Osteopontin binds multiple calcium ions with high affinity and independently of phosphorylation status

Bone 66 (2014) 90–95

Contents lists available at ScienceDirect

Bone journal homepage: www.elsevier.com/locate/bone

Rapid Communication

Osteopontin binds multiple calcium ions with high affinity and independently of phosphorylation status Eva Kläning a,c, Brian Christensen a, Esben S. Sørensen a, Thomas Vorup-Jensen c, Jan K. Jensen a,b,⁎ a b c

Department of Molecular Biology and Genetics, Aarhus University, Denmark Danish-Chinese Centre for Proteases and Cancer, Aarhus University, Denmark Department of Biomedicine, Aarhus University, Denmark

a r t i c l e

i n f o

Article history: Received 20 January 2014 Revised 15 May 2014 Accepted 27 May 2014 Available online 10 June 2014 Edited by Mark Johnson Keywords: Calcium EDTA Intrinsically disordered Magnesium Metal affinity Milk protein

a b s t r a c t Osteopontin (OPN) is an acidic, intrinsically disordered extracellular matrix protein with a capacity to modulate biomineralization in vitro and in vivo. The role of posttranslational modification of osteopontin has been intensively studied. Phosphorylation of OPN has been demonstrated to play a role in inhibition of biomineral formation and growth in vitro. Here, we used isothermal titration calorimetry (ITC) to investigate the ability of OPN to bind the divalent cations Ca2+ and Mg2+, both essential components of inorganic minerals in vivo. We found, that bovine OPN binds ~10 Ca2+ ions with an apparent affinity ~50-fold tighter than Mg2+, both regardless of OPN phosphorylation, and with affinities significantly stronger than previously reported. These results were confirmed using human derived OPN. This implies that a majority of the acidic residues within OPN must be engaged in calcium interaction under physiological conditions. © 2014 Elsevier Inc. All rights reserved.

Introduction The formation and deposition of biominerals is of great importance in normal tissue as well as under pathological conditions. Osteopontin (OPN) has been demonstrated to be involved in both scenarios. OPN is an acidic intrinsically disordered extracellular glycoprotein found in many human tissues and body fluids including bone, skin, urine, milk and blood [1,2]. In the normal kidney, OPN is secreted by renal epithelial cells after it has been produced by renal tubular cells [3]. OPN is present at a concentration of approximately 0.1 μM in normal human urine [4]. In kidney stone formers, the expression of OPN is upregulated and OPN has been identified as an integral part of kidney stones [5]. OPN levels have also been implicated in the risk of calcification of renal allografts [6]. Calcium oxalate is a major constituent of kidney stones [7] and phosphorylated OPN is a more potent inhibitor of calcium oxalate monohydrate crystal growth in vitro than the non-phosphorylated counterpart [8,9]. During the in vitro formation of hydroxyapatite, the major component of human bone [10], phosphorylated OPN can play a modulatory Abbreviations: OPN, osteopontin; dOPN, dephosphorylated osteopontin; ITC, isothermal titration calorimetry. ⁎ Corresponding author at: Gustav Wieds Vej 10C, 8000 Aarhus C, Denmark. E-mail address: [email protected] (J.K. Jensen).

http://dx.doi.org/10.1016/j.bone.2014.05.020 8756-3282/© 2014 Elsevier Inc. All rights reserved.

