Interaction of bone proteoglycans and proteoglycan components with hydroxyapatite

Biochimica et Biophysica Acta 1568 (2001) 118^128

www.bba-direct.com

Interaction of bone proteoglycans and proteoglycan components with hydroxyapatite Sarah G. Rees *, Diana T. Hughes Wassell, Rachel J. Waddington, Graham Embery Department of Basic Dental Science, Dental School, University of Wales College of Medicine, Heath Park, Cardi¡ CF14 4XY, UK Received 6 June 2001; received in revised form 4 September 2001; accepted 5 September 2001

Abstract The small leucine-rich proteoglycans (SLRPs) of bone interact with hydroxyapatite (HAP) and are proposed to play an important role in the regulation of the mineralisation process. The present study has examined the interaction of bone SLRPs, purified, liberated bone glycosaminoglycan (GAG) chains and core proteins, as well as commercial chondroitin 4-sulphate (C4S) with HAP. Isotherm data (0.02 M sodium acetate) revealed that the intact proteoglycans (PGs) and bone GAGs showed greater binding onto HAP with higher adsorption maxima than the constituent core proteins and commercial C4S. Adsorption was dependent on pH and ionic strength, increasing with decreasing pH and in the presence of calcium whilst decreasing in the presence of phosphate, suggesting that electrostatic effects are important. The data indicates that PG/GAG chemistry and conformation in solution are significant determinants in the adsorption process and provides important information concerning interfacial adsorption phenomena between the organic-inorganic phases of mineralised systems. ß 2001 Elsevier Science B.V. All rights reserved. Keywords : Decorin; Biglycan; Glycosaminoglycans ; Hydroxyapatite; Adsorption ; Mineralization

1. Introduction The matrix proteins of bone and cartilage have been implicated in a number of biological processes including the regulation of matrix organisation, cell adhesion, matrix turnover and mineral deposition. Speci¢cally, the small leucine-rich proteoglycans (SLRPs) decorin and biglycan have been implicated in the regulation of collagen ¢brillogenesis [1,2] and have been shown to bind to a number of other connective tissue constituents including transforming growth factor-L [3], ¢bronectin [4] and heparin cofactor II [5]. In addition, bone SLRPs have been implicated in the regulation of mineralisation, di¡using across the osteoid and binding directly to the mineralising front [6] ; studies have also shown that PGs disappear from the initial sites of mineralisation [7^9]. Furthermore, the expression and localisation of decorin and biglycan has been demonstrated in adult human bone [10] and developing human bone [11]. Decorin has also been implicated in heterotropic ossi¢cation [12] and has been shown to bind * Corresponding author. Connective Tissue Biology Laboratories, Cardi¡ School of Biosciences, Cardi¡ University, Museum Avenue, Cardi¡ CF10 3US, UK. Fax: +44-29-2087-4594. E-mail address : reessg1@cardi¡.ac.uk (S.G. Rees).

to a speci¢c face (100) of HAP crystals thus controlling of the morphogenesis of such crystals [13]. In addition, decorin and biglycan have been demonstrated to regulate hydroxyapatite formation in a gelatin gel system [14]. It is the a¤nity of such extracellular matrix components for HAP that is the basis for any control mechanism in£uencing the nucleation, orientation, size and growth rate of the calcium phosphate formed [15]. HAP mineral is amphoteric in nature, presenting a complex mosaic of charges due to its calcium, phosphate and hydroxyl groups and thus facilitates the binding of both basic and acidic proteins. However, anionic molecules such as PGs are more readily adsorbed [16]. Electrostatic interactions occur between the cationic sites on the mineral (calcium) and the anionic domains on the macromolecule (carboxyl and sulphate) [17]. Multi-point binding due to the presence of a number of functional groups on the macromolecule may also occur in the adsorption process [18]. In addition, protein conformation is an important determinant in HAP interaction since only native proteins bind to HAP e¡ectively; a di¡use native structure renders binding di¤cult [19]. Interactions of PGs and GAGs with HAP are also signi¢cant in the process of osseointegration. Apatites are excellent biomaterials due to their biocompatibility; the

0304-4165 / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 0 4 - 4 1 6 5 ( 0 1 ) 0 0 2 0 9 - 4

BBAGEN 25247 6-12-01

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

119

bone-implant interface comprises a so-called bonding zone composed of a calcium- and phosphorus-rich proteinaceous matrix [20]. It is generally accepted that the initial event in the interaction of implant materials with tissues is the adsorption of blood proteins and tissue PGs to the surface [21,22]. However, knowledge of the interface between bone and synthetic apatite surfaces is not clearly understood. Few studies have been carried out to elucidate the mechanisms of interfacial adsorption between bone SLRPs and HAP and the role of PG constituents in the process. The aim of the present work, therefore, was to examine the interaction of bone SLRPs and PG constituents (i.e., the liberated GAG chains and core proteins isolated from the native molecule) with HAP, as well as adsorption under di¡erent solution conditions. Due to the fact that such interactive phenomena represent a two-way exchange process, desorption of the PGs was also examined. The adsorption of commercial chondroitin 4-sulphate (C4S) was also included by way of comparison in order to aid interpretation as chondroitin sulphate is the predominant GAG of bone [23^25].

using colorimetric assay with dimethylmethylene blue (DMMB ; Serva) [26]. Puri¢ed bone PGs were characterised using sodium dodecyl sulfate^polyacrylamide gel electrophoresis (SDS^PAGE), cellulose acetate electrophoresis, amino acid analysis and Western blotting (with monoclonal antibodies (mAbs) CS-56, recognising chondroitin sulphate (Sigma) ; mAb 70.6, recognising bovine decorin core protein [27]; and mAb PR85, recognising a biglycan C-terminal epitope [28].

