake in skeletal tissues [37], Tnap^−/− mice revealed a severe disturbance of the growth plate, suggesting abnormal endochondral ossification [39]. It was shown that depletion of Tnap gene results not only in hypomineralization of the skeleton, but also in a severe disorder of mineral crystal alignment in growing long bones with a disordered bone matrix architecture [40]. Murshed et al. have reported that transgenic mice expressing Tnap in the dermis showed extracellular mineralization consisting of hydroxyapatite crystals [41]. Given the evidence, it seems that TNAP activity is essential for mineralization in bone, but it is still unknown why the cell membranes with intense TNAP activity in bone are not identical to those forming matrix vesicles.

Recently, basic research on TNAP provided a promising tool for clinicians. The FDA approved asfotase alfa (Strensiq) in 2015 for the treatment of hypophosphatasia caused by a rare hereditary mutation in the alkaline phosphatase gene. Patients with either perinatal or infant onset of the disease who were treated with asfotase alfa showed improvement in overall survival: 97% patients receiving the drug were alive at age 1 year compared with 42% of control patients selected from a natural history study group. Patients with juvenile-onset hypophosphatasia also experienced improved growth and bone health compared with patients in a natural history database [42], [43].

3.2. Biological action of ecto-nucleotide pyrophosphatase/phosphodiesterase 1, ENPP 1 in bone mineralization

ENPP 1 is a member of the ENPP family of proteins, which are conserved in vertebrates and hydrolyze pyrophosphate or phosphodiester bonds in various extracellular compounds such as nucleotides and phospholipids. ENPP1 is composed of two N-terminal somatomedin B-like domains (SMB1 and SMB2), a catalytic domain and nuclease-like domain (Fig. 5). ENPP1 participates in different biological processes through distinct sets of domains: the catalytic and nuclease-like domains act in bone mineralization, while the SMB-like domains for insulin signaling. The crystalline structure analysis of ENPP1 suggests that the nucleotides are able to be accommodated in a pocket formed by an insertion loop in the catalytic domain, explaining why ATPs are ENPP1’s preferred substrate [44].

In bone mineralization, ENPP1’s catalytic activity generates PPi – presumably using extracellular ATPs – and the resultant PPi inhibits mineralization by binding to incipient hydroxyapatite crystals and preventing their overgrowth [45], [46], [47] (Fig. 6). In our observations, ENPP1 immunoreactivity was seen throughout the cytoplasm and on the secretory (osteoidal) pole of mature osteoblasts, as well as in osteocytes (data not shown). The localization of ENPP1 predominantly on the osteoidal surface of osteoblasts may suggest that it inhibits the overgrowth of mineral crystals. The histological evidence that TNAP is not intensely localized on the osteoidal surface appears to be consistent with the distribution of ENPP1 on osteoblasts.

Severe ENPP1 deficiency was recently shown to be related to a syndrome of spontaneous infantile arterial and periarticular mineralization [48], [49]. They suggested that the PPi generated by ENPP1 activity in vascular smooth muscle cells and chondrocytes disrupts the growth of hydroxyapatite crystals. In addition, the linkage of genetic ENPP1 dysfunction to infantile arterial mineralization suggests abnormal PPi metabolism seems to be an important regulatory factor or inducer of osteoblastic differentiation in vascular smooth muscle cell.

Mice with Enpp1 gene depletion (Enpp1^−/− mice), also known as tiptoe walking (ttw/ttw) mice, develop progressive ankylosing intervertebral and peripheral joint hyperostosis and articular cartilage mineralization [50], [51], [52], [53], [54]. Mackenzie et al. [54] have recently reported on the details of the skeletal deformities seen in Enpp1^−/− mice: mineralization in arteries and in the kidney, ectopic formation of cartilage in the joints and spine, despite the reduced circulating calcium and phosphate levels. Also, there was hypermineralization of the talocrural joint and hypomineralization of the femur and tibia with reduced trabecular number, trabecular bone volume and trabecular/cortical thickness. Consistent with the marked increased in fibroblast growth factor (FGF) 23 mRNA, circulating FGF23 was significantly elevated in Enpp1^−/− mice [54].

3.3. Putative function of ankylosis protein (ANK) in pyrophosphate transport in cells

ENPP1 can be found not only on the cell surface but also in cytoplasmic regions, generating PPi in the both regions of the cell (Fig. 6). Ankylosis protein, ANK, which is encoded by the mouse progressive ankylosis (Ank) gene, appears to function as a multiple-pass, transmembrane PPi-channelling protein, allowing PPi molecules to pass through the plasma membrane from the cytoplasm to the outside of the cell [55], [56]. In mice, Ank mRNA is expressed in several tissues including the heart, brain, liver, spleen, lung, muscle and kidney, as well as in bone and cartilage such as the articular cartilage in the shoulder, elbow, wrist and digits. Since PPi and its derivatives are naturally potent mineralization inhibitors both in vivo and in vitro, ANK-mediated regulation of PPi levels provides a room for inhibition of excessive mineralization in several tissues [55]. Considering that infants carrying Ank mutations display a three- to five-fold decrease in extracellular PPi, the Ank gene appears to regulate both intra- and extracellular levels of an important inhibitor of hydroxyapatite crystal formation [55]. Thus, local PPi production naturally inhibits hydroxyapatite deposition, blocking undesirable mineralization in articular cartilage and other tissues. With loss of ANK activity, however, extracellular PPi levels attenuate, intracellular PPi levels rise, and unregulated mineralization takes place.

3.4. Ca²⁺ transport through annexins

Matrix metalloproteinase (MMP)-3 [57], TNAP [5], [16], [17], [32], [57], [58], [59], [60], annexins [61], phospholipase A2, carbonic anhydrase II and lactate dehydrogenase [62] are some of the relevant enzymes and proteins involved in matrix vesicle-mediated mineralization.

Among these molecules, annexins are Ca²⁺- and lipid-binding proteins involved in Ca²⁺ transport and serving as ion channels in the matrix vesicles (Fig. 6). Three annexins were identified in the matrix vesicles: annexin A2, A5 and A6 [63], [64], [65], [66]. In the first phase of matrix vesicle-mediated mineralization, mineral complexes appears on the inner leaflet of the matrix vesicle’s membrane. The affinity of phophatidylserine for Ca²⁺ is quite strong in the inner leaflet of the matrix vesicle membrane, which is enriched with anionic lipids [67], [68]. The annexin A5 shows Ca²⁺-dependent phosphatidylserine-binding properties and may play a central role in mineralization. However, Annexin A5^−/− mice did not demonstrate abnormal skeletal development; thus, other annexins could replace the biological functions of annexin A5 in knockout mice. Thus, further studies seem to be necessary to clarify the precise role of annexins during bone mineralization.