role [11]. Evidence suggests that OPN interacts with both calcium oxalate monohydrate and hydroxyapatite crystals in ways involving negatively charged residues coordinating divalent cations on the surface of the crystals [12]. The peptide chain of OPN is highly negatively charged with one-fourth of residues being acidic. In addition, OPN is varyingly phosphorylated, with the highest level identified in milk-derived OPN of both human and bovine origin [13,14]. In humans, OPN is phosphorylated by the secreted kinase Fam20C [15]. It remains unclear how the addition of phosphates to the already highly acidic OPN can influence biomineral formation [8,16]. OPN could fulfill the Flexible Polyelectrolyte Hypothesis, proposed in [12], stating that the overall density of negative charges and protein flexibility may be determining factors in protein biomineral interactions. Ca2+ and Mg2+ ions have important physiological functions including cellular signaling, as enzyme cofactors and as modulators of biomineral properties by incorporation into hydroxyapatite of bone. In body fluids, both ions are present as free ions or as protein-bound forms and are found in concentrations of 2.2–2.6 mM and 0.65–1.05 mM for Ca2 + and Mg2 +, respectively [17]. In two pioneering studies, membrane-immobilized OPN was shown to bind Ca2+ in competition with Mg2 +, via an aspartate-rich N-terminal domain of OPN [18,19]. Moreover, both cations bound a synthetic OPN fragment, indicating that phosphorylation is irrelevant for binding [18]. An average affinity (KD) of 1 mM was reported for binding of Ca2 + to rat OPN [19]. A

E. Kläning et al. / Bone 66 (2014) 90–95

more recent isothermal titration calorimetric (ITC) study suggested that bovine OPN binds Ca2+ with an even weaker affinity (KD ~2.8 mM) [20]. In the present study we designed a method to obtain the apo-form (cation free) of native bovine and human OPN. Using ITC, we show that more than 10 Ca2+-binding sites, with an average affinity orders of magnitude higher than previously reported, reside in OPN. We provide strong evidence that phosphorylation is irrelevant for the divalent cation binding. We conclude that OPN in vivo is always saturated with Ca2 +. Our observations are not only important for understanding of physiological functions of OPN, but also represents a fundamental knowledge when studying OPN in vitro.

91

titration curves were observed (Figs. 1A and B) and (Figs. 2A and B). These results clearly confirm previous reports using membrane-coated OPN, stating that phosphorylation has no effect on Ca2 + and Mg2 + binding [18,19]. OPN prepared from native source without EDTAtreatment yielded no signal (Fig. 1C). This illustrates that OPN generated by a standard purification procedure remains saturated with divalent cations. As a control, CaCl2 titration into the buffer from the final dialysis step did not yield a signal (Fig. 1D), confirming that the EDTA used in the initial dialysis steps was effectively removed by the final dialysis step. OPN binds multiple Ca2+ ions with high affinity

Methods and materials Purification and dephosphorylation of bovine OPN and preparation of human OPN OPN from bovine milk was purified as described in [21] (for further information see the Supplementary data (SD)). After purification, OPN still contained various cleavage fragments, amounting to approximately 40% of the total protein, as evaluated by SDS–PAGE (data shown in SD). Dephosphorylated OPN (dOPN) was prepared using bovine alkaline phosphatase (Sigma) (30 milliunits/μg OPN) in 10 mM NH4HCO3 pH 8.5 overnight at 37 °C. Complete dephosphorylation was confirmed by mass spectrometry (data not shown). Unfragmented human OPN (see SDS–PAGE in SD) was used for supporting experiments and was purified as described in [13]. Apo-forms of bovine OPN and dOPN as well as human OPN were generated under calcium free environment by extensive EDTA dialysis as described in SD. Buffer from the final dialysis step was used to prepare the metal solutions. Calcium and magnesium concentrations were below the detectable concentration of 10 nM in our ITC apo-OPN stock and milliQ water, as determined by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES, Spectro Analytical Instrument GmbH) analysis on an instrument located at Department of Chemistry, Aarhus University (See SD for further information). ITC of OPN and dOPN ITC experiments were performed on a VP-ITC instrument (Microcal, Northampton, MA, USA) in 20 mM HEPES pH 7.5, 150 mM NaCl at 25 °C. Titrations were performed by stepwise 5 μl injections of buffer containing 1.5 mM MgCl2 or CaCl2 into the ~1.4 ml sample cell. The molar binding enthalpies (ΔH) were obtained by integration of the injection peak areas using the manufacturer's software. Average association constants (KA = K−1 D ) for binding were estimated by fit to a 1:1 binding isotherm of the integrated values as a function of metal:OPN ratio. OPN and dOPN concentrations of 13.4 μM and 14.8 μM were used, respectively. Due to limitation in human OPN preparation, a concentration of 1.6 μM was used in the ITC. Protein concentration was determined by dry-weight, OD280 (ε = 22,910, 84,080 M−1 cm−1, human and bovine respectively) theoretical extinction coefficients were predicted using the ProtParam tool (http://web.expasy.org/protparam/) and BCA quantification (Pierce), (33,500 g/mol for bovine OPN and 45,000 g/mol for human OPN). Results and discussion OPN phosphorylation does not influence binding of divalent cations To generate OPN starting material (apo-forms) completely free of divalent cations, we designed an extensive EDTA-dialysis routine (see SD). Apo-forms of bovine OPN and dOPN were titrated with CaCl2 and MgCl2, the prevalent physiological relevant cations, in separate ITC experiments and for both metal ions, saturable and indistinguishable

