Phosphorylation-dependent inhibition of mineralization by osteopontin ASARM peptides is regulated by PHEX cleavage

Abstract

The SIBLING family (small integrin-binding ligand N-linked glycoproteins) of mineral-regulating proteins, which includes matrix extracellular phosphoglycoprotein (MEPE) and osteopontin (OPN), contains an acidic serine- and aspartate-rich motif (ASARM). X-linked hypophosphatemia caused by inactivating mutations of the PHEX gene results in elevated mineralization-inhibiting MEPE-derived ASARM peptides. Although the OPN ASARM motif shares 60% homology with MEPE ASARM, it is still unknown whether OPN ASARM similarly inhibits mineralization. In this study we have examined the role of OPN ASARM and its interaction with PHEX enzyme using an osteoblast cell culture model, mass spectrometry, mineral-binding assays, and computational modeling. MC3T3-E1 osteoblast cultures were treated with differently phosphorylated OPN ASARM peptides [with 5 phosphoserines (OpnAs5) or 3 phosphoserines (OpnAs3)] or with control nonphosphorylated peptide (OpnAs0). Phosphorylated peptides dose-dependently inhibited mineralization, and binding of phosphorylated peptides to mineral was confirmed by a hydroxyapatite-binding assay. OpnAs0 showed no binding to hydroxyapatite and did not inhibit culture mineralization. Computational modeling of peptide-mineral interactions indicated a favorable change in binding energy with increasing phosphorylation consistent with hydroxyapatite-binding experiments and inhibition of culture mineralization. Addition of PHEX rescued inhibition of mineralization by OpnAs3. Mass spectrometry of cleaved peptides after ASARM-PHEX incubations identified OpnAs3 as a PHEX substrate. We conclude that OPN ASARM inhibits mineralization by binding to hydroxyapatite in a phosphorylation-dependent manner and that this inhibitor can be cleaved by PHEX, thus providing a mechanistic explanation for how loss of PHEX activity in X-linked hyposphosphatemia can lead to extracellular matrix accumulation of ASARM resulting in the osteomalacia. © 2010 American Society for Bone and Mineral Research

Introduction

The extracellular matrix of bone and other mineralized tissues contains abundant noncollagenous proteins that influence the nucleation, growth, and shape of the carbonate-substituted hydroxyapatite crystals found within these tissues.1 These noncollagenous proteins are often particularly acidic, this state being conferred by a preponderance of aspartate and glutamate amino acid residues and by abundant posttranslational modifications—specifically phosphorylations and glycosylations. The resulting net negative charge of these proteins results in their ability to bind abundant calcium ions and to bind to the surfaces of calcium-containing inorganic mineral crystals found in bone, cartilage, teeth, and otoconia (and in pathologic calcification sites). Thus calcium and/or mineral-binding proteins either may promote crystal nucleation by sequestering calcium ions and increasing local calcium concentrations beyond a critical stabilization point that also involves other ions (critical nucleus size) or may inhibit crystal growth either by stabilizing nascent crystal nuclei or precursor mineral phases or by adsorbing onto growing crystal surfaces. Structural attributes of posttranslationally modified osteopontin (OPN), a highly phosphorylated acidic glycoprotein found in bones and teeth and at pathologic calcification sites, confer the potential to perform any of these roles.2 OPN is particularly abundant in bone but is also present in a variety of hard and soft tissues and body fluids, where it acts as a key regulator of both physiologic and pathologic calcification.2, 3

Cell-secreted human OPN consists of 298 amino acid residues, of which 75 (25%) are acidic.4 Recent experimental data have determined that human milk OPN possesses 36 phosphorylations.5 Similarly, rat bone OPN has been shown to contain an average of 11 phosphorylations per molecule distributed over 29 potential sites,6 whereas OPN purified from the murine MC3T3-E1 osteoblast cell line contains an average of 20 phosphorylations over 27 potential sites.7 Phosphorylation of OPN at serine residues plays a key role in its mineral-binding and crystal growth-regulating properties. Surprisingly, a recent study has shown that bone-derived OPN containing 13 phosphates inhibited hydroxyapatite formation, whereas milk OPN containing 28 phosphates promoted hydroxyapatite formation.8

In addition to its role in mineralization, OPN also has been shown to be involved in a variety of cellular processes in bone remodeling, cell adhesion, inflammation, and cancer metastasis.9 For these roles, and apart from the phosphorylations, a number of peptide domains in OPN appear to be functionally active and include an RGD integrin-binding motif, CD44 receptor recognition sites, a transglutaminase cross-linking site, and a polyaspartate mineral-binding domain. In addition, proteolytic cleavage sites in OPN have been identified for thrombin, matrix metalloproteinase 2 (MMP2), MMP3, and MMP7.10, 11 In addition to the mineral-regulating activities of OPN phosphorylation, evidence likewise has been reported that posttranslational modifications are essential to these other functions as well. More specifically, phosphorylation of OPN has been demonstrated to be required for osteoclast adhesion, cell migration, bone resorption, and integrin binding.9