2. Materials and methods

2.4. Isolation and characterisation of proteoglycan core proteins

2.1. Materials HAP powder, with a surface area of 26 m2 g31 and a calcium/phosphate ratio of 1.61, was obtained from Unilever, UK. The theoretical calcium/phosphate ratio predicted by the structural formula of HAP is 1.67; however, empirical measurement of this ratio generally gives lower results. The HAP used in this work is thus termed `calcium de¢cient'. C4S (Sigma, UK; catalogue no. C-8529) is a sodium salt extracted from bovine trachea (molecular mass 46 000) and is approximately 70% C4S, the balance being chondroitin 6-sulphate (C6S). 2.2. Isolation and characterisation of alveolar bone proteoglycans The extraction, puri¢cation and characterisation of PGs from ovine alveolar bone has been described previously [25]. Brie£y, bone was treated twice with collagenase-dispase (0.1 and 0.8 U/ml in phosphate-bu¡ered saline; Boehringer Corp., UK) to facilitate the removal of soft, adherent tissue. Powdered bone was demineralised for 14 days with 10% EDTA (trisodium salt) and non-collagenous proteins extracted from the organic matrix with 4 M guanidinium chloride, 0.05 M sodium acetate. PGs were subsequently puri¢ed by two-stage anion exchange chromatography using a Q-Sepharose column (HiLoad 16/10) followed by a Resource-Q column (1 ml) incorporated into an FPLC system (Pharmacia Biotech, UK) and PG-rich fractions measured by sulphated GAG content

2.3. Isolation and characterisation of glycosaminoglycan chains Samples of puri¢ed PGs (10 g l31 ) were treated with non-speci¢c protease (10 g l31 ; Sigma, type XIV) in 0.2 M Tris^HCl, 10 mM calcium chloride (pH 7.5) for 18 h at 55³C. The protein was separated from the GAG chains by precipitation with 50% trichloroacetic acid and the supernatant recovered by microcentrifugation, desalted using a PD-10 column (Pharmacia, UK) and the digest recovered by lyophilisation. The puri¢ed bone GAGs were characterised by cellulose acetate electrophoresis as previously described [25,29].

Core proteins were obtained from intact bone PGs through chondroitinase ABC digestion (protease free, Seikagaku Corp., Japan); equal volumes of PGs (5.0 g l31 ) and chondroitinase ABC (0.5 units ml31 ) were incubated together in a 0.1 M Tris^HCl, 0.03 M sodium acetate bu¡er (pH 8.0) at 37³C for 40 min. The digest was exhaustively dialysed against double-distilled water containing protease inhibitors and lyophilised. Separation of core proteins and chondroitinase ABC was achieved using gel ¢ltration chromatography on a pre-packed Sephadex 75 HR 10/30 column, incorporated into an FPLC system (Pharmacia Biotech, UK) using a 0.05 M Tris^HCl, 0.1 M sodium chloride bu¡er, pH 7.2. Absorbance was monitored at 280 nm and fractions (1 ml) pooled, dialysed and lyophilised. SDS^PAGE was used in order to con¢rm the separation of the PG core proteins and chondroitinase ABC following gel ¢ltration. Prior to analysis, samples were lyophilised and redissolved in SDS^PAGE dissociating bu¡er (0.5 M Tris^HCl bu¡er (pH 6.8) containing 10% SDS, 0.5% bromophenol blue, 10% glycerol, and 5% 2-L-mercaptoethanol) at a concentration of 10 g l31 . Samples were subsequently separated using a Pharmacia Phast System on 3 cm pre-formed 4^15% gradient gels. A low electrical output (100 V, 1 mA, 1 W) was initially applied for 4 Vh to draw the samples through the stacking gel. The output was subsequently increased (250 V, 3 mA, 3 W, 66 Vh) for the separation step. The separated components were then stained using a silver staining kit (Sigma).