To estimate the average binding parameters, a One Set of Sites model (Origin software) was used to fit the integrated ITC data (Figs. 1E and F) and (Figs. 2C and D). The resulting molar enthalpies (ΔH), affinity constants (KD) and the stoichiometries (N) are reported in Table 1. Assuming that all sites are equivalent, OPN and dOPN interacts with Ca2+ with an average KD of ~30–50 nM and a stoichiometry (Ca2+:OPN) of 9–12. In the case of Mg2 +, an average KD of ~ 2 μM was estimated and a stoichiometry of 12–13 (Table 2). Both KD values must be considered as estimates due to the high number of binding sites, however, it can be directly observed that OPN is completely saturated at ten-fold surplus of Ca2 + in the ITC experiment (~ 130 μM, injection number ~ 30), demonstrating that an average KD must be N100-fold below this concentration (Figs. 1E and F). Importantly, our ITC data show that Mg2 + and Ca 2 + must bind with markedly different affinities (Figs. 1 and 2). At a near-constant OPN concentration, the steepness of the curves approaching saturation illustrate the difference in the overall affinity for Mg2 + and Ca2 + according to the Wiseman value [22] also termed the c-value (c = [OPN] ∗ KA) as described in the ITC protocol (Microcal). In fact, given the apparent large c-value (~500) in the case of Ca2+, the N value can be extrapolated from the molar ratio at which saturation is complete. Hence, based on estimated c-values, Ca2+ affinity is likely to be ~50-fold stronger than for Mg2+ (c ~ 10). Unlike the previously reported mM affinities for Ca2+ [19,20], we here report much stronger binding with KD values in the mid-nanomolar range. Using standard purification procedures, the previously reported low affinities must stem from cation binding to additional low-affinity sites. Distinctive differences in the thermodynamic parameters for Ca2+ and Mg2 +-binding suggest a similarity between the metal binding sites of OPN and EDTA OPN titration with Mg2+ results in strong endothermic (positive injection peaks) signals as opposed to the strong exothermic (negative injection peaks) signals observed for the Ca2+ titration (Figs. 1 and 2). Interestingly, this observation and the overall binding parameters and estimated affinities reported here correlate well with what reported for EDTA. Here, a KD of 22.3 nM and 1.7 μM was reported for EDTA interaction with Ca2 + and Mg2+, respectively [23]. These values are very compatible with the approximate average KD-values, 30–50 nM and 2 μM, for OPN interaction with Ca2+ and Mg2+ we report. Additionally, the ΔH-values reported for EDTA, −6.0 kcal/mol and 4.4 kcal/mol for the interaction with Ca2+ and Mg2+ [23], respectively, are also similar to those reported here. In EDTA, four carboxyl groups and two amines chelate the divalent cation (Fig. 3A) [24]. These observations and the fact the fact that OPN is flexible and contain ~24% acidic residues, suggests that OPN may interact with divalent cations in a manner similar to EDTA. Another calcium-binding protein, calsequestrin, has been demonstrated to bind multiple Ca2+ ions via an intrinsically disordered, acidic stretch located in the C-terminal part of the protein. However, the affinities reported for the Ca2+ interaction with the C-terminal part of calsequestrin are markedly lower than those reported here for OPN [25]. Sites with higher affinity were also reported for calsequestrin,