Based largely on the work of Rowe and colleagues investigating matrix extracellular glycoprotein (MEPE), a new mineral-inhibiting functional domain has been identified in bone noncollagenous proteins—the acidic serine- and aspartate-rich motif (ASARM).12–14 This motif occurs in the mineral-regulating family of small integrin-binding ligand N-linked glycosylated proteins (SIBLINGs) that includes dentin matrix protein 1 (DMP1), bone sialoprotein (BSP), MEPE, and OPN (15). The only ASARM motif thoroughly characterized thus far is the MEPE ASARM.12, 14, 16, 17 We, and others, have shown that levels of the MEPE-derived ASARM peptide are elevated in the bones and serum of patients and mice with X-linked hypophosphatemia (human XLH, Hyp mouse model) having characteristic osteomalacia (hypomineralization and osteoidosis).14, 18 X-linked hypophosphatemia is caused by inactivating mutations in PHEX, a gene encoding for a protease responsible for the degradation and clearance of mineralization-inhibiting ASARM peptides.14 Confirmation for this comes from the demonstration that phosphorylated MEPE ASARM peptide inhibits osteoblast mineralization in vitro and in vivo and that this inhibition can be rescued by PHEX cleavage of the inhibitory peptide.14, 17, 19 Given this potent antimineralization activity of the MEPE ASARM peptide and the fact that other SIBLING proteins share extensive conserved homology (about 60%) in this region,13 it seems important to consider similar actions for this peptide derived from other noncollagenous bone and tooth proteins. Indeed, recent data have shown increased release of ASARM peptides from SIBLING proteins other than MEPE in X-linked hypophosphatemia.20

In this study we have examined the antimineralization effects of synthetic, differently phosphorylated forms of the OPN ASARM peptide and their interactions with the enzyme PHEX using an extracellular matrix–producing and –mineralizing osteoblast cell culture model, in vitro mineral-binding assays, and mass spectrometry. In addition, computational molecular modeling using Monte Carlo–based simulations was performed for the peptides in solution and when adsorbed onto the {100} face of hydroxyapatite. These experimental and modeling approaches have allowed us to determine a mineral-regulating function for the OPN ASARM peptide that can be modified via its enzymatic cleavage by PHEX. Moreover, we show the importance of the phosphate groups (as phosphoserine residues) of the OPN ASARM peptide in this regulation and that this phosphorylation leads to direct binding to hydroxyapatite crystals. Finally, we describe possibilities for the mechanism underlying this binding that involve multiple phosphate group linkages with calcium atoms in different lattice planes of the {100} crystallographic face of hydroxyapatite.

Materials and Methods

ASARM peptides and PHEX protein

Synthetic peptides derived from the human osteopontin ASARM motif sequence (residues 115 to 132)13 were synthesized (NeoMPS, Inc., San Deigo, CA, USA) by Fmoc solid-phase chemistry according to standard peptide synthesis procedures with either 0 phosphoserines (DDSHQSDESHHSDESDEL), 3 phosphoserines (DDSHQpSDESHHpSDEpSDEL), or 5 phosphoserines (DDpSHQpSDEpSHHpSDEpSDEL); these peptides were termed OpnAs0, OpnAs3, and OpnAs5, respectively. Phosphorylations at specific serine residues in our chosen peptide sequence were achieved using preformed, protected phosphoserine amino acids. Peptides were purified by high-performance liquid chromatography, and the integrity of the peptides was verified by mass spectroscopy. A similarly sized peptide sequence from a more neutrally charged portion of OPN (residues 198 to 215, LNGAYKAIPVAQDLNAPS, and designated OpnN) was synthesized and used as a control. The peptides used in this study and some of their properties are summarized in Table 1. A recombinant, secreted soluble form of human PHEX (courtesy of Drs P Crine and T Loisel, Enobia Pharma, Montreal, Canada) was expressed and purified as described previously.21 Heat-inactivated PHEX (iPHEX) was prepared by heating PHEX at 56°C for 1 hour.

Table 1. Amino Acid Sequences of the Peptides Used in this Study

Peptide	Sequence	Description	Phosphates	pI
OpnAs0	DDSHQSDESHHSDESDEL	Unphosphorylated ASARM (a.a. 115–132)	0	4.0
OpnAs3	DDSHQ(pS)DESHH(pS)DE(pS)DEL	Phosphorylated ASARM (a.a. 115–132)	3	3.3
OpnAs5	DD(pS)HQ(pS)DE(pS)HH(pS)DE(pS)DEL	Phosphorylated ASARM (a.a. 115–132)	5	2.6
OpnN	LNGAYKAIPVAQDLNAPS	Neutrally charged OPN peptide (a.a. 198–215)	0	6.7