BBAGEN 25247 6-12-01

120

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

2.5. Adsorption of bone proteoglycans, glycosaminoglycans and core proteins to hydroxyapatite

2.6. Desorption of bone proteoglycans and commercial C4S from hydroxyapatite

Adsorption isotherms were based on the method of Hughes Wassell et al. [30]; however, in order to minimise the amount of puri¢ed PGs and PG constituents used in adsorption experiments, samples of HAP powder (0.001 g) were pre-equilibrated with 0.09 ml of 0.02 M sodium acetate (pH 5.0^6.8) for 1 h at 37³C, prior to the addition of the adsorbate, allowing the bu¡er solution to become saturated with respect to the HAP. Samples of PGs, PG core proteins or puri¢ed bone GAGs (0.01 ml) ranging in concentration from 0.2 to 10.0 g l31 range were added to the HAP, giving a ¢nal concentration in solution of 0.02^ 1.0 g l31 . The tubes (in triplicate) were incubated for 4 h at 37³C (it has previously been demonstrated that equilibrium has been achieved in the adsorption experiments by this time point [31]), the samples microcentrifuged for 5 gmin and the supernatants assayed for sulphated GAG or protein using the DMMB [26] or Lowry [32] methods, respectively. Standard calibration curves were established for both assays in the 0^1.0 g l31 concentration range using PGs or PG constituents as standards (which yielded correlation coe¤cients (r2 ) of s 0.9) from which experimental concentrations were subsequently calculated. Controls were set up (in the absence of HAP) at each PG or PG constituent concentration, allowing corrections to be made for GAG losses in the system. The amount of binding per speci¢c surface area was calculated by solution depletion, i.e., the di¡erence between the GAG/protein content of the supernatant and the original GAG/protein content of the solution. These value were corrected for the amount of HAP present and expressed as speci¢c surface area (mg m32 ). To assess the e¡ect of pH on the adsorption of PGs onto HAP, adsorption isotherms were established as described with 0.02 M sodium acetate in the pH range 5.0^ 7.1. For the pH-dependent adsorption of commercial C4S, samples of HAP (0.01 g) were pre-equilibrated with 0.9 ml of 0.02 M sodium acetate (pH 5.0^7.1) for 1 h at 37³C and C4S samples (0.1 ml) in the 0.2^10.0 g l31 concentration range subsequently added. In order to exclude the possibility of HAP dissolution over the time-course and pH range of the experiments, the pH of 0.02 M sodium acetate bu¡er (pH 5.0 and 7.1) was monitored at hourly intervals over a 4-h period and ¢nally at 24 h following the addition of HAP and incubation at 37³C. The e¡ect of ionic strength on adsorption was also examined where experiments were carried out in 0.02 M sodium acetate (pH 6.8) with the addition of calcium acetate (0.002^0.2 M) and di-sodium orthophosphate (0.002^ 0.02 M). A limitation in the amount of puri¢ed PG constituents (i.e., core proteins and GAG chains) precluded an investigation of the e¡ect of both pH and ionic strength on adsorption.

PG samples (0.04^1.0 g l31 , n = 8 at each concentration) were adsorbed onto HAP using 0.02 M sodium acetate (pH 6.8) as described. Desorption was achieved through the addition of either (a) 0.5 ml of 0.05 M sodium acetate, pH 6.8, giving a ¢nal solution concentration of 0.02 M or (b) 0.5 ml of 0.86 M di-sodium orthophosphate, pH 6.8, giving a ¢nal solution concentration of 0.36 M. The samples were placed on a rotating wheel and incubated at 37³C for 2^24 h. After 2 h, samples were microcentrifuged for 5 min and a small volume removed for sulphated GAG determination after which they were replaced at 37³C on a rotating wheel for a further 22 h, and the assay procedure repeated. Desorption was expressed as the di¡erence (percentage) between the GAG content of the PG solution at the start and end of the 2- and 24-h time periods. Controls were set up at each PG concentration, allowing corrections to be made for PG losses in the system. The pH-dependent desorption of C4S was also examined where samples of C4S (0.02^0.2 g l31 , n = 8 at each concentration) were adsorbed onto HAP using 0.02 M sodium acetate (pH 5.0^7.1) in the manner described. Desorption was achieved through the addition of either (a) 0.5 ml of 0.05 M sodium acetate, pH 5.0^7.1, giving a ¢nal solution concentration of 0.02 M, or (b) 0.5 ml of 0.86 M di-sodium orthophosphate, pH 5.0^7.1, giving a ¢nal solution concentration of 0.36 M. 2.7. Analysis of adsorption data The plot of the amount adsorbed per unit area of the adsorbent versus the equilibrium concentration of the adsorbate in solution constitutes the adsorption isotherm for the experimental adsorbent and adsorbate. The isotherm is an experimental result independent of any adsorption models. However, most protein adsorption data ¢t reasonably well to the Langmuir-type model [33]. The classical Langmuir theory was originally used to describe gas adsorption onto solids, but can be applied to adsorption from solutions if su¤ciently dilute. Langmuir treatment of adsorption has a number of assumptions: only one molecule can be adsorbed per site (commonly called the monolayer assumption) ; only one type of site is present (there is a homogeneous surface); the adsorption of one molecule does not a¡ect the adsorption of other molecules (no lateral interactions or cooperativity); only one adsorbing species is present; the solution is dilute; the adsorption is reversible. However, there are limitations to this model as deviations from the ideal situation occur, for example, it is frequently found that adsorption is irreversible. Data can be expressed in terms of the Langmuir equation. For the equilibrium : P ‡ AIPA

BBAGEN 25247 6-12-01

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

121

Fig. 1. Elution pro¢le of core protein digest separated by gel ¢ltration chromatography.

where P is protein in solution, A the unoccupied surface site and PA the occupied surface site; Langmuir-type adsorption gives K‰PŠ‰AT Š=1 ‡ K‰PŠ ˆ ‰PAŠ

…1†

Linearisation of Eq. 1 gives 1=‰PAŠ ˆ 1=K‰AT Š‰PŠ ‡ 1‰AT Š

…2†

where [AT ] is the maximum number of adsorption sites per unit of surface area of the protein (mg m32 ) and K re£ects the a¤nity that the protein molecules have for the adsorption sites (l g31 ). Based on this model, the experimental data should satisfy the linearised form of Eq. 1. According to Eq. 2, a plot of 1/[PA] against 1/[P] should yield a straight line from which K and AT can be determined from the slope and intercept of such a line. The best ¢t of the linear plots is determined according to a leastsquares criterion. The correlation coe¤cients (r2 ) of the individual linear regressions indicate the degree of agreement between Eq. 2 and the experimental data.

