Changes to the Disordered Phase and Apatite ... - ACS Publications

Jul 24, 2015 - from the bone/dentin protein osteonectin. The mineral formed comprises needle-shaped hydroxyapatite crystals like in dentin and a stabl...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/Biomac

Changes to the Disordered Phase and Apatite Crystallite Morphology during Mineralization by an Acidic Mineral Binding Peptide from Osteonectin Taly Iline-Vul,‡ Irina Matlahov,‡ Judith Grinblat, Keren Keinan-Adamsky, and Gil Goobes* Department of Chemistry, Bar Ilan University, Ramat Gan 52900, Israel S Supporting Information *

ABSTRACT: Noncollagenous proteins regulate the formation of the mineral constituent in hard tissue. The mineral formed contains apatite crystals coated by a functional disordered calcium phosphate phase. Although the crystalline phase of bone mineral was extensively investigated, little is known about the disordered layer’s composition and structure, and less is known regarding the function of noncollagenous proteins in the context of this layer. In the current study, apatite was prepared with an acidic peptide (ON29) derived from the bone/dentin protein osteonectin. The mineral formed comprises needle-shaped hydroxyapatite crystals like in dentin and a stable disordered phase coating the apatitic crystals as shown using X-ray diffraction, transmission electron microscopy, and solid-state NMR techniques. The peptide, embedded between the mineral particles, reduces the overall phosphate content in the mineral formed as inferred from inductively coupled plasma and elemental analysis results. Magnetization transfers between disordered phase species and apatitic phase species are observed for the first time using 2D 1H−31P heteronuclear correlation NMR measurements. The dynamics of phosphate magnetization transfers reveal that ON29 decreases significantly the amount of water molecules in the disordered phase and increases slightly their content at the ordered-disordered interface. The peptide decreases hydroxyl to disordered phosphate transfers within the surface layer but does not influence transfer within the bulk crystalline mineral. Overall, these results indicate that control of crystallite morphology and properties of the inorganic component in hard tissue by biomolecules is more involved than just direct interaction between protein functional groups and mineral crystal faces. Subtler mechanisms such as modulation of the disordered phase composition and structural changes at the ordered−disordered interface may be involved.



INTRODUCTION Calcium phosphate mineralization in tissue such as bone and dentin is an intriguing process that is regulated down to the molecular level.1,2 In recent years, high-resolution information on bone and dentin structure was gained through the combination of advanced analysis techniques. Assembled collagen with the additive polyaspartate was shown to form an active scaffold for mineralization in vitro promoting bone-like apatite precipitation.3 Later, it was shown that mineralization occurs in collagen fibrils without additives,4 leaving the question regarding the role of noncollagenous proteins (NCPs) open. The mineral of bone, known to contain platelet shaped apatite crystals was reported to have lower content of hydroxyl groups than stoichiometric calcium hydroxyapatite (HAP).5−7 Individual platelets were shown to be coated by a disordered calcium phosphate phase8−11 containing water molecules that orient the platelets.12 This phase was also shown, using solid-state NMR, to contain citrate molecules, which bind to the apatite surface and bridge different platelets.13−15 © XXXX American Chemical Society

Dentin is characterized by a similar organic/inorganic ratio to bone.16 It has some NCPs in common with bone, for example, osteonectin (ON), osteocalcin, and others that are unique to the dental tissue.17−20 However, collagen fibrils assemble into a different scaffold in dentin subsequently influencing the final structure of the mineral precipitated.21 Electron microscopy (EM) studies have shown that in dentin, unlike in bone, apatite crystals grow in thin elongated (needle-like) morphology.22 It was also shown that in dentin tissue22 and in in vitro models,23 mineral transformation from octacalcium phosphate (OCP) to HAP may take place. However, fewer details are known regarding the presence of water or other molecules at the interface with the mineral and about their role in its organization. Mineral formation and homeostasis are performed by bone and dentin cells through the function of NCPs. These processes Received: April 11, 2015 Revised: June 26, 2015

A

DOI: 10.1021/acs.biomac.5b00465 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

findings to mechanisms of mineral regulation in hard tissue are discussed.