92

E. Kläning et al. / Bone 66 (2014) 90–95

Fig. 1. ITC analysis of Ca2+ binding to OPN. OPN (A) and dOPN (B) were titrated with CaCl2. The raw data represent the differential power (μcal/s) as a function of time (s), n = 3. (C) OPN without any prior dialysis against EDTA, titrated with CaCl2, n = 3. (D) Buffer from the final EDTA dialysis step titrated with CaCl2, n = 1. The integrated data (E and F) represent the enthalpy change per mole of injectant, ΔH, in units of (kcal/mol) as a function of the molar ratio. Data points and fitted data are overlaid, n = 3.

although the actual affinities remain unknown (saturated at lowest used concentration in titration experiments), however these sites were located in structured parts of the protein. We are therefore convinced that different mechanisms govern the calcium-binding properties of the two proteins. Our observation with OPN suggests that in order to have multiple sites with high affinity for calcium in an unstructured protein, acidic residues need to be clustered in patches of 5–8 residues across the linear sequence in regions with a truly flexible polypeptide main chain (Fig. 3B). The relative binding affinity of individual sites may be further modulated by the actual spacing between Asp and Glu within the patch and the surrounding sequence motifs. The disordered C-terminal of calsequestrin is different almost entirely consisting of acidic amino acids, no patches or spacers, which may be disruptive of EDTA-like chelation resulting in somewhat lower affinity. Our data show that OPN binds ~9–12 Ca2+ and ~12–13 Mg2+ per molecule. Bovine OPN contains 62 acidic residues, and as 3–4 acidic residues would most likely be needed to chelate a divalent cation as seen for EDTA, more than ten hypothetical metal sites can be postulated as illustrated in Fig. 3B. Possibly, water molecules, protein backbone carbonyls or other amino acid side chains could form the remainder of the coordination bonds, as is seen for the calcium-binding EF-hand proteins [26]. Our stoichiometry for Ca2+-binding correlates well with the reported ~8 sites of human milk OPN [18].

OPN is known to engage biomineral surfaces and various binding partners, e.g., a range of integrins in the human body [12,1], and it is therefore crucial to understand the nature of the native form of OPN when investigating these interactions. As the affinity for Ca2+ described here is far stronger than previously known, and given the accessibility of Ca2 + in the extracellular environment [17], we expect that OPN will always be fully saturated with calcium. A likely direct consequence of calcium binding, will be shielding of the numerous negative charges in the small protein, allowing for condensation of the molecule due to minimized internal electrostatic repulsion. This again, may shield OPN from premature or unwanted adsorption to extracellular matrix components or binding to other proteins due to unspecific ionic interactions. Although highly speculative, the binding of OPN to surface exposed calcium ions on biomineral surfaces could happen by a “zipper-like” manner, by which OPN-bound calcium ions will be sequentially replaced by the mineral surface calcium causing an “unwinding” of OPN onto the mineral surface. During this process, phosphorylation and glycosylation status of OPN may play important regulatory roles. We also speculate that local increases in Ca2+ may play a direct role in the local binding equilibrium between OPN and biomineral surfaces by competing for the acidic residues. Our observations are important for understanding the influence of OPN phosphorylation, as this type of posttranslational modification could provide the primary accessible negative charges in

E. Kläning et al. / Bone 66 (2014) 90–95

93

Fig. 2. ITC analysis of Mg2+ binding to OPN. OPN (A and C) and dOPN (B and D) were titrated with MgCl2. The raw data (A and B) represent the differential power (μcal/s) as a function of time (s). The integrated data (C and D) represent the enthalpy change per mole of injectant, ΔH, in units of (kcal/mol of injectant) as a function of the molar ratio. Data points and fitted data are overlaid, n = 2.