the GAG moiety released from the PG species by protease digestion was identi¢ed by cellulose acetate electrophoresis, revealing that chondroitin sulphate represented approximately 86% of the total GAG present whilst dermatan sulphate contributed to the remaining 14%. The total GAG represented 16.5 þ 2.1% by dry weight of the PG fraction. The elution pro¢le for the separation of the PG-rich fraction following digestion with chondroitinase ABC (utilising gel ¢ltration chromatography) is shown in Fig. 1. The elution pro¢le for the separation of chondroitinase ABC only produced discrete absorbance peaks at an elution volume of 16^24 ml. Following chondroitinase ABC

3. Results 3.1. Characterisation of bone proteoglycans The extraction and puri¢cation procedures described yielded a PG extract associated with the mineralised matrix of alveolar bone as previously reported by our laboratory [25]. In summary, electrophoretic separation of the intact PG-rich fraction by SDS^PAGE revealed protein staining material ranging in molecular mass from 250 kDa to s 29 kDa. Western blots following separation by SDS^PAGE revealed native chondroitin sulphate-containing material (immunoreactive with mAb CS-56), and positive immunostaining with mAbs 70.6 and PR85 (speci¢c for decorin and biglycan, respectively). The nature of

Fig. 2. SDS^PAGE analysis of chondroitinase ABC, chondroitinase ABC/PG digest (prior to separation) and PG core proteins (following separation by gel ¢ltration chromatography).

BBAGEN 25247 6-12-01

122

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

Fig. 3. Adsorption isotherms for intact proteoglycans, glycosaminoglycans and commercial chondroitin 4-sulphate onto hydroxyapatite.

digestion of the core protein sample a new absorbance peak of lower molecular mass was also apparent (31^ 36 ml elution volume), representing the core proteins. This was con¢rmed through SDS^PAGE analysis which revealed a band at 49 kDa (Fig. 2), corresponding in molecular mass to the core proteins of PGs previously characterised from human alveolar bone [29] and bovine cortical bone [23]. 3.2. Adsorption of intact proteoglycans, glycosaminoglycan chains and core proteins The shape of the adsorption isotherms for intact PGs, alveolar bone GAG chains and commercial C4S were not continuous with decreases in adsorption occurring after certain solution concentrations; however, repeat experi-

ments showed this discontinuity to be reproducible (adsorption experiments were repeated at least four times and values represent the means of the replicates). Maximum adsorption (Admax ) values are means þ the standard error of the mean (S.E.M.). Intact PGs showed greatest adsorption onto HAP, followed by the alveolar bone GAG chains and commercial C4S, in descending order (Fig. 3). The binding curves did not follow the Langmuir model, with double reciprocal plots of [PA] against [P] yielding r2 values of 0.433, 0.009 and 0.063 for intact PGs, bone GAG chains and commercial C4S, respectively. The corresponding Admax values were 0.775 þ 0.021 (n = 8), 0.658 þ 0.027 (n = 5) and 0.316 þ 0.031 (n = 8), respectively. The bone GAG chains and commercial C4S showed increased adsorption to the maximum value at solution concentrations of approximately 0.2 and 0.35 g l31 , respec-

Fig. 4. Adsorption isotherms for intact proteoglycans and proteoglycan core proteins onto hydroxyapatite.

BBAGEN 25247 6-12-01

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

tively, after which adsorption decreased. Intact PGs showed a small decrease in adsorption after a solution concentration of approximately 0.2 g l31 ; however, a plateau was reached at the higher concentrations. Isotherms were obtained for the adsorption of isolated core proteins onto HAP and samples of both intact PGs and core proteins assayed for protein content to allow for comparison (Fig. 4). The intact PGs showed notably higher adsorption than the constituent core proteins with Admax values of 1.488 þ 0.15 (n = 5) and 0.519 þ 0.03 (n = 4), respectively, and binding was non-Langmuir in character for both the intact PGs (AT = 34.148) and the core proteins (r2 = 0.009). 3.3. pH-Dependent adsorption of proteoglycans and chondroitin 4-sulphate Isotherms were obtained for the adsorption of intact PGs and commercial C4S onto HAP at varying pH values and repeat experiments showed these trends to be reproducible. The adsorption of PGs onto HAP in 0.02 M sodium acetate at pH 5.0 was greater than at pH 6.8 with Admax values of 2.069 þ 0.031 mg m32 (n = 5) versus 0.775 þ 0.021 mg m32 (n = 8), respectively (Fig. 5). Binding at pH 5.0 was also non-Langmuir, with the double reciprocal plot yielding an r2 value of 0.631. The adsorption of commercial C4S showed a similar trend, increasing with decreasing pH and showing non-Langmuir character (Fig. 6A). The Admax values at pH 5.0, pH 6.8 and pH 7.1 were 0.570 þ 0.025 mg m32 (n = 6), 0.316 þ 0.021 mg m32 (n = 8) and 0.270 þ 0.020 mg m32 (n = 6), with r2 values of 0.063, 0.001 and 0.192, respectively. A general trend emerged for each pH value, where C4S adsorption increased up to a solution concentration of approximately 0.3 g l31 , then decreased. In order to con¢rm the stability of HAP in 0.02 M sodium acetate bu¡er (pH 5.0 and 7.1), the solution pH