Cell culture

MC3T3-E1 (subclone 14) murine calvarial osteoblasts.22, 23 (courtesy of Dr R Franceschi, University of Michigan, Ann Arbor, MI, USA) were maintained in modified essential medium (MEM, Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, Hyclone, Logan, UT, USA) and 1% penicillin-streptomycin (Invitrogen) at 37°C in a humidified atmosphere of 5% CO₂. All experiments were carried out at a plating density of 50,000 cells/cm². Cell differentiation and matrix mineralization were initiated at 24 hours after plating of the cells by replacing the medium with fresh medium supplemented with 50 µg/mL ascorbic acid (Sigma, Oakville, ON, Canada) as a cofactor for collagen fibrillogenesis and 10 mM β-glycerophosphate as a source of phosphate for mineralization. Medium, with or without peptide (see above), was changed every 48 hours over a 12 day time period.

Quantification of mineralization

After 12 days of culture, mineral was visualized by von Kossa staining using 5% silver nitrate solution (Sigma). For quantification of mineralization by measuring calcium deposited within the cell/matrix layer, cultures were decalcified with 0.5 N HCl, and calcium in the supernatant was determined spectrophotometrically (absorbance at 595 nm) using a calcium assay kit (Diagnostic Chemicals, Charlottetown, PEI, Canada). The calcium content of the cell/matrix layer was normalized to protein content as determined by the bicinchoninic acid (BCA) protein assay kit (Pierce, Rockford, IL, USA).

Assay for cell proliferation

Cell proliferation and viability in the presence of the peptide treatments were measured using the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide) assay.24 Briefly, cells were incubated with 0.5 µg/µL MTT in medium for 3 hours, solubilized with DMSO, and absorbance monitored at 550 nm.

Assay for collagen deposition

To assess cell differentiation in terms of extracellular matrix production and assembly, collagen matrix deposition was quantified by picrosirius red staining followed by extraction with 0.1 N NaOH and spectrophotometric measurement (at wavelength 562 nm) of released stain, as described previously.25 Bovine calf skin collagen type I (Sigma) was used as a standard.

Assay for alkaline phosphatase activity

To measure alkaline phosphatase enzymatic activity as a marker of osteoblast differentiation, cultures were washed three times with PBS and solubilized in 10 mM Tris, pH 7.4, 0.2% Igepal (Sigma), and 2 mM phenylmethylsufonyl fluoride. After sonication and centrifugation, alkaline phosphatase activity in the supernatant was determined spectrophotometrically (wavelength 405 nm) using p-nitrophenylphosphate (Sigma) as a substrate and bovine calf intestinal alkaline phosphatase (Sigma) as a standard. One unit of alkaline phosphatase hydrolyzes 1 µmol of p-nitrophenylphosphate/min at 37°C.

RNA isolation and RT-PCR

Total RNA was isolated using TRIzol reagent (Invitrogen) following the manufacturer's protocol. RNA was treated with DNase (Invitrogen), and 1 µg was reverse-transcribed and amplified using SuperScript one-step reverse-transcriptase polymerase chain reaction (RT-PCR) with Platinum Taq polymerase (Invitrogen). PCR products for Bsp (bone sialoprotein), Alpl (tissue-nonspecific alkaline phosphatase, also Tnap, Tnsalp, and Akpt2), Gapdh, and Ocn (osteocalcin, also bone Gla protein) were analyzed by 2% agarose gel electrophoresis. Primers (Invitrogen) used were as follows14: Bsp, 5′-AACAATCCGTGCCACTCA-3′ and 5′-GGAGGGGGCTTCACTGAT-3′; Alpl, 5′-GGGGACATGCAGTATGAGTT-3′ and 5′-GGCCTGGTAGTTGTTGTGAG-3′; Gapdh, 5′-CCACTCTTCCACCTTCG-3′ and 5′-GTGGTCCAGGGTTTCTTAC-3′; and Ocn, 5′-TGAACAGACTCCGGCG-3′ and 5′-GATACCGTAGATGCGTTTG-3′. Annealing temperature and cycles performed were 58°C/24 h for Ocn and Bsp and 55°C/30 h for Gapdh and Alpl.

Biochemical assay for protein binding to hydroxyapatite

Hydroxyapatite crystals (courtesy of Drs D Eanes and B Fowler, National Institute of Standards and Technology, Bethesda, MD, USA) were suspended in a reaction volume of 100 µL of buffer containing 20 mM Tris HCl and 150 mM NaCl, at pH 7.4, to which 60 µM of peptide was added. After a 1 hour room-temperature incubation followed by centrifugation at 10,000 × g for 5 minutes, protein concentration in the supernatant was measured using the BCA protein assay (Pierce).

