Research article

Post-translational modification and proteolytic processing of urinary osteopontin

Brian Christensen, Torben E. Petersen, Esben S. Sørensen

Abstract

OPN (osteopontin) is a highly phosphorylated glycoprotein present in many tissues and body fluids. In urine, OPN is a potent inhibitor of nucleation, growth and aggregation of calcium oxalate crystals, suggesting that it has a role in the prevention of renal stone formation. The role of OPN in nephrolithiasis is, however, somewhat unclear, as it may also be involved in urinary stone formation, and it has been identified among the major protein components of renal calculi. Most likely, the function of OPN in urine is dependent on the highly anionic character of the protein. Besides a very high content of aspartic and glutamic residues, OPN is subjected to significant PTM (post-translational modification), such as phosphorylation, sulfation and glycosylation, which may function as regulatory switches in promotion or inhibition of mineralization. In the present study, we have characterized the PTMs of intact human urinary OPN and N-terminal fragments thereof. MS analysis showed a mass of 37.7 kDa for the intact protein. Enzymatic dephosphorylation and peptide mass analyses demonstrated that the protein contains approximately eight phosphate groups distributed over 30 potential phosphorylation sites. In addition, one sulfated tyrosine and five O-linked glycosylations were identified in OPN, whereas no N-linked glycans were detected. Peptide mapping and immunoblotting using different monoclonal antibodies showed that the N-terminal fragments present in urine are generated by proteolytic cleavage at Arg228–Leu229 and Tyr230–Lys231.

  • O-glycosylation
  • osteopontin
  • phosphorylation
  • proteolytic processing
  • sulfation
  • uropontin

INTRODUCTION

OPN (osteopontin) is a phosphorylated glycoprotein containing an integrin-binding RGD (Arg-Glu-Asp) sequence. The protein was originally purified from the mineralized matrix of bovine bone [1]; however, the expression of OPN is not limited to mineralized tissues, but extends to most of the tissues and body fluids, including milk, blood and urine [2]. OPN consists of approx. 300 amino acid residues with a very high content of aspartic and glutamic residues. The protein is involved in a broad range of functions in both physiological and pathological processes such as cytokine production, tumour growth, inflammation and mineralization [35]. Furthermore, OPN is a key molecule in bone remodelling [5,6] and it hinders ectopic calcification by inhibiting the formation of hydroxyapatite [7,8], which is the main mineral phase of bone and teeth, as well as calcium oxalate [9], the main mineral in renal calculi.

OPN is found at high levels in urine with an average of approx. 4 mg secreted into human urine daily [10]. In contrast, OPN is undetectable in normal kidney tissues [11], indicating that the OPN produced in the renal tubules is immediately secreted in the urine. OPN can inhibit nucleation, growth and aggregation of calcium oxalate crystals in vitro [9,12,13], and it blocks binding of the crystals to renal epithelial cells [14]. In addition, COM (calcium oxalate monohydrate) crystal growth and aggregation are inhibited by 50% at an OPN concentration between 16 and 28 nM, a concentration that is an order of magnitude lower than the OPN concentration of 131 nM found in normal urine [13]. OPN also favours the formation of COD (calcium oxalate dihydrate) over COM crystals [12]. The induction of COD by OPN may be protective against calcium oxalate nephrolithiasis, since COD has demonstrated less affinity for attachment to renal epithelial cells than COM [12]. Besides reducing crystal affinity for renal cells, OPN may act as an intracrystalline protein assisting stone prevention by rendering crystals more susceptible to intracellular degradation [15]. Furthermore, OPN-knockout mice fed with a crystal-inducing diet developed renal calcium oxalate stones, whereas wild-type mice were completely unaffected, perhaps due to an up-regulated OPN expression in response to the induced hyperoxaluria, suggesting a protective role of OPN [16]. Similarly, a recent study showed that 10% of mice lacking OPN spontaneously formed renal stones and 65% developed renal crystals under experimentally induced hyperoxaluria, whereas these events were not observed in wild-type mice [17].

There are many indications that OPN is an important inhibitor of renal stone formation; however, the protein has also been implicated as a promoter of stone formation. One study showed increased calcium oxalate crystal adherence to kidney cells when they were incubated with OPN [18]. Furthermore, inhibition of OPN synthesis has been shown to suppress the deposition and adhesion of calcium oxalate crystals to rat renal cells [19], and addition of polyclonal OPN antibodies reduced the deposition of crystals on the surface of kidney cells [20]. It has also been suggested that OPN is involved in urinary stone formation [21], and the protein has been identified as a prominent constituent of the major matrix of renal calculi [22]. Reduced OPN levels in stone formers compared with non-stone formers have been reported in one study [23], while no difference in urinary OPN levels has been reported in other studies [24,25].