123

Table 1 Stability of hydroxyapatite in bu¡er Time (h) 0.2 M sodium acetate, pH 5.0 0.2 M sodium acetate, pH 7.1 1 2 3 4 24

pH pH pH pH pH

5.27 5.29 5.31 5.30 5.31

pH pH pH pH pH

7.04 7.02 7.04 7.04 7.08

was monitored over a 24-h period following addition of HAP (Table 1). A change at pH 5.0 occurred at 1 h due to initial equilibration of the bu¡er with HAP following its addition; however, pH was then maintained at a steady state, con¢rming that negligible HAP dissolution was occurring. This is supported by studies which show that dissolution of calcium does not vary signi¢cantly between pH 5.0 and pH 7.0 [34]. 3.4. Determination of ionic strength dependence Isotherms were obtained for the adsorption of intact PGs and commercial C4S onto HAP with varying calcium acetate concentrations. A limitation in the amount of puri¢ed PGs precluded adsorption with a greater array of calcium acetate concentrations. The presence of 0.02 M calcium acetate increased the adsorption of PGs producing an Admax value of 2.089 þ 0.019 mg m32 (n = 8, r2 = 0.320) compared with 0.775 þ 0.021 (n = 8, r2 = 0.433) in the absence of calcium acetate (Fig. 5). The presence of calcium acetate also increased C4S adsorption in a dose-dependent manner with the exception of 0.2 M calcium (Fig. 6B). The Admax values for the control (no calcium acetate), 0.002, 0.02 and 0.2 M calcium acetate were 0.316 þ 0.031 mg m32 (n = 8), 0.954 þ 0.031 mg m32 (n = 5), 1.469 þ 0.05 mg m32 (n = 5), and 0.560 þ 0.041 mg m32 (n = 5), respectively. Adsorption did not follow the Langmuir model yielding r2 values of 0.001, 0.170, 0.524 and 0.019, respectively.

Fig. 5. Adsorption isotherms for intact proteoglycans onto hydroxyapatite showing pH and ionic strength dependence.

BBAGEN 25247 6-12-01

124

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

Fig. 6. Adsorption isotherms for commercial chondroitin 4-sulphate onto hydroxyapatite showing (A) pH, (B) calcium, and (C) phosphate.

BBAGEN 25247 6-12-01

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

125

Table 2 Desorption of proteoglycans and chondroitin 4-sulphate from hydroxyapatite [PG]/[C4S] (g l31 )

pH

% Desorption After 2 h 0.02 M CH3 COONa

0.04 PG 0.4 PG 1.0 PG 0.02 C4S

0.08 C4S

0.2 C4S

6.8 6.8 6.8 7.1 6.8 5.0 7.1 6.8 5.0 7.1 6.8 5.0

^ ^ ^ ^ ^ ^ ^ ^ ^ 54 41 ^

After 24 h 0.36 M Na2 HPO4 93 95 100 100 100 78 100 100 100 100 100 100

Isotherms were obtained for the adsorption of PGs and C4S onto HAP in the presence of varying sodium di-orthophosphate concentrations. For commercial C4S, a dose-dependent decrease in adsorption occurred with increasing phosphate concentrations (Fig. 6C), producing an Admax value of 0.123 þ 0.015 mg m32 (n = 5) for 0.002 M sodium di-orthophosphate. Adsorption was abrogated for both C4S and intact PGs in the presence of 0.02 M phosphate (Fig. 6C and 5, respectively). 3.5. Desorption of proteoglycans and chondroitin 4-sulphate Desorption data for intact PGs and commercial C4S (at varying pH values) in the presence of 0.02 M sodium acetate or 0.36 M di-sodium orthophosphate is shown in Table 2. Desorption at three di¡erent concentrations is shown, corresponding to the concentration range of the adsorption isotherm. The limited amount of puri¢ed PGs precluded desorption at varying pH values. In the presence of 0.02 M sodium acetate there was no desorption after 2 or 24 h for intact PGs; however, in the presence of di-sodium orthophosphate desorption occurred much more readily, with 100% desorption after 24 h at all concentrations. In the presence of 0.02 M sodium acetate, some desorption occurred after 24 h at pH 6.8 and pH 7.1 for commercial C4S at 0.2 g l31 , whilst no desorption occurred at pH 5.0. However, desorption occurred more readily at this pH after 24 h. In the presence of disodium orthophosphate, 100% desorption occurred at all concentrations and pH values after 2 h, with the exception of 0.02 g l31 C4S, pH 5.0, where notably less desorption occurred (78%). Overall, desorption occurred more readily at higher pH values and solution concentrations of C4S. 4. Discussion The results presented herein examined the adsorption of bone SLRPs and PG components (i.e., constituent GAG