Mass spectrometry

ASARM peptides were dissolved in water at a concentration of 10 µg/µL. Reactions were carried out in 100 µL reaction volumes using 250 µM peptide in zinc-free cleavage buffer (20 mM Tris, 150 mM NaCl, pH 7.4) and 25 µM purified recombinant PHEX. After a 1 hour incubation at 37°C, the reaction was quenched by the addition of 1 µL of 500 mM EDTA (pH 8.0) solution. Peptides were analyzed by liquid chromatography–mass spectrometry (LC-MS, Waters Corporation, Milford, MA, USA) or by matrix-assisted laser desorption/ionization quadruple time of flight (MALDI-QTOF). For MALDI-QTOF, samples were spotted and mixed with a matrix solution of 10 mg/mL α-cyano-4-hydroxycinnamic in 50% acetonitrile and containing 0.1% trifluoroacetic acid. After drying, samples were washed with two volumes of 2.5 µL 0.1% trifluoroacetic acid before analysis with a MALDI-QTOF Ultima instrument (Waters Corporation). Data acquisition and analysis were done with Masslynx software Version 4.0 (Waters Corporation). Tandem mass spectrometry (MS/MS) data analysis for peptide fragment sequence assignment was done using the Masslynx software companion BioLynx.

Energy calculation and structure prediction of OPN peptides

All simulations were performed using the RosettaSurface Monte Carlo plus-minimization structure-prediction program.26 Each execution of the program folds a peptide from a fully extended conformation and results in one energy-minimized candidate solution- and adsorbed-state structure. Large structural ensembles of 10⁵ candidate solution- and adsorbed-state structures were generated from which the 100 lowest-energy structures from each state were chosen for further analysis.

Results

Phosphorylated OPN ASARM peptide inhibits mineralization of MC3T3-E1 osteoblast cultures

To examine the role of OPN ASARM peptides (see Table 1) on extracellular matrix mineralization in osteoblast cultures, MC3T3-E1 cells were treated with the differently phosphorylated peptides (and the OpnN peptide) for 12 days, and mineral was visualized by von Kossa staining and quantified by a biochemical assay for calcium. As shown in Fig. 1A, phosphorylated peptides (OpnAs5 and OpnAs3) dose-dependently inhibited mineralization of the cultures, with OpnAs5 being more potent than OpnAs3. Inhibition by OpnAs5 was maximal after 1 µM, whereas inhibition by OpnAs3 was maximal after 20 µM. Nonphosphorylated ASARM (OpnAs0) and the neutrally charged control peptide (OpnN) had no effect on mineralization at comparable doses (see Fig. 1B). To further elucidate the mechanism by which phosphorylated peptides inhibit mineralization, MC3T3-E1 cultures were treated with 20 µM peptide either during days 0 to 6 (the matrix deposition and assembly stage), during days 6 to 12 (the matrix mineralization stage), or for the entire 12 day duration of the culture period. As shown in Fig. 1C, OpnAs5 inhibited mineralization dramatically only when the peptide was present during days 6 to 12 (i.e., the matrix mineralization stage). OpnAs3 also inhibited mineralization only when the peptide was present during days 6 to 12 (see Fig. 1D). This suggests that phosphorylated ASARM peptides block culture mineralization by affecting crystal formation and not by adversely affecting osteoblast differentiation and matrix deposition during the earlier matrix assembly stage.

Figure 1. Effect of OPN ASARM peptides on mineralization of osteoblast cultures. MC3T3-E1 osteoblast cultures were incubated with phosphorylated OPN ASARM (OpnAs5 and OpnAs3) or control (OpnAs0 and OpnN) peptides at the indicated concentrations for 12 days followed by (A) von Kossa staining for mineral in culture dishes and (B) calcium content determination from cell–matrix layer extracts (after acid hydrolysis/decalcification) expressed as a percentage of untreated control cultures. (C, D) von Kossa staining and calcium assays after 12 days where cultures were treated with 20 µM OpnAs5 (C) or 20 µM OpnAs3 (D) for either the first or the last 6 days of the 12 day culture period. Inhibition of mineralization correlates with the mineralization stage (days 6 to 12) and not the cell differentiation and matrix assembly stage (days 0 to 6). Data are presented as means ± SE.

Absence of an effect of ASARM peptides on osteoblast differentiation

Cell proliferation, as measured by the MTT assay, was normal after treatment with all peptides (Fig. 2); thus the peptides did not appear to be cytotoxic to the osteoblasts. To verify that the mineralization-inhibiting effect of the phosphorylated ASARM peptides was not attributable to a disruption in osteoblast differentiation and matrix secretion and assembly, which, in turn, would limit the amount of mineralization, we examined the effect of phosphorylated ASARM on collagen deposition and alkaline phosphatase activity—two prominent markers of osteoblast differentiation and function. As demonstrated in Fig. 3A, collagen deposition was abundant in all cases but slightly decreased in the OpnAs5-treated cultures. Alkaline phosphatase activity (see Fig. 3B) and mRNA expression (see Fig. 3C) were largely unaffected by all peptides. Later markers of osteoblast differentiation—osteocalcin and bone sialoprotein—also were examined by RT-PCR, and as shown in Fig. 3C, their mRNA levels showed no changes after peptide treatment. These data demonstrate that OPN ASARM peptides do not disrupt the ability of osteoblasts to assemble a collagenous matrix, express alkaline phosphatase, and differentiate.