Ca2 +-saturated OPN. The model we propose does not exclude that calcium-bound OPN cannot still behave as a flexible, elongated protein. However, these hypotheses will have to be investigated further. Bovine OPN is a suitable substitute for human derived OPN Bovine OPN is widely used as a substitute for human OPN, due to the more ready supply as well as a high degree of similarity in both sequence and posttranslational modification [13]. To verify and generalize our results reporting a high affinity for Ca2+ observed for bovine OPN, we decided to test calcium binding to highly purified, unfragmented human milk OPN in the ITC (Fig. 4) (See SD for SDS–PAGE). The results confirm that as observed for bovine OPN, human milk OPN binds Ca2+ with high affinity, and with similar strong exothermic signal which reaches full saturation at ~ 90 μM (Fig. 4). In separate experiments, a strong endothermic signal for human OPN interaction with Mg2 + was observed (data not shown), further illustrating the similarity between the data obtained for human and bovine OPN. The number of binding-sites within human OPN appears to be increased as compared to that of bovine OPN, as indicated by the molar ratio at which saturation is reached (Fig. 4C). We suggest that this may in part be due to the presence of 13 additional acidic residues in human OPN, but may also be contributed by mechanisms not fully explained by our

Table 1 ITC Ca2+-binding parameters.

simple linear model proposed for bovine OPN in Fig. 3B. This is not in complete agreement with the results describing ~8 Ca2+ binding sites in human milk OPN [18], but most likely due to the variations in the techniques applied. Furthermore, we cannot exclude that the OPN glycosylations can contribute to Ca2 + binding. As evaluated by the SDS–PAGE (Supplementary figure), the majority of the bovine OPN is full-length protein, although some OPN is represented by distinct fragments of rather large sizes. Given the size of the fragments and according to our hypothesis of chelating patches in OPN, we believe that, on average not more than a single binding site in the subfraction of fragmented protein will be affected, allowing us to draw overall conclusions although we may underestimate the total number of binding sites in bovine OPN. We must also stress, that the estimation of the actual number of binding-sites becomes less and less precise with increasing degeneracy of the data resulting from the increasing number of binding-sites. Based on the increased size of human relative to bovine OPN, resulting in more clusters of acidic residues, in combination with the above mentioned criteria for Ca2+ coordination (which are rather restrictive), we postulated that a realistic number of binding sites is in the range of 15–19 sites, a number compatible with our ITC results suggesting at least 20 sites. In summary, our data support the use of bovine OPN as a valid substitute for the human protein with respect to metalion binding.

Table 2 ITC Mg2+-binding parameters.

Ca2+-binding parameters (mean ± SD, n = 3) Sample

ΔH (kcal/mol)

KD (1/KA) (nM)

N

OPN dOPN

−7.3 ± 0.15 −7.4 ± 0.17

35 ± 11 38 ± 11

10.6 ± 1.4 10.2 ± 0.9

Mg2+-binding parameters (mean ± SD, n = 4) Sample

ΔH (kcal/mol)

KD (1/KA) (μM)

N

OPN/dOPN

4.3 ± 0

2.1 ± 0.3

12.9 ± 0.8

94

E. Kläning et al. / Bone 66 (2014) 90–95

Fig. 3. Representation of Ca2+ binding EDTA and OPN. (A) Schematic model of EDTA chelating Ca2+, drawn in ChemSketch (Advanced Chemistry Development, Inc., Toronto, On, Canada). (B) Sequence of bovine OPN illustrating the 28 phosphorylation sites (blue circles) and three o-glycosylation sites (green circles) [14]. Acidic residues are highlighted in yellow. Proposed Ca2+-binding sites are indicated by black lines connecting acidic residues to calcium ions (red circles). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Conclusions