0.02 M CH3 COONa

0.36 M Na2 HPO4

^ ^ ^ 16 11 ^ ^ ^ ^ 54 41 39

100 100 100 100 100 100 100 100 100 100 100 100

chains and core proteins) onto HAP. The in£uence of solution parameters such as pH and ionic strength on both adsorption and desorption processes have also been examined. Intact PGs and their constituent GAG chains showed greater adsorption onto HAP than commercial C4S, with notably higher Admax values. In addition, intact PGs showed considerably greater adsorption than the constituent core proteins, demonstrating that the £exible GAG chains with their anionic groups are signi¢cant in the adsorption process. The high a¤nity of SLRPs for HAP may be attributed to their three-dimensional structure. The central domain of the core protein is composed of approximately 10 repeats of 25 residues, featuring leucine-rich motifs [35] and similar leucine-rich repeats exist in a large number of proteins [36]. An intracellular ribonuclease inhibitor has been demonstrated to contain a leucine-rich repeat domain with a three-dimensional structure composed of alternating Khelices and L-sheets which are arranged in a coil, stabilised by lateral interactions [37]. The protein has a non-globular shape of a horseshoe and represents a new class of K/L protein folds. Scho«nherr et al. [38] and Svensson et al. [1] proposed that SLRPs may also have a similar structure and studies using rotary shadowing-electron microscopy suggested that these PGs are indeed horseshoe shaped, with the large number of tandem leucine-rich repeats forcing the two parallel lines of K-helices and L-turns into such a con¢guration [39]. This may cause greater exposure of the GAG chains, thus enhancing the binding properties of the molecule. In addition, in the intact molecule, the GAG chains and the protein core may act synergistically, exposing the anionic groups and producing a favourable conformation for adsorption. This study demonstrates that the bone GAGs play a particularly important role in the adsorption process. Although the GAG chains represent approximately 16% of the PG molecule, the Admax value obtained was comparable to that of intact PGs (0.658 þ 0.027 and 0.775 þ 0.021 mg m32 , respectively). The bone GAGs

BBAGEN 25247 6-12-01

126

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

also showed greater adsorption onto HAP than commercial C4S. The binding properties of the former may be attributed to their low molecular mass and high content of anionic groups resulting in a high charge density. The small size of the GAG chains compared with the larger, more extended conformation of commercial C4S, may result in reduced electrostatic repulsion between the molecules of the former, thus favouring adsorption. The respective GAG chains may also vary in their chemical properties as they originate from di¡erent sources, i.e., bovine trachea and ovine alveolar bone. Other studies have also examined the interaction of PGs and HAP. The CS component was proposed to account for the ability of the large cartilage PG aggrecan to inhibit extraliposomal calcium phosphate precipitation, mediated through the calcium linkage of exposed sulphate and carboxyl groups on the CS side-chains of aggrecan to growth sites on the apatite surface [40]. However, following enzymatic removal of CS, the core protein remained e¡ective in the inhibition process, suggesting that the core protein and oligosaccharides remaining attached to it have latent inhibitory properties. The probable ligands having such binding strength would either be sulphate and/or phosphate esters and an appreciable number of each of these calcium-binding ligands is associated with the core structure. Solution parameters also in£uenced the interactions of PGs/GAGs with HAP in the present study. The adsorption isotherms for PGs and commercial C4S onto HAP at varying solution pH demonstrated that adsorption increased with decreasing pH. Both C4S and PGs are negatively charged over the working pH range; C4S, consisting of repeating disaccharide units of N-acetylgalactosamine and D-glucuronic acid, may have a maximum of 4 anionic groups present per tetrasaccharide (two carboxyl and two sulphate) [41] whilst PGs have a high proportion of acidic residues associated with the core protein [25,29,42,43] as well as the negatively charged GAG chains. The HAP surface, in contrast, has a net positive charge at pH 6.8 [44] and electrostatic interactions occur between the cationic sites on HAP (calcium) and ionic domains on the macromolecule (carboxyl and sulphate). Therefore, as the solution pH is decreased, the surface will become more positively charged due to an excess of positive potential determining ions in solution, with a reduction in the activities of the negative species. The increased adsorption of C4S with decreasing pH may also be attributed to a pH-induced conformational change in the molecule, resulting in a more compact conformation, thus increased adsorption. This is supported by the fact that the rate of oxidation of C6S has been shown to decrease in acid solutions compared with neutral solutions, the resistance of certain groups to oxidation suggesting a more compact, stable structure [45,46]. In the present study, C4S exhibited a decrease in adsorption after a certain solution concentration at all pH values which could be due