OPN is extensively altered through PTM (post-translational modification), such as phosphorylation, sulfation, glycosylation and proteolytic processing, which have significant implications on the structure and the biological function of the protein [26]. The PTMs of OPN from bovine and human milk as well as rat bone have been thoroughly characterized [2729]. Recently, a comparative study of OPN produced by murine ras-transformed fibroblasts and differentiating osteoblasts demonstrated remarkably different phosphorylation patterns between OPNs originating from the same species, which translated into significant functional differences in cellular adhesion [30].

The PTMs of OPN provide a basis for its regulatory functions in mineralization processes, as dephosphorylated OPN has been shown to lose the ability to inhibit hydroxyapatite formation [7,8]. Emphasizing the importance of the phosphorylations, it was recently shown that highly phosphorylated milk OPN promoted hydroxyapatite formation and growth, whereas the less phosphorylated bone OPN inhibited these processes [31]. Similarly, recombinant (non-phosphorylated) OPN increases mineralization of vascular smooth-muscle cells, whereas enzymatically phosphorylated OPN is a potent inhibitor of this process [32]. Recently, it was shown that only phosphorylated OPN dose-dependently inhibited mineralization of MC3T3-E1 osteoblast cell cultures [33]. Phosphorylation of OPN peptides also markedly enhanced the inhibition of hydroxyapatite and COM crystal growth [34,35], and C-terminal fragments of urinary OPN can inhibit COM growth as efficiently as the intact protein, suggesting that different parts of the protein can regulate crystal formation [36].

Western blotting of human urinary OPN has shown the presence of different forms in the 40–70 kDa range [15,24,25]. The structural variation underlying these differences in migration is unknown. Normal rat kidney cells have been shown to produce both phosphorylated and non-phosphorylated OPN, which also diverge in glycosylation and physiological properties [37]. Furthermore, low-molecular-mass OPN variants probably resulting from cleavage by serine proteases are present at a higher frequency in the urine from individuals with kidney stone disease than in the urine from healthy individuals [24]. Recently, a series of C-terminal OPN fragments were identified by two-dimensional electrophoresis and MS in patients with ovarian cancer, but not in healthy controls [38]. In the present study, we have separated and characterized different forms of human urinary OPN with regard to PTMs and proteolytic processing. These modifications are likely to be key factors in the regulatory roles of OPN in renal stone formation.

EXPERIMENTAL

Materials

The monoclonal mouse antibodies 1H3 and 3D9 [39] were a gift from Dr David T. Denhardt (Rutgers University). Swine anti-rabbit and rabbit anti-mouse immunoglobulins conjugated to alkaline phosphatase were obtained from Dako (Glostrup, Denmark). Bovine pancreas trypsin and staphylococcal (Staphylococcus aureus) V8 protease were obtained from Worthington Biochemical (Freehold, NJ, U.S.A.). Thermolysin, ALP (bovine alkaline phosphatase), human plasma thrombin and arylsulfatase were from Sigma (St. Louis, MO, U.S.A.). The μRPC (μ reverse-phase chromatography) C2/C18 PC 2.1/10 column, Protein A–Sepharose CL-4B resin, DEAE-Sepharose material and PD-10 desalting columns were from GE Healthcare (Uppsala, Sweden). Vydac C4 and C18 reverse-phase resins were obtained from The Separations Group (Hesperia, CA, U.S.A.). Reagents used for sequencing were purchased from Applied Biosystems (Warrington, U.K.). 2,5-Dihydroxybenzoic acid was from LaserBio Labs (Sophia-Antipolis Cedex, France). The PNGase F (peptide N-glycosidase F) kit was obtained from New England Biolabs (Beverly, MA, U.S.A.). All other chemicals used were of analytical grade.

Polyclonal OPN antibodies

Polyclonal antiserum against human milk OPN was raised in rabbits at Dako. OPN-specific polyclonal antibodies were isolated from the rabbit serum on a Protein A–Sepharose column. The specificity of the antibodies was checked and verified by Western-blot analyses of milk samples and purified OPN.

Western blotting

OPN (50 ng/lane) and urine (10 μl) were loaded on to 16% Tris/Tricine gels, fractionated by SDS/PAGE and electrophoretically transferred on to Hybond-P PVDF membranes (GE Healthcare) for immunodetection. The membranes were blocked in 2% Tween in Tris-buffered saline before addition of either polyclonal rabbit OPN antibodies (5 μg/ml) or monoclonal mouse OPN antibodies 1H3 and 3D9 (1.5 μg/ml). OPN was detected with alkaline phosphatase-conjugated secondary immunoglobulins.