Figure 2. Effect of OPN ASARM peptides on cell proliferation in MC3T3-E1 osteoblast cultures. Osteoblasts were cultured in the presence of 20 µM peptide for 12 days, during which the number of viable cells was assessed by the MTT assay. Data are presented as means ± SE.

Figure 3. Effect of OPN ASARM peptides on MC3T3-E1 osteoblast differentiation, collagen matrix assembly, and gene expression. Cultures were incubated with or without 20 µM peptide for 8 days, after which (A) collagen deposition was determined by quantification of picrosirius red staining and (B) alkaline phosphatase activity was measured using p-nitrophenylphosphate as a substrate. Data are presented as means ± SE. *p < .05; **p < .01 from Student's t-test relative to the untreated control. (C) MC3T3-E1 osteoblasts were treated for 12 days with peptide as indicated, after which RNA was extracted and RT-PCR analysis performed to examine gene expression of tissue-nonspecific alkaline phosphatase (Alpl), bone sialoprotein (Bsp), and osteocalcin (Ocn) relative to the housekeeping gene Gapdh.

Phosphorylation of OPN ASARM peptides is required for binding to hydroxyapatite

To quantify ASARM binding to hydroxyapatite, peptide was first incubated with synthetic hydroxyapatite crystals and then centrifuged to separate bound and unbound peptide. Increasing the amount of hydroxyapatite in the incubations caused a significant depletion of OpnAs5 and OpnAs3, whereas OpnAs0 and OpnN were unaffected (Fig. 4). These data demonstrates that the phosphoserine residues are required for the binding of OPN ASARM to hydroxyapatite.

Figure 4. OPN ASARM peptide binding to synthetic hydroxyapatite. Graph depicts peptide depletion from the supernatant after incubation of 60 µM peptide with differing amounts of hydroxyapatite (as indicated) in a 100 µL suspension for 1 hour at room temperature. Binding of peptide to hydroxyapatite is shown as a percentage of starting peptide concentration, as determined by BCA protein assay. Data are presented as means ± SE. *p < .05; ***p < .001 from Student's t test relative to the OpnN control.

Computational modeling of OPN ASARM binding to hydroxyapatite

To investigate the mechanism of OPN ASARM binding, we performed computational modeling of peptides binding to the hydroxyapatite {100} crystal face using a Monte Carlo plus-minimization structure-prediction algorithm. This approach has been used previously to characterize binding of the salivary protein statherin to hydroxyapatite, with results being consistent with experimental solid-state NMR data.26, 27 Calculated binding energies (Fig. 5A) show a trend of increasing adsorption energies with increasing peptide phosphate group content, which predicts that OpnAs5 binds more strongly than OpnAs3, which, in turn, binds much more strongly than OpnAs0. Addition of phosphoserine residues thus contributes favorably to the adsorption free energy of the peptide. Control peptide OpnN, without phosphate and having neutral charge, had the lowest binding energy. These binding energies are consistent with the experimental observations made in Fig. 4, demonstrating that phosphoserine residues enhance hydroxyapatite binding. Computational modeling also predicts that the OPN ASARM peptides are largely unstructured in solution but display a small increase in α-helical secondary structure on binding (see Fig. 5B). In contrast, the control peptide OpnN becomes less structured on adsorption. Figure 5C–F shows representative structures of OpnAs5, OpnAs3, OpnAs0 and OpnN bound to hydroxyapatite. The predicted helix structure of OpnAs3 and OpnAs0 also can be observed in Fig. 5D, E, respectively. Phosphoserine residues bind calcium in the crystal surface very strongly, although the carboxylate groups of the aspartate and glutamate also adsorb well. Glutamine, serine, and histidine residues also contribute moderate hydrogen bonding with the crystal surface. The transparent surface around the peptides represents the van der Waals boundary of the peptide, whose profile shows a high degree of surface complementarily of the bound phosphopeptides to the crystal surface. Figure 5G shows details of an interaction between a phosphate group in OpnAs3 coordinating with three calcium atoms in the crystal surface. In Fig. 5H, a degree of lattice matching of OpnAs3 is observed as a row of carboxylate and phosphate groups aligned with calcium atoms in the crystal surface. Coordinate files in Protein Data Bank format for the 10 lowest-scoring models for each of the four OPN peptides bound to the {100} face of hydroxyapatite are available as supplemental data.