to a conformational change in the molecules, where some become completely detached in favour of the `spreading' of others. Such molecules require a larger surface area at the expense of others that may be less tightly bound to the surface. This is supported by similar studies examining C4S/HAP adsorption [31,47]. Importantly, the non-Langmuir character of the adsorption isotherms in the present study may not be attributed to an insu¤cient amount of HAP and/or too low a concentration of PGs. It has previously been demonstrated that the binding of dermatan sulphate-containing decorin and biglycan from bovine articular cartilage and foetal bovine skin was Langmuir in character using as little as 0.5 mg of HAP and 2^4 mg ml31 of puri¢ed PGs [14]. Furthermore, it has previously been demonstrated that the PGs are not degraded during incubation, a factor which would otherwise a¡ect binding. The hexuronic acid pro¢les of bone SLRPs following gel ¢ltration chromatography after incubation in sodium acetate bu¡er only for 24 h at 37³C (controls) showed no evidence of degradation [25]. Furthermore, SDS^PAGE and Western blot analyses in the same study con¢rmed this ¢nding. In keeping with the present study, previous work has also shown decreasing adsorption of GAGs and proteins to HAP with increasing pH. Hughes Wassell and Embery demonstrated that the adsorption of BSA onto HAP was pH-dependent [30]. Similar trends were reported for the adsorption of phosphorylated organic compounds and BSA adsorption to HAP [48], human serum albumin adsorption on precipitated HAP [49], as well as acidic amino acid adsorption [50]. Dentine protein isolates also showed an increase in maximum adsorption with decreasing pH, which was attributed to reduced repulsion between the proteins due to protonation of the acidic groups, a factor that could play a part in pH-dependent adsorption of PGs [51]. In addition to solution pH, ionic strength was also important in the attachment of PGs/GAGs to HAP where the presence of calcium ions increased the adsorption of both intact PGs and commercial C4S. While phosphate addition makes the mineral more negatively charged, calcium makes it more positively charged under all pH conditions [44]. Calcium ions present in the medium may also reinforce the binding by enabling the formation of additional calcium bridges between sulphate and carboxyl residues on the PGs/GAGs and the phosphate sites on the HAP surface [40,52]. Conversely, the presence of phosphate ions decreased the adsorption of both intact PGs and commercial C4S. Phosphate groups are reported to have a higher a¤nity for HAP surfaces than carboxyl groups, the parameters for the adsorption of various amino acids indicating that the strength of the phosphate bond is 20 times greater [33]. A possible explanation for such strength of binding is the fact that phosphate is more hydrophobic, a factor which is more favourable for the removal of water from the mineral surface on adsorption. Therefore, the decrease in adsorption can be attributed to

BBAGEN 25247 6-12-01

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

successful competition by phosphate ions for calcium sites on the HAP surface. Previous studies have also demonstrated the ionic strength-dependent adsorption of macromolecules onto HAP. During the adsorption of BSA onto HAP both the maximum amount of protein adsorbed (AT ) and the a¤nity constant (K) increased with increasing concentrations of calcium chloride and decreased with di-sodium orthophosphate [30]. Similar trends were reported for BSA adsorption onto HAP in the presence of phosphorylated organic compounds [48] as well as amino acid adsorption onto HAP [50]. The desorption of PGs and C4S occurred more readily in the presence of di-sodium orthophosphate than sodium acetate bu¡er in the present study. The desorption studies suggest that C4S was more tightly bound at the lower concentrations (0.02 and 0.08 g l31 ), which can be attributed to the fact that at these concentrations, the molecules have su¤cient time and space to accommodate to the surface by slow, structural rearrangements. At higher concentrations (e.g., 0.2 g l31 ), collision with the surface is so high that the molecules do not have time to optimise their interaction with HAP [53]. The desorption process for C4S appears to be dependent on pH, with the strongest binding occurring at pH 5.0 due to an increase in the activities of the positive species on the mineral surface. In addition, a more compact con¢guration (as suggested by the C4S pHdependent isotherms) would make desorption more di¤cult due to tighter binding. The present study has provided important information concerning the interactions of PGs and the component GAG chains and protein cores with HAP. It has provided evidence that in the intact molecule, constituent GAG chains and protein cores may act synergistically, producing a suitable conformation for binding. Solution parameters such as pH and ionic strength have also been shown to have a signi¢cant in£uence on the process. These data also provide greater insight into the mechanisms of biomineralisation in which PGs/GAGs adsorption to the mineral surface plays an important part in the regulation of the process. Acknowledgements This work was supported by the Medical Research Council, UK. References [1] L. Svensson, D.K. Heinega®rd, A. Oldberg, J. Biol. Chem. 270 (1995) 20712^20716. [2] V.G. Vogel, J.A. Trotter, Coll. Relat. Res. 7 (1987) 105^114. [3] Y. Yamaguchi, D.M. Mann, E. Ruoslahti, Nature 346 (1990) 281^ 284. [4] D.J. Bidanset, R. LeBaron, L.C. Rosenberg, J.E. Murphy-Ulrich, M.J. Hook, Cell Biol. 118 (1992) 1523^1531.