Author contributions

We have demonstrated that OPN phosphorylation does not affect the Mg2+ and Ca2+ binding capacity of OPN in solution. Furthermore, we show that OPN interacts with both ions with previously unknown high affinity and through ~10 sites which share high similarity to that of EDTA. The biological function of soluble OPN binding of Ca2+ is not clear. We have identified that OPN binds Ca2+ with a high affinity and that the Ca2+-saturated form of the protein is therefore the actual native form of OPN. We propose that as divalent cations will occupy the majority of the acidic residues within OPN, the phosphate groups are left free to engage the surface of a biomineral crystal. On the other hand, binding to calcium embedded in biominerals may be a competition with the bound calcium ions for the metal sites in OPN. This may help explain why the phosphorylated OPN isoforms are more potent inhibitors of biomineral formation and growth in vitro than the nonphosphorylated counterparts [8,16] as the phosphorylations may contribute independently or regulate binding to biominerals. Indeed, this is a relevant scenario, as the physiological concentrations of Ca2+ and Mg2 + would render OPN fully saturated with Ca2 +, considering the overall KD-values of 30–50 nM for Ca2+ and 2 μM for Mg2+ reported in this study. Our results on human OPN verify a similarity in behavior of the protein between to mammalian species, again supporting a common evolutionary conserved function of OPN.

E.K. prepared dOPN and apo-forms, performed all experiment and wrote the manuscript. B.C. performed mass spectroscopy. E.S.S. provided native OPN and funding of E.K. T.V.-J. initiated the project and commented on the manuscript. J.K.J. designed experiments and wrote the manuscript. All authors have given approval to the final version of the manuscript. Acknowledgments This work was supported by Arla Foods Ingredients (E.K.), The Danish Council for Independent Research (E.S.S.), The Lundbeck Foundation Nanomedicine Center for Individualized Management of Tissue Damage and Regeneration (LUNA) (T.V.-J.) and the Danish Cancer Society (R56-A2997-12-S2) and the Danish Natural Research Foundation (26331-6) (J.K.J.). We would like to thank the Danish National Research Foundation (DNRF93) Center for Materials Crystallography for the help with the ICP-OES analysis. Especially, we are grateful for the competent ICP assistance provided by Peter Hald. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.bone.2014.05.020.

Fig. 4. ITC analysis of Ca2+ binding to human milk OPN. The raw data (A) represent the differential power (μcal/s) as a function of time (s) and the integrated data (C) represent the enthalpy change per mole of injectant, ΔH, in units of (kcal/mol of injectant) as a function of the molar ratio for human OPN (1.6 μM ~10 fold lower concentration than for bovine OPN ITC experiments) titration with CaCl2, n = 3. CaCl2 titration into the buffer from the final dialysis step did not yield a signal (B), confirming that the EDTA used in the initial dialysis steps was effectively removed by the final dialysis steps, n = 1.

E. Kläning et al. / Bone 66 (2014) 90–95

References [1] Sodek J, Ganss B, McKee MD. Osteopontin. Crit Rev Oral Biol Med 2000;11(3):279–303. [2] Platzer G, Schedlbauer A, Chemelli A, Ozdowy P, Coudevylle N, Auer R, et al. The metastasis-associated extracellular matrix protein osteopontin forms transient structure in ligand interaction sites. Biochemistry 2011;50(27):6113–24. [3] Kohri K, Yasui T, Okada A, Hirose M, Hamamoto S, Fujii Y, et al. Biomolecular mechanism of urinary stone formation involving osteopontin. Urol Res 2012;40(6):623–37. [4] Asplin JR, Arsenault D, Parks JH, Coe FL, Hoyer JR. Contribution of human uropontin to inhibition of calcium oxalate crystallization. Kidney Int 1998;53(1):194–9. [5] Kohri K, Suzuki Y, Yoshida K, Yamamoto K, Amasaki N, Yamate T, et al. Molecular cloning and sequencing of cDNA encoding urinary stone protein, which is identical to osteopontin. Biochem Biophys Res Commun 1992;184(2):859–64. [6] Lorenzen JM, Martino F, Scheffner I, Brocker V, Leitolf H, Haller H, et al. Fetuin, matrix-Gla protein and osteopontin in calcification of renal allografts. PLoS ONE 2012;7(12):e52039. [7] Bushinsky DA. Kidney stones. Adv Intern Med 2001;47:219–38. [8] Hunter GK, Grohe B, Jeffrey S, O'Young J, Sorensen ES, Goldberg HA. Role of phosphate groups in inhibition of calcium oxalate crystal growth by osteopontin. Cells Tissues Organs 2009;189(1–4):44–50. [9] O'Young J, Chirico S, Al Tarhuni N, Grohe B, Karttunen M, Goldberg HA, et al. Phosphorylation of osteopontin peptides mediates adsorption to and incorporation into calcium oxalate crystals. Cells Tissues Organs 2009;189(1–4):51–5. [10] Lowenstam HA, Weiner S. On Biomineralization. Oxford: Oxford University Press; 1989. [11] de Bruyn JR, Goiko M, Mozaffari M, Bator D, Dauphinee RL, Liao Y, et al. Dynamic light scattering study of inhibition of nucleation and growth of hydroxyapatite crystals by osteopontin. PLoS ONE 2013;8(2):e56764. [12] Hunter GK, O'Young J, Grohe B, Karttunen M, Goldberg HA. The flexible polyelectrolyte hypothesis of protein–biomineral interaction. Langmuir 2010;26(24):18639–46. [13] Christensen B, Nielsen MS, Haselmann KF, Petersen TE, Sorensen ES. Posttranslationally modified residues of native human osteopontin are located in clusters: identification of 36 phosphorylation and five O-glycosylation sites and their biological implications. Biochem J 2005;390(Pt 1):285–92.