Figure 5. Molecular modeling of OPN ASARM peptide binding to the {100} face of hydroxyapatite. (A) Peptide-hydroxyapatite adsorption energies in Rosetta energy units (REU). (B) Percent predicted peptide secondary structure in its solution state (Free) and hydroxyapatite-adsorbed state (Bound) averaged over the 100 lowest-energy structures. Representative models of hydroxyapatite-adsorbed (C) OpnAs5, (D) OpnAs3, (E) OpnAs0, and (F) OpnAsN. (G) Close-up view of a phosphate residue (PO₄) in OpnAs3 coordinating with hydroxyapatite calcium (Ca) atoms. (H) Close-up view of a segment of OpnAs3 demonstrating lattice matching of carboxyl (COOH) and phosphate (PO₄) side chains to aligned calcium (Ca) atoms. (Ca, green; P, orange; O, red; H, white). Images are oriented with the N-terminal of the peptides to the left and the C-terminal of the peptides to the right.

Soluble secreted PHEX rescues OPN ASARM inhibition of mineralization

The accumulation of ASARM peptides in the osteoid of XLH patients and hypophosphatemic (Hyp) mice is attributable to the absence of functional PHEX. In our bone cell culture model, addition of exogenous PHEX (20 µg/mL) to OpnAs3-inhibited cultures rescued the inhibition of mineralization (Fig. 6A). Interestingly, PHEX failed to rescue the inhibition of mineralization in OpnAs5-inhibited cultures (see Fig. 6B), indicating that PHEX might selectively modulate OPN ASARM activity in a phosphorylation-dependent manner. To confirm that the observed rescue of mineralization in OpnAs3-inhibited cultures was caused by PHEX enzymatic activity, a heat-inactivated PHEX control (iPHEX) was included. iPHEX did not rescue mineralization of OpnAs3- or OpnAs5-inhibited cultures.

Figure 6. Effect of PHEX on OPN ASARM inhibition of MC3T3-E1 osteoblast culture mineralization. (A) Cultures were treated for 12 days with either 20 µM OpnAs3 peptide alone or 20 µg/mL PHEX alone and 20 µg/mL heat-inactivated PHEX (iPHEX) alone or with both OpnAs3 and PHEX combined. Control cultures had no treatment with either OpnAs3 or PHEX. von Kossa staining for mineral in culture dishes (lower panel) and quantification of calcium content in acid-hydrolyzed/decalcified cell-matrix extracts (upper panel) show rescue of OpnAs3 inhibition of mineralization by PHEX. (B) Cultures were treated for 12 days with either 20 µM OpnAs5 peptide alone or 20 µg/mL PHEX alone and 20 µg/mL heat-inactivated PHEX (iPHEX) alone or with both OpnAs5 and PHEX combined. Control cultures had no treatment with either OpnAs5 or PHEX. von Kossa staining for mineral in culture dishes (lower panel) and quantification of calcium content in acid-hydrolyzed/decalcified cell-matrix extracts (upper panel) show an inability of PHEX to rescue OpnAs5 inhibition of mineralization. Data are presented as means ± SE. ***p < .001 from Student's t test relative to the untreated control.

Mass spectrometry showing proteolytic cleavage of OpnAs3 by PHEX

Since PHEX rescued OpnAs3 inhibition of mineralization, and given that we have shown previously that PHEX digests MEPE ASARM peptides,14 we examined whether OpnAs3 is a substrate for PHEX. For this, PHEX was incubated with OpnAs3 for 1 hour at 37°C, and cleavage products were analyzed by mass spectrometry. OpnAs3 was completely cleaved by PHEX at two distinct positions, resulting in the degradation products shown in Fig. 7. MS/MS analysis confirmed the peptide sequence of the resulting fragments (data not shown). At both cleavage sites, digestion occurred at an amide linkage between phosphoserine and aspartate. Cleavage of OpnAs0 and OpnAs5 by PHEX also was observed (data not shown).

Figure 7. Mass spectrometry showing cleavage of OpnAs3 peptide by PHEX. Addition of recombinant, soluble secreted human PHEX to OpnAs3 yields complete cleavage of all peptide at discrete amide linkages between phosphoserine and aspartate residues to generate the indicated peptide fragments.

Discussion

The OPN ASARM motif is a negatively charged stretch of amino acids that is highly homologous (60% amino acid sequence homology) with a region found in MEPE and other members of the SIBLING family of mineral-regulating proteins.13 Recent studies have demonstrated that the MEPE ASARM peptide is a potent inhibitor of mineralization in vitro and in vivo and also contributes to the hypomineralization (e.g., osteomalacia, osteoidosis) occurring in the pathologic condition of X-linked hypophosphatemia (XLH).14, 17, 20 Prior to this study, no reports existed on the function of ASARM peptides in other SIBLING proteins. Here, our data provide new insights into the role of the ASARM peptide from OPN on mineralization. Specifically, we demonstrate (1) that phosphorylated OPN ASARM inhibits extracellular matrix mineralization in an osteoblast culture model, (2) that OPN ASARM binding to mineral crystals depends on phosphorylation of serine residues, (3) that phosphoserine residues increase the adsorption free energy of the peptides, and (4) that inhibition of mineralization by phosphorylated OPN ASARM peptides can be rescued by proteolytic digestion by PHEX. While inhibition of mineralization by OPN peptides is not unique to the ASARM motif—other OPN phosphopeptides are known to inhibit hydroxyapatite formation in constant-composition autotitration asays28 and steady-state agarose gels29—an important difference relevant to the OPN ASARM work reported here is that the peptide is observed in vivo, and clinical correlations have been made with osteomalacia in mice and in human patients.