Purification of OPN

Urine samples were collected from seven healthy donors with normal renal function and no history of urinary disease and frozen at −80 °C. Thawed urine samples were centrifuged at 4500 g for 30 min and the resulting supernatant was incubated with DEAE-Sepharose at pH 8.1 for 2 h at room temperature (25 °C). Adsorbed proteins were batch eluted with stepwise increasing concentrations of ammonium bicarbonate (pH 8.1). The batch containing OPN was freeze-dried, followed by passage through a PD-10 column in 20% formic acid. The salt-free eluate was freeze-dried and subjected to further purification by RP-HPLC (reverse-phase HPLC) on a Vydac C4 column connected to a GE Healthcare LKB system. Separation was carried out at 40 °C in 0.1% TFA (trifluoroacetic acid) (buffer A) and the proteins were eluted with a gradient of 80% acetonitrile in 0.1% TFA (buffer B) developed over 50 min (0–5 min, 0% B; 5–40 min, 0–50% B; 40–50 min, 50–95% B) at a flow rate of 0.85 ml/min. The proteins were detected in the effluent by measuring the absorbance at 226 nm. The fraction containing OPN was identified by immunodetection and N-terminal sequencing. The full-length protein (OPN60) and N-terminal fragments (OPN45 and OPN50) were separated by RP-HPLC on a Vydac C18 column and eluted with stepwise increasing concentrations between 10 and 30% of 75% (v/v) propan-2-ol in 0.1% TFA. The fractions were analysed by N-terminal sequencing and SDS/PAGE, and proteins were visualized by CBB (Coomassie Brilliant Blue) staining or by Western blotting. The amount of purified OPN was determined by amino acid analysis.

Phosphate content of OPN

Native and dephosphorylated OPN isoforms were analysed by MS using a Voyager DE-PRO MALDI–TOF (matrix-assisted laser-desorption ionization–time-of-flight) mass spectrometer (Applied Biosystems). Samples for MS analyses were prepared by mixing the sample with a saturated solution of 2,5-dihydroxybenzoic acid in a 1:1 ratio directly on the MS-target probe. All spectra were obtained in positive linear-ion mode using a nitrogen laser at 337 nm and an acceleration voltage of 20 kV. Typically, 50–100 laser shots were added per spectrum and calibrated with external standards. The masses were assigned using the half-height width method. For dephosphorylation, 2.5 μg of OPN was incubated with 0.07 unit of ALP in 10 mM ammonium bicarbonate (pH 8.5) overnight at 37 °C. The average phosphate content estimated by MS was confirmed with a Phosphoprotein Phosphate Estimation Assay kit (Pierce, Rockford, IL, U.S.A.).

Carbohydrate analyses

Full-length OPN (OPN60) was quantified by amino acid analysis and hydrolysed in 2 M TFA for 4 h at 100 °C under oxygen-free conditions. α-L-Rhamnose was added as an internal standard. For quantifying the more stable amino sugars, GalNAc and GlcNAc, an additional OPN sample was hydrolysed in 4 M HCl for 6 h at 110 °C. For sialic acid analysis, the protein was hydrolysed in 0.1 M HCl for 60 min at 80 °C. Monosaccharides were separated by high-pH anion-exchange chromatography using a CarboPac PA1 column (4 mm×250 mm; Dionex, Sunnyvale, CA, U.S.A.) and monitored by pulsed electrochemical detection. The neutral monosaccharides were eluted isocratically with 16 mM NaOH, whereas sialic acid was eluted with 100 mM NaOH and 150 mM sodium acetate. Standard monosaccharide mixtures of α-L-fucose, D-galactosamine, D-glucosamine, D-galactose, D-glucose (all 0.1 mM), D-mannose (0.5 mM), α-L-rhamnose (0.2 mM) and NeuAc (0.02 mM) (all from Sigma) were used. Linearity was observed between the injected amounts of monosaccharide standards and the peak areas, and hence the amounts of the individual monosaccharides in the samples were deduced from standard curves.

Removal of potential N-linked glycosylations was performed as described by the manufacturer using a PNGase F deglycosylation kit. OPN was incubated with 500 units of PNGase F at 37 °C for 20 h. The reaction products from the enzymatic deglycosylation were analysed by SDS/PAGE.

Generation and separation of peptides

OPN was digested with trypsin by using an enzyme-to-substrate ratio of 1:30 (w/w) in 0.1 M ammonium bicarbonate at 37 °C for 6 h. Tryptic peptides were separated by RP-HPLC on a μRPC C2/C18 PC 2.1/10 column connected to a GE Healthcare SMART system. Separation was carried out in 0.1% TFA (buffer A) and eluted with a gradient of 60% acetonitrile in 0.1% TFA (buffer B) developed over 54 min (0–9 min, 0% B; 9–49 min, 0–50% B; 49–54 min, 50–100% B) at a flow rate of 0.15 ml/min. The peptides were detected in the effluent by measuring the absorbance at 214 nm. A fragment (residues Gln36/Ser62–Arg143) from the tryptic digest was further digested with thermolysin in 0.1 M pyridine acetate (pH 6.5) and 5 mM CaCl2 at 56 °C for 6 h. The resulting peptides were separated as described for the tryptic digest. A glycopeptide (residues Leu116–Thr139) from the thermolysin digest was further digested with staphylococcal (S. aureus) V8 protease in 0.1 M ammonium bicarbonate (pH 8.1) at 37 °C for 3.5 h.