95

[14] Sorensen ES, Hojrup P, Petersen TE. Posttranslational modifications of bovine osteopontin: identification of twenty-eight phosphorylation and three O-glycosylation sites. Protein Sci 1995;4(10):2040–9. [15] Tagliabracci VS, Engel JL, Wen J, Wiley SE, Worby CA, Kinch LN, et al. Secreted kinase phosphorylates extracellular proteins that regulate biomineralization. Science 2012;336(6085):1150–3. [16] Boskey AL, Christensen B, Taleb H, Sorensen ES. Post-translational modification of osteopontin: effects on in vitro hydroxyapatite formation and growth. Biochem Biophys Res Commun 2012;419(2):333–8. [17] Jahnen-Dechent W, Ketteler M. Magnesium basics. Clin Kidney J 2012;5(Suppl. 1): i3–i14. [18] Singh K, Deonarine D, Shanmugam V, Senger DR, Mukherjee AB, Chang PL, et al. Calcium-binding properties of osteopontin derived from non-osteogenic sources. J Biochem 1993;114(5):702–7. [19] Chen Y, Bal BS, Gorski JP. Calcium and collagen binding properties of osteopontin, bone sialoprotein, and bone acidic glycoprotein-75 from bone. J Biol Chem 1992;267(34):24871–8. [20] Yamniuk AP, Burling H, Vogel HJ. Thermodynamic characterization of the interactions between the immunoregulatory proteins osteopontin and lactoferrin. Mol Immunol 2009;46(11–12):2395–402. [21] Sorensen ES, Petersen TE. Purification and characterization of three proteins isolated from the proteose peptone fraction of bovine milk. J Dairy Res 1993;60(2):189–97. [22] Wiseman T, Williston S, Brandts JF, Lin LN. Rapid measurement of binding constants and heats of binding using a new titration calorimeter. Anal Biochem 1989;179(1):131–7. [23] Henzl MT, Larson JD, Agah S. Estimation of parvalbumin Ca(2+)- and Mg(2 +)binding constants by global least-squares analysis of isothermal titration calorimetry data. Anal Biochem 2003;319(2):216–33. [24] Skoog DE, West DM, Holler J. Fundamentals of Analytical Chemistry. Harcourt Brace College Publishers; 1995. [25] Sanchez EJ, Lewis KM, Danna BR, Kang C. High-capacity Ca2+ binding of human skeletal calsequestrin. J Biol Chem 2012 Mar 30;287(14):11592–601. [26] Biekofsky RR, Martin SR, Browne JP, Bayley PM, Feeney J. Ca2+ coordination to backbone carbonyl oxygen atoms in calmodulin and other EF-hand proteins: 15N chemical shifts as probes for monitoring individual-site Ca2+ coordination. Biochemistry 1998;37(20):7617–29.