127

[5] H.C. Whinna, H.U. Choi, L.C. Rosenberg, F.C. Church, J. Biol. Chem. 268 (1993) 3920^3924. [6] C.W. Prince, F. Rahemtulla, W.T. Butler, Biochem. J. 216 (1983) 589^595. [7] J.E. Scott, M. Haigh, Biosci. Rep. 5 (1985) 71^81. [8] L.W. Fisher, J.D. Termine, S.W.J. Dejter, S.W. Whitson, M.J. Yanagishita, J. Biol. Chem. 258 (1983) 6588^6594. [9] T. Kazama, M. Takagi, T. Ishii, Y. Toda, Histochem. J. 24 (1992) 747^755. [10] R.T. Ingram, B.L. Clarke, L.W. Fisher, L.A. Fitzpatrick, J. Bone Miner. Res. 258 (1993) 1019^1029. [11] P. Bianco, L.W. Fisher, M.F. Young, J.D. Termine, P.J. Gehron Robey, Histochem. Cytochem. 38 (1990) 1549^1563. [12] A. Bosse, H. Kresse, K. Schwartz, K.M. Mu«ller, Calcif. Tissue Int. 54 (1994) 119^124. [13] R. Fujisawa, Y. Kuboki, Biochim. Biophys. Acta 1075 (1991) 56^60. [14] A.L. Boskey, L. Spevak, S.B. Doty, L. Rosenberg, Calcif. Tissue Int. 61 (1997) 298^305. [15] A.L. Boskey, Clin. Orthop. Rel. Res. 281 (1992) 244^273. [16] G. Bernadi, M.G. Giro, C. Gaillard, Biochim. Biophys. Acta 278 (1972) 409^420. [17] S. Chander, D.W. Fuerstenau, in: D.N. Misra (Ed.), Adsorption On and Surface Chemistry of Hydroxyapatite, Plenum Press, New York, 1984 pp. 29^49. [18] A.L. Kamashyni, Russ. J. Phys. Chem. 55 (1981) 319^330. [19] M.J. Gorbuno¡, S.N. Timashe¡, Anal. Biochem. 136 (1984) 440^445. [20] C.A. van Blitterswijk, J.J. Grote, W. Kuijpers, C.J.G. Blok-van Hoek, W.T. Deams, Biomaterials 6 (1985) 243^251. [21] K.E. Healy, P. Ducheyne, Biomaterials 13 (1992) 553^561. [22] C.B. Johnsson, H.A. Hansson, T. Albrektsson, Biomaterials 11 (1990) 277^280. [23] L.W. Fisher, J.D. Termine, M.F. Young, J. Biol. Chem. 264 (1989) 4571^4576. [24] R.J. Waddington, G. Embery, K.S. Last, Arch. Oral Biol. 34 (1989) 587^589. [25] R. Moseley, R.J. Waddington, G. Embery, S.G. Rees, Connect. Tissue Res. 37 (1998) 13^28. [26] R.W. Farndale, D.J. Buttle, A.J. Barrett, Biochim. Biophys. Acta 883 (1986) 173^177. [27] D. Bidanset, C. Guidry, L. Rosenberg, H.J. Choi, J. Biol. Chem. 267 (1992) 5250^5256. [28] P. Roughley, L. Melching, A. Recklies, Matrix Biol. 14 (1994) 51^ 59. [29] R.J. Waddington, G. Embery, Arch. Oral Biol. 36 (1991) 859^866. [30] D.T. Hughes Wassel, R.C. Hall, G. Embery, Biomaterials 16 (1995) 697^702. [31] S.G. Rees, D.T. HughesWassell, G. Embery, Biomaterials 23 (2002) 481^489. [32] O.H. Lowry, N.J. Rosenberg, A.L. Farr, R.J. Randall, J. Biol. Chem. 193 (1951) 265^275. [33] E.C. Moreno, M. Kresak, D.I. Hay, Calcif. Tissue Int. 36 (1984) 48^ 59. [34] P. Ducheyne, C.S. Kim, S.R. Pollack, J. Biomed. Mater. Res. 26 (1992) 147^168. [35] R.V. Iozzo, A.D. Murdoch, FASEB J. 10 (1996) 598^614. [36] T. Krusius, E. Ruoslahti, Proc. Natl. Acad. Sci. USA 83 (1986) 7683^ 7687. [37] B. Kobe, J. Deisenhofer, Trends Biochem. Sci. 19 (1994) 415^421. [38] E. Scho«nherr, P. Witsh-Prehm, B. Harrach, H. Robenek, J. Rauterberg, H. Kresse, J. Biol. Chem. 270 (1995) 2776^2783. [39] J.E. Scott, Biochemistry 35 (1996) 8795^8799. [40] E.D. Eanes, A.W. Hailer, R.J. Midura, V.C. Hascall, Glycobiology 2 (1992) 571^578. [41] U. Lindahl, M. Ho«o«k, Annu. Rev. Biochem. 47 (1978) 385^417. [42] L.W. Fisher, J.D. Termine, Clin. Orthop. Rel. Res. 200 (1985) 362^ 385. [43] P.M. Bartold, J. Dent. Res. 69 (1990) 7^19.

BBAGEN 25247 6-12-01

128

S.G. Rees et al. / Biochimica et Biophysica Acta 1568 (2001) 118^128

[44] P. Somasundaran, Y.H.C. Wang, in: D.N. Misra (Ed.), Adsorption On and Surface Chemistry of Hydroxyapatite, Plenum Press, New York, 1984, pp. 129^149. [45] L.A. Fra®nsson, Carbohdr. Res. 36 (1974) 339^345. [46] N. Di Ferrante, P.V. Donnelly, R.K. Berglund, Biochem. J. 124 (1971) 549^553. [47] D.T. Hughes Wassell, G. Embery, Biomaterials 18 (1997) 1001^1007. [48] S. Shimabayashi, Y. Tanizawa, K. Ishida, Chem. Pharm. Bull. 39 (1991) 2183^2188.

[49] V. Hlady, H. Furedi-Milhofer, J. Colloid Interface Sci. 69 (1979) 460^468. [50] H. Tanaka, K. Miyajima, M. Nakagaki, S. Shimabayashi, Chem. Pharm. Bull. 37 (1989) 2897^2901. [51] W.D. Boonstra, J.J. Ten Bosch, J. Arends, J. Biol. Buccale 20 (1992) 111^116. [52] E.A. MacGregor, J.M. Bowness, Can. J. Biochem. 49 (1971) 417^425. [53] J.D. Andrade, V. Hlady, A.P. Wei, C.G. Golander, Croatica Chim. Acta 3 (1990) 527^538.

BBAGEN 25247 6-12-01