OPN45/OPN50 and OPN60 were digested with thrombin (0.05 unit/μg of OPN) in 0.1 M ammonium bicarbonate at 37 °C for 1 h. The generated fragments were separated by RP-HPLC on a Vydac C18 column connected to a GE Healthcare LKB system. Separation was carried out in 0.1% TFA (buffer A) and proteins were eluted with a gradient of 75% propan-2-ol in 0.1% TFA (buffer B) developed over 80 min (0–5 min, 0% B; 5–70 min, 0–50% B; 70–80 min, 50–95% B) at a flow rate of 0.85 ml/min.

Characterization of peptides

Peptides and fragments were characterized by MS and amino acid sequence analyses. All MS spectra were obtained in both positive reflector-ion and positive linear-ion mode as described above. The theoretical peptide masses and sequence coverage were calculated using the GPMAW program (Lighthouse Data, Odense, Denmark). Amino acid sequence analyses were performed on an Applied Biosystems PROCISE HT protein sequencer with online identification of PTH (phenylthiohydantoin)-derivatives. Glycosylated threonine residues were identified by the lack of a PTH derivative in the cycles where these amino acids are modified.

Sulfate analysis

Modified peptides containing both potential phosphorylation (serine and threonine residues) and sulfation (tyrosine residues) sites were dephosphorylated with ALP as described above and analysed by MALDI–TOF-MS. The peptides Phe158–Lys187 and Arg160–Lys187 were further treated with 0.5 unit of arylsulfatase in 0.2 M sodium acetate (pH 5.0) at 37 °C for 2 h and then re-analysed by MS.

RESULTS

Purification of OPN from urine

Urine samples from seven healthy donors, with normal renal function and no history of urinary disease, were analysed by Western blotting to examine the molecular forms of OPN in urine (Figure 1). Urinary OPN was detected with variable intensity in all individuals by a polyclonal antibody. The observed OPN patterns show that all analysed urine samples contain OPN forms of relative homogeneous molecular mass in the range of 45–60 kDa. No low-molecular-mass OPN forms were observed in any of the tested individuals.

Figure 1 Western-blot analysis of human urinary OPN

Urine obtained from seven healthy donors analysed on a 16% Tris/Tricine gel by SDS/PAGE and detected by a polyclonal OPN antibody. Lane 1, molecular mass standards; lanes 2–8, urine samples from seven individuals.

OPN was purified from human urine by anion exchange followed by RP-HPLC separation on a Vydac C4 column (Figure 2). Sequencing of the OPN-containing fraction showed two sequences: a major, I1PVKQA6, and a minor, V3KQADS8, both corresponding to the N-terminal part of human OPN. SDS/PAGE revealed the presence of three bands migrating at approx. 45, 50 and 60 kDa respectively and Western blotting confirmed their OPN nature (Figure 3A, lanes 1 and 2). In the following, these components will be referred to as OPN45, OPN50 and OPN60.

Figure 2 RP-HPLC chromatography of human urine proteins eluted from DEAE material

Separation was performed on a Vydac C4 column in 0.1% TFA and proteins were eluted with a gradient of 80% acetonitrile (dashed line) at a flow rate of 0.85 ml/min. Proteins were monitored at 226 nm (solid line). Peaks containing OPN and a trypsin inhibitor (HI30) are indicated. The inset shows the separation of proteins in the OPN-containing peak on a Vydac C18 column by stepwise increasing concentrations of 75% propan-2-ol.

Figure 3 Analysis of urinary OPN

Samples were separated on 16% Tris/Tricine gels by SDS/PAGE, and proteins were visualized by CBB staining or Western blotting with a polyclonal antibody or the monoclonal antibodies 1H3 or 3D9. (A) Lane 1, CBB staining of the OPN-containing peak (from Figure 2); lane 2, Western-blot detection with a polyclonal OPN antibody of the OPN-containing peak (from Figure 2); lane 3, CBB staining of the peak containing OPN45 and OPN50 (from the inset in Figure 2); lane 4, CBB staining of the peak containing OPN60 (from the inset in Figure 2). (B) Western-blot detection with the monoclonal antibody 1H3 of peaks containing OPN45/OPN50 (lane 1) and OPN60 (lane 2) (from the inset in Figure 2). Western-blot detection with the monoclonal antibody 3D9 of peaks containing OPN45/OPN50 (lane 3) and OPN60 (lane 4) (from the inset in Figure 2).