Previously, it has been demonstrated that MEPE ASARM inhibits in vitro mineralization of MC3T3-E1 osteoblast cultures,14 2T3 osteoblast cultures,12 and bone marrow stromal cell cultures.16 In this study, we demonstrate that the ASARM motif from OPN is also a potent inhibitor of extracellular matrix mineralization in osteoblast cultures, thus providing the first functional evidence that this conserved peptide motif likely has a similar role across all members of the SIBLING protein family.

In terms of the potency of phosphorylated OPN ASARM peptide in inhibiting mineralization, maximal inhibition by OpnAs5 and OpnAs3 was observed at 1 and 20 µM, respectively, which is comparable with the inhibition reported previously by a triphosphorylated MEPE ASARM peptide, where inhibition was maximal at 20 µM. Importantly, this micromolar concentration range is similar to the observed ASARM serum levels in XLH patients and Hyp mice (15 and 23 µM, respectively).18

Phosphorylated OPN peptides also inhibit growth of calcium oxalate monohydrate crystals,30–32 a mineral phase typical of kidney stones. A common property among these investigated peptides is the repeated presence of clusters of negatively charged aspartate, glutamate, and/or phosphoserines. The presence of several such regions within OPN, and their conservation across the different SIBLING proteins, suggests a complementary role for these peptides plus a degree of functional redundancy. Indeed, despite having slight pertubations in mineralization, transgenic mice deficient in individual SIBLING proteins continue to show a high level of skeletal and dental mineralization whose regulation generally produces crystals with characteristics similar to normal crystals from wild-type mice.33, 34 While at most skeletal sites mineralization appears essentially normal in Opn-deficient mice, hypermineralization and slight increases in crystal maturity and size have been reported at selected osseous sites, as would be expected from the loss of a mineralization inhibitor.33 Likewise, in cell-free crystal growth studies using different model systems, full-length OPN (and OPN peptides) potently inhibit hydroxyapatite formation.28, 29, 35–37

Phosphorylation of OPN is ordered, with clusters of three to five phosphoserine residues that favor its interaction with calcium within the hydroxyapatite crystal lattice.5, 38 The observed phosphorylation sites match the recognition motifs for the enzymes Golgi kinase/mammary gland casein kinase and casein kinase II.39–41 The phosphorylation motifs found within the human OPN ASARM peptide include three Golgi kinase motifs and five casein kinase II motifs. Our results, showing a correlation in the number of phosphoserine residues to mineral binding and to potency of mineralization inhibition in osteoblast cultures, are consistent with previous data demonstrating that phosphorylation of MEPE ASARM and OPN is required for inhibition of in vitro mineralization.8, 14, 42 In support of this, dephosphorylation of OPN lowers its crystal growth-inhibiting potency in hydroxyapatite gelatin- and agarose-gel crystal growth systems,8, 36 in osteoblast42 and vascular smooth muscle cell culture models,43 and likewise in calcium oxalate growth systems.44 Differently phosphorylated forms of OPN have been detected in rat bone cell cultures,45 murine fibroblasts, and murine osteoblasts.7 Collectively, these data strongly suggest that the phosphorylation status of OPN and its peptides likely has an important mineralization-regulating function.

Phosphorylation of OPN appears directly related to its binding to hydroxyapatite, which, in turn, likely provides the mechanism for its inhibition of hydroxyapatite crystal growth. Besides the increased charge density provided by the phosphoserine residues, which clearly plays a role in this binding and inhibition, little is known beyond this electrostatic interaction. As shown in Table 1, addition of phosphoserine residues further lowers the isoelectric point of the peptide, thereby increasing its net negative charge. Our present data showing that nonphosphorylated OPN ASARM (OpnAs0) and a more neutrally charged OpnN peptide bind very minimally to hydroxyapatite and fail to inhibit mineralization are analogous to our previous results for MEPE ASARM peptide14 and together support the notion that negative charge density is particularly important for this binding and inhibition.31