To separate the different forms, the OPNs were subjected to RP-HPLC on a Vydac C18 column and eluted with stepwise increasing concentrations of propan-2-ol. The elution profile revealed two fractions (inset in Figure 2). As judged by SDS/PAGE, the first fraction contained OPN45 and OPN50 and the second fraction contained OPN60 (Figure 3A, lanes 3 and 4). The monoclonal antibody 1H3 (epitope: S218AETHSH224 [39]) recognized all three OPN forms (Figure 3B, lanes 1 and 2), whereas Western blotting with the monoclonal antibody 3D9 (epitope: K283FRISHELDSASSEVN298 [39]) showed that only OPN60 contains the extreme C-terminal part of OPN (Figure 3B, lanes 3 and 4). This indicates that OPN60 represents the full-length OPN molecule from human urine, since it contains both the N- and C-termini, whereas OPN45 and OPN50 are truncated or cleaved forms lacking a C-terminal part of the protein.

Identification of cleavage sites in OPN

To identify the C-terminus and thereby the exact cleavage site that generates OPN45 and OPN50, these proteins were subjected to thrombin cleavage. RP separation of the products from the thrombin digest followed by sequence and MS analyses showed the presence of two N-terminal fragments, Ile1–Arg152 and Val3–Arg152, and two C-terminal fragments, Ser153–Arg228 and Ser153–Tyr230 (Table 1). This indicates that OPN45 and OPN50 are generated by cleavage at Arg228–Leu229 and/or Tyr230–Lys231 in human urine. In an additional experiment, digests of OPN45/OPN50 and OPN60 were compared by peptide mapping analysis. Peptides from the N-terminal part (Ile1–Leu229/Arg232) were found in all OPN isoforms, whereas peptides from the C-terminal part (Leu229–Asn298) only were identified in OPN60, further supporting that OPN45 and OPN50 are generated by cleavage in the region around Leu229–Arg232.

View this table:
Table 1 Thrombin cleavage of OPN45/OPN50

The fragments were identified by N-terminal sequence and MALDI–TOF-MS analyses. The observed molecular masses were determined by MALDI–TOF-MS in linear mode. Mass difference is that between the observed and calculated average masses. The type of modification corresponding to the mass difference is given in parentheses: phosphorylation (P), sulfation (S) and oxidation of methionine (Mox).

To test whether OPN45/OPN50 could be generated from OPN60 by thrombin cleavage, the full-length protein was incubated with this protease. This resulted in two major fractions in RP-HPLC containing the N-terminal part (Ile1–Arg152 and Val3–Arg152) and the expected C-terminal part of OPN (Ser153–Asn298). Furthermore, we observed a minor fraction representing a fragment corresponding to Ser153–Arg228/Tyr230. Judged by RP-HPLC and MS analyses, this fraction only represents a very minor portion of the cleavage products and, hence, we cannot, on this basis, conclude that thrombin would cleave OPN at this site in vivo.

Molecular masses

The masses of the OPNs were determined by linear MALDI–TOF-MS. MS analysis showed a mass of approx. 37.7 kDa for OPN60 (Figure 4A), whereas the sample containing OPN45 and OPN50 showed one broad mass peak with an average value of approx. 29.3 kDa (Figure 4B). To estimate the total number of phosphate groups present, OPN60 and the fragments OPN45/OPN50 were treated with ALP. The molecular mass of dephosphorylated OPN60 was 37.1 kDa, corresponding to a loss of approximately eight phosphate groups (Figure 4A). Phosphatase treatment of the N-terminal OPN fragments reduced the mass to 28.8 kDa, corresponding to a loss of approximately six phosphorylations (Figure 4B). This was confirmed by use of a Phosphoprotein Phosphate Estimation kit, revealing a total of 7.8 phosphate groups in OPN60 and 6.9 phosphate groups in the sample containing OPN45/OPN50. Subtraction of the theoretical mass of the human OPN polypeptide (33714 Da) from the observed average mass of dephosphorylated OPN60 (37.1 kDa) leaves approx. 3.4 kDa, which must then be accounted for by other PTMs.

Figure 4 MALDI–TOF-MS analysis of native and dephosphorylated urinary OPNs

(A) The average mass peaks at ∼37.7 kDa (black trace) and 37.1 kDa (grey trace) represent OPN60 before and after treatment with ALP. (B) The average mass peaks at ∼29.3 kDa (black trace) and 28.8 kDa (grey trace) represent the N-terminal OPN fragments before and after treatment with ALP.

Phosphorylations

To identify potential phosphorylation sites in urinary OPN, the OPN fraction (Figure 2) was digested with trypsin and the resulting peptides were separated by RP-HPLC (results not shown). A large acidic fragment of OPN (Gln36/Ser62–Arg143) not susceptible to tryptic cleavage was further digested with thermolysin and the resulting peptides were separated by RP-HPLC. All fractions from the RP-HPLC separations of peptides were analysed by MALDI–TOF-MS (Table 2; Supplementary Tables 1 and 2 at http://www.BiochemJ.org/bj/411/bj4110053add.htm). In total, the approximately eight phosphate groups in OPN60 were distributed over 30 potential phosphorylation sites. Some of the phosphopeptides contained more serine/threonine residues than the observed number of phosphorylations. In these cases, the phosphate groups were assigned to residues fitting the target sequence of the Golgi kinase/mammary-gland casein kinase [S/T-X-E/S(P)/D], based on the localization of phosphorylated residues in other OPN isoforms [2729]. Data from the peptide phosphorylation analysis are summarized in Table 2 and the resulting map of PTMs is shown in Figure 5, which also contains a comparison with the PTMs of human milk OPN.

View this table:
Table 2 Characterization of phosphorylated peptides

The peptides from the tryptic and thermolytic (Gln36/Ser62-Arg143) digests of OPN were characterized by MALDI–TOF-MS. The number of phosphates indicates the different phosphorylation variants of each individual peptide observed by MS. MGCK, mammary-gland casein kinase.

Figure 5 PTMs in human urinary and milk OPNs [28]

Phosphorylations (black), a sulfation (black) and glycosylations (grey) are highlighted. The glycan structures observed in urinary OPN are indicated: GalNAc (black squares), galactose (grey diamonds) and sialic acid (white circles). The cleavage sites observed in OPN45, OPN50 and OPN60 from urine are indicated by black arrows. Known proteolytic cleavage sites for MMP-3 and MMP-7 [40] and thrombin [41] are indicated by grey arrows.

Sulfation of OPN

The linear MALDI–TOF-MS spectrum of a peptide covering R160PDIQYPDATDEDITSHMESEELNGAYK187 showed m/z values of 3307.18, 3387.53 and 3466.42 (Figure 6A). These masses correspond to Arg160–Lys187 with excess masses of ∼80, ∼160 and ∼240 Da respectively. These masses could correspond to different phosphorylation variants, or they could represent a combination of phosphate and sulfate groups on the peptide. After treatment with ALP, an excess mass of ∼80 Da remained associated with the peptide (Figure 6B), strongly indicating that Arg160–Lys187 is also sulfated. The presence of a sulfate group was verified by treatment of the dephosphorylated peptide with arylsulfatase, which removed the excess mass of 80 Da (Figure 6C). The peptide Arg160–Lys187 contains two potential sites for sulfation (Tyr165 and Tyr186). The residue corresponding to Tyr165 has been shown to be sulfated in OPN from rat bone [29], and it is therefore most likely that this residue is sulfated in urinary OPN. Two other tyrosine-containing peptides, Gln5–Lys35 and Gly205–Lys225, were observed with excess masses that could correspond to both phosphorylations and/or sulfate groups (Supplementary Table 1). Treatment of these peptides with ALP showed that they only contained phosphate groups. Peptides containing Tyr165 in human milk OPN were also analysed for sulfation. However, no indication of this modification was observed in the milk isoform (results not shown).

Figure 6 MALDI–TOF-MS in linear mode of a sulfated peptide

(A) MS spectrum of a fraction containing the peptides Gln36–Lys61 and Arg160–Lys187. The mass at m/z 3466.42 corresponds to the peptide Arg160–Lys187 modified by three phosphate/sulfate groups. (B) MS after treatment with ALP. The mass at m/z 3306.43 corresponds to Arg160–Lys187 with an excess mass of ∼80 Da, indicating the presence of a sulfate group. (C) MS after treatment with alkaline phosphatase and subsequently arylsulfatase. The mass loss of 80 Da from Arg160–Lys187 upon treatment with the sulfatase shows that this peptide contained two phosphorylations and one sulfate group.

Glycosylations

All O-glycosylated residues in OPN from urine were located in a single thermolytic peptide, Leu116–Thr139. This peptide was further digested with staphylococcal (S. aureus) V8 protease, resulting in the two peptides, L116VTDFPTDLPATE128 and V129FTPVVPTVDT139, which were separated by RP-HPLC. The peptide L116VTDFPTDLPATE128 was observed in linear MS with m/z values at 3679.83, 3970.97 and 4261.24 (Table 3). These masses correspond to Leu116–Glu128 with three GalNAc-galactose units and four to six sialic acid units attached. The lack of a PTH-threonine at the positions of Thr118, Thr122 and Thr127, when subjected to Edman sequencing, indicates that these threonine residues are glycosylated in the peptide. Similarly, MS analysis of the peptide V129FTPVVPTVDT139 showed that this peptide contained glycosylations composed of up to two GalNAc, one GlcNAc, two galactose and five sialic acid units (Table 3). Edman sequencing showed that Thr131 and Thr136 are modified. Collectively, these results show that Thr118, Thr122, Thr127, Thr131 and Thr136 are glycosylated in urinary OPN. The absence of any trace of PTH-amino acids in these positions indicates that they are fully glycosylated. The masses of the glycopeptides show that each glycosylated threonine residue can be modified by different glycan structures, but based on the major mass peaks for both Leu116–Glu128 and Val129–Thr139 they seem mainly to consist of the disialylated core-1 O-glycan sialic acid–galactose–[sialic acid]–GalNAc structure.