Interestingly, atomic-scale molecular dynamics simulation of an OPN peptide (not ASARM) binding to a crystal surface (in this case calcium oxalate monohydrate) determined that although phosphorylation enhanced peptide adsorption, it was in fact the aspartate and glutamate residues that were predicted to be in closest contact with the crystal surface.31 It was suggested that the negatively charged phosphate groups initially attract the peptide to the crystal surface but that it is the acidic carboxylate groups of glutamate and aspartate that form the stable interactions. Our Monte Carlo–based predictions suggest that although aspartate and glutamate adsorb well, phosphoserine residues bind more strongly and contribute favorably to the binding energy. NMR analysis of OPN in solution determined that the protein was essentially completely unstructured46—this lack of conformation was suggested to be advantageous to a protein that binds to multiple protein partners related to its role in various cell and matrix biology processes and to different calcium-containing mineral chemistries related to its role in crystal growth regulation. In this study, we likewise show in our computational peptide simulations that OPN ASARM peptides, with or without phosphorylation, lack any readily identifiable structure in solution. This would suggest that phosphorylation does not favor a peptide-crystal interaction by inducing conformational change in the peptide structure in solution prior to mineral docking. These data are consistent with observations made by Hoyer and colleagues30 showing that although phosphorylation of OPN peptides enhanced inhibition of calcium oxalate monohydrate crystal growth, circular dichroism analysis failed to detect any changed in peptide secondary structure. Because of its consistent presence in mineral-binding proteins in various biomineralizing systems, poly-Asp is often used as a model compound for studying protein-mineral interactions. Circular dichroism of poly-Asp by Goldberg and colleagues36 also revealed no secondary structure, although poly-Glu, a nucleator of hydroxyapatite formation in cell-free agarose gel systems, displayed a partial left-handed helix. Lack of secondary structure in solution does not necessarily mean that the peptide also would be unordered once bound to mineral, and for the salivary protein statherin, a significant conformational change occurs on binding.26, 27, 47 For full-length OPN, using attenuated total reflection infrared spectroscopy, Gericke and colleagues8 have shown that binding of OPN to hydroxyapatite induces a slight increase in the percentage of β-sheet structure. In our study, computational modeling shows that the OPN ASARM peptides exhibit a small increase in α-helical secondary structure on adsorption to hydroxyapatite. Phosphorylation appears to decrease the percentage of α-helical structure, which is consistent with the observation that peptides are increasingly extended with increasing phosphorylation. Grohe and colleagues31 observed similar effects of phosphorylation on OPN peptides and suggested that electrostatic repulsion of phosphate groups within the peptide maintains the peptide in an extended conformation.

Considering the apparent functional redundancy of different OPN phosphopeptides as revealed in the literature, particularly significant in this study is the finding that the proteolytic activity of PHEX can rescue ASARM inhibition of mineralization. Indeed, mutations in PHEX lead to accumulation of SIBLING ASARM peptides in the hypomineralized bones of XLH patients and Hyp mice,47, 48 yet identification of a substrate for PHEX remained elusive for many years until we recently demonstrated that PHEX digests the MEPE ASARM peptide.14 Indeed, previous attempts to investigate PHEX-MEPE interactions revealed that MEPE binds to PHEX in a nonproteolytic manner48, 49 and, in doing so, protects MEPE from proteolytic cleavage by cathepsin B. PHEX activity thus represents a regulatory mechanism to limit and prevent inhibition of mineralization by ASARM peptides in vivo. Our data confirm this hypothesis and extend it to other SIBLING proteins and their peptides by demonstrating that OPN ASARM is also a substrate for PHEX. In our previous study on the function and degradation of MEPE ASARM, digestion by PHEX occurred immediately at the N-terminal side of aspartate or glutamate residues,14 and in this paper we likewise observed cleavage on the N-terminal side of aspartate residues in OPN ASARM. Additional significance deriving from these results comes from the observation that MEPE ASARM peptide is particularly resistant to proteases12 and that PHEX belongs to a family of zinc metalloendopeptidases whose endogenous substrates are usually small peptides of less than 3 kDa.50

In summary, we provide new data on biomineralization describing a potential molecular basis for peptide and specific amino acid interactions with hydroxyapatite in the context of a family of known mineral-regulating noncollagenous proteins. Our in vitro observations on the function of OPN ASARM peptide, when combined with computational molecular modeling simulations, indicate that phosphorylated OPN ASARM peptides bind to hydroxyapatite crystals and inhibit extracellular matrix mineralization in osteoblast cultures in a manner that depends on phosphorylation of serine residues. Moreover, this inhibition is regulated by proteolytic digestion of OPN ASARM by PHEX. Further investigation is needed to determine potential mineralization-inhibiting contributions from ASARM peptides derived from other SIBLING proteins and whether PHEX and other enzymes likewise modulate their activity. Collectively, such information could be used beneficially for the therapeutic management of mineralization disorders.

Disclosures

MDM is a consultant for and receives research funding from Enobia Pharma. All the other authors state that they have no conflicts of interest.

Acknowledgements

We thank Drs P Crine, T Loisel, D Eanes, and B Fowler for contributing reagents to this study and Dr P Rowe for insightful discussions during the course of this work. These studies were funded by a grant from the Canadian Institutes of Health Research (MT11360 to MDM) and by the Arnold and Mabel Beckman Foundation. MDM is a member of the McGill Centre for Bone and Periodontal Research and the Jamson T. N. Wong Laboratories for calcified tissue research. WNA was partially funded by a studentship from the CIHR Strategic Training Program in Skeletal Health Research.