View this table:
Table 3 Characterization of glycosylations on urinary OPN

(a) The peptides were characterized by Edman sequencing and MALDI–TOF-MS. Glycosylated residues are indicated with boldface letters. The observed molecular masses were determined by MALDI–TOF-MS in linear mode. Mass difference is that between the observed and calculated average masses. Based on the MS data, the composition of the individual glycans is suggested in parentheses. Sia, sialic acid. (b) The carbohydrate composition of purified OPN was determined as described in the Experimental section. n.d., Not detected.

OPN60 was subjected to mild acid hydrolysis to determine the monosaccharide composition of the glycans on the protein and thereby evaluate the structures observed in MS. The obtained monosaccharides were separated by high-pH anion-exchange chromatography and subsequently monitored by pulsed electrochemical detection. The hydrolysates of OPN60 showed the presence of four types of monosaccharides: GalNAc, GlcNAc, galactose and sialic acid (Table 3). The detected amounts of monosaccharides give approx. 1 mol of GalNAc/galactose and 2 mol of sialic acid per glycosylated threonine on OPN, corresponding well to the glycans observed by MS.

The human OPN sequence contains two asparagine residues (Asn63 and Asn90) in putative N-glycosylation motifs. MS analysis of peptides containing these residues showed that none of these asparagine residues are glycosylated in urinary OPN. In addition, incubation of urinary OPN with PNGase F did not result in altered migration when subjected to SDS/PAGE (results not shown). Furthermore, no trace of mannose was detected in the analysis of OPN monosaccharides (Table 3), substantiating the absence of N-linked glycosylations in human urinary OPN.

DISCUSSION

The present study describes the purification and characterization of urinary OPN. Western blotting of urine obtained from seven healthy donors shows that urinary OPN exists in different forms migrating between 45 and 60 kDa (Figure 1). In the OPN peak eluted from the Vydac C4 column, three dominating forms were observed by CBB staining and immunoblotting (Figure 3A). The strongly staining top band observed by Western blotting may consist of more than one isoform. This assumption is supported by Western-blot detection of two bands in the samples containing OPN60 using the monoclonal antibodies 1H3 (faint) and 3D9 (Figure 3B).

The present study indicates that the N-terminal fragments OPN45 and OPN50 are generated from the full-length molecule by proteolytic cleavage at Arg228–Leu229 and/or Tyr230–Lys231 in urine. Although two bands are seen in SDS/PAGE, representing OPN45 and OPN50, only one broad peak is observed by MS, suggesting that the mass difference between the fragmented OPNs is not large enough to give separation in linear MALDI–TOF-MS. No heterogeneity at the identified O-glycosylation sites was observed. However, differences in the glycans at the individual sites might account for the observed migration of OPN45 and OPN50 during SDS/PAGE.

OPN is present in human urine in different fragmented forms, and it is therefore relevant to speculate on how these forms are generated. The protein has previously been shown to be a substrate for MMP-3 (matrix metalloproteinase-3) and MMP-7 [40] and thrombin [41]. In Figure 5, the cleavage sites of these proteolytic enzymes are compared with the observed fragmentation patterns of the characterized urinary OPN, showing that neither MMP-3, MMP-7 nor thrombin have been reported to cleave OPN at the sites reported in the present study. It is not clear whether OPN45/OPN50 is the result of cleavage by specific proteases in the urinary secretory system. The cleavage sites identified in the present study point in two directions. The cleavage of Arg228–Leu229 could be catalysed by a thrombin-type protease, whereas the cleavage site Tyr230–Lys231 fits the specificity of a chymotrypsin-type protease. However, it remains to be shown which proteases are actually responsible for the cleavage of OPN in urine. Kidney cells have been reported to produce an OPN peptide that is thought to represent tissue-specific intracellular processing of the intact protein [42], which supports the hypothesis that OPN45 and OPN50 are proteolytically derived fragments of the intact protein. Alternative mRNA splicing could also be speculated to be responsible for the presence of the truncated OPN forms. However, the region containing the cleavage sites is encoded by exon 7 of the OPN gene, and no intron–exon junction is located in the vicinity. It is therefore a much more plausible explanation that the N-terminal fragments are products of proteolytic activity rather than alternative mRNA splicing.

Interestingly, no C-terminal fragments corresponding to Leu229/Lys231–Asn298 were detected by the polyclonal antibody in the Western blot of urine (Figure 1). This indicates that the C-terminus of urinary OPN is completely degraded when OPN45 and OPN50 are generated from OPN60. In support of this, no low-molecular-mass C-terminal OPN fragments were detected in Western blots of human urine using different monoclonal OPN antibodies with epitopes mapped to the C-terminal part of the protein [39]. OPN forms below the 40 kDa electrophoretic mobility range have been observed in urine from patients with kidney stones [24], whereas normal individuals had OPN forms migrating between 55 and 66 kDa, as observed in the present study. These low-molecular-mass forms of OPN were generated by serine proteases and could reflect that augmented cleavage of urinary OPN could be a critical step in the formation of kidney stones.

OPN60 was shown to contain approximately eight phosphorylations and six phosphate groups were estimated to decorate OPN45/OPN50. This indicates that the region Leu229/Lys231–Asn298 lacking in these cleaved forms contains approx. two phosphate groups. It is not clear whether some serine or threonine residue in OPN are phosphorylated to a higher degree than others. However, the fragment Leu229/Lys231–Asn298 that is lacking in OPN45/OPN50 constitutes approximately one-fourth of the amino acids in OPN. So, the observation that this fragment contains two out of eight phosphorylations suggests that the phosphate groups decorating the intact molecule are evenly distributed. The degree of phosphorylation on urinary OPN is significantly lower than the levels observed in milk OPN [27,28], showing tissue-specific phosphorylation. A comparison of OPN from human milk and urine with regard to PTMs is shown in Figure 5. It is interesting to observe that of the 36 phosphorylation sites previously located in human milk OPN, 30 sites were also identified in human urinary OPN. However, the average phosphate content of the urinary isoform is only approximately one-third of the milk isoform. In both urine and milk, the highly anionic regions of OPN composed of multiple acidic amino acids and phosphorylations could constitute potential binding sites for minerals and calcium salts. These anionic regions could therefore enable OPN to form soluble complexes with calcium ions and thereby inhibit unintentional calcium precipitation and crystallization. It is not known whether the different degrees of phosphorylation of human milk and urinary OPNs are due to tissue-specific phosphorylation or dephosphorylation.

Urinary OPN was shown to be sulfated at Tyr165. This residue has recently been shown to be sulfated in rat bone OPN [29], and OPN produced from canine kidney cells has also been shown to be sulfated by metabolic labelling [42]. Tyr165 is conserved in mouse, rat and rabbit but not present in cow, sheep, pig and chicken, indicating that sulfation of the protein is species-dependent. No indication of a sulfate group on Tyr165 was observed in human milk OPN, showing that this modification is also tissue-specific.

In the present study, urinary OPN has been shown to contain five O-glycosylated threonine residue (Thr118, Thr122, Thr127, Thr131 and Thr136). As shown in Figure 5, all O-linked glycans identified in human urine OPN are also present in human milk OPN. It appears that the glycans in urinary OPN predominantly consist of a disialylated GalNAc-galactose core. This glycan structure has also been described in OPN from rat bone and mouse fibroblasts and osteoblasts [29,30]. No N-linked glycosylations were identified on urinary OPN at the two putative motifs (Asn63 and Asn90). This is consistent with data on milk, bone and fibroblast OPNs [2730]. Normal rat kidney cells have been shown to produce both phosphorylated and non-phosphorylated forms of OPN in vitro, and the non-phosphorylated isoform contained N-linked carbohydrates [37]; whether this is also the case in vivo remains to be shown.

In summary, human urine contains different forms of OPN representing the full-length protein and truncated forms lacking the C-terminal fragment Leu229/Lys231–Asn298. The intact protein contains approximately eight phosphate groups distributed over 30 phosphorylation sites, whereas the fragments contain approximately six phosphate groups. In addition, one sulfate group and five O-glycosylated threonine residues were identified in urinary OPN. The PTMs of urinary OPN reported here are likely to play key roles in the protein's regulation of mineral crystallization in urine. Further studies are necessary to investigate whether different PTMs could account for the pleiotropic role of OPN as both an inhibitor and promoter of renal stone formation.

Abbreviations: ALP, bovine alkaline phosphatase; CBB, Coomassie Brilliant Blue; COD, calcium oxalate dihydrate; COM, calcium oxalate monohydrate; MALDI–TOF-MS, matrix-assisted laser-desorption ionization–time-of-flight MS; MMP, matrix metalloproteinase; OPN, osteopontin; PNGase F, peptide N-glycosidase F; PTH, phenylthiohydantoin; PTM, post-translational modification; μRPC, μ reverse-phase chromatography; RP-HPLC, reverse-phase HPLC; TFA, trifluoroacetic acid

References

View Abstract