are thought to be regulated both spatially24−31 and temporally32 on a molecular level by the NCPs. ON is a noncollagenous matrix protein highly abundant in bone tissue.33−36 ON stimulates cell proliferation and diminishes focal cell adhesions17,37,38 and is a potent cytokine, found in other tissues in the body. In bone, it is implicated with the important function of mediating cell and extracellular matrix interactions in the process of mineralization. It exhibits affinity to HAP, collagen, and Ca2+ ions.39−42 Although ON is also associated with recruitment of cells in dentin just like in bone, it is, nevertheless, beneficial to understand the effect of its mineral binding domain on the final shape of apatite crystallite in relation to the disparate morphology in different mineralized tissues. The existence of a disordered phase coating the crystalline apatite phase in bone raises some new questions regarding the means by which the NCPs regulate mineralization. It is not known yet whether they affect crystallization directly or through manipulating the structure or composition of the disordered phase. It is also unclear whether they can operate within the confines of the assembled collagen matrix or only regulate extra-fibrillar mineralization. Moreover, recent results have suggested that interfaces between inorganic mineral phosphates and organic molecules do not include NCPs at all,39 underscoring the need for high-resolution studies examining NCPs’ impact on the mineral formed and remodeled.40,42 Such studies can also assist in design of improved biomimetic bone replacement materials and biocompatible prosthetics.43 Hydroxyapatite precipitated with an acidic salivary peptide and studied using solid-state NMR provided viable molecular information regarding the peptide’s interaction with different surfaces of the mineral. 44 Studies of ON’s effect on hydroxyapatite crystallization have shown retardation of crystal nucleation45 and growth,46 whereas attachment of the protein to collagen could either inhibit47 or promote48 mineral formation. Inside collagen−ON complexes, hydroxyapatite crystallizes as nanosized spindles with variable length.49 However, the effects of the protein on the disordered phase forming on the crystals of apatite and its implication to the crystallization process have not been investigated and can provide further evidence for the activity of NCPs in organic− inorganic interfaces in hard tissue. Revealing the molecular structure of these interfaces can shed light on the role of NCPs in the formation of inorganic structures in hard tissue. In the present study, apatite mineral precipitation was performed in the presence of a 29-residue peptide (ON29) derived from ON’s mineral binding domain. ON29 contains the highest density of glutamate residues in the full mineral binding domain APQQEALPDE TEVVEETVAE VTEVSVGANP VQVEVGEFDD GAEETEEEVV. The inorganic solid was precipitated at 60 °C (synthesis route 1), reported to improve crystallization, and at body temperature 37 °C (synthesis route 2) under controlled pH, temperature, and rate of reactant addition.50,51 The crystalline phase and the disordered hydrated surface phase formed with and without ON29 in the precipitate were characterized by EM, X-ray diffraction (XRD), Brunauer− Emmett−Teller (BET) isotherm, inductively coupled plasma (ICP), elemental analysis, and magic-angle spinning (MAS) NMR experiments. From these measurements, it is shown that the peptide modifies the content of the disordered phase and induces growth of needle-like crystals. The implications of these



EXPERIMENTAL SECTION

Materials. Diammonium hydrogen phosphate (NH4)2HPO4 99% and calcium nitrate tetrahydrate Ca(NO3)2·4H2O 99% were purchased from Merck and Sigma-Aldrich, respectively. Ammonium hydroxide NH4OH was purchased from BioLab Ltd. Resin and amino acids used in the synthesis of ON29 were purchased from GL Biochem (Shanghai) Ltd. Methods. Synthesis and Purification of ON29. An ON peptide (ON29) derived from the N-terminus apatite-binding domain of the protein, residues T21−A50, was prepared. The peptide’s sequence, with a deletion of V49, is as follows: TEVSVGANPVQVEVGEFDDGAEETEEEVA. It has a molecular weight of 3.04 kDa and a calculated pI of 3.21. The peptide was synthesized in 100 μmol quantities using standard solid phase peptide synthesis (SPPS) techniques based on 9fluorenylmethoxycarbonyl (FMOC) chemistry on Fmoc-Ala-Wang resin. The product was purified on a Waters HPLC using a VYDAC C18 reverse-phase column for separation based on the hydrophobic index of the products and on an Acta Explorer FPLC using GE HP and FF Q-sepharose anion exchange columns due to high density of charged residues. Mass spectrometry analysis of crude and purified product composition on a matrix-assisted laser desorption/ionization (MALDI) mass spectrometer equipped with an Autoflex III smartbeam by Bruker confirmed the identity of the peptide. Preparation of Apatite in the Presence of ON29. Apatite was precipitated in the presence of ON29 by titrating 16 mL of 0.075 M (NH4)2HPO4 into 56 mL of 0.040 M calcium nitrate tetrahydrate and 0.2 M (0.8%) ammonium hydroxide solution in which 29.6 mg of peptide were dissolved. Synthesis without ON29 followed identical conditions simply without the peptide in the Ca(NO3)2-NH4OH solution. All syntheses were carried out under air. The titration was performed over a duration of 5 h at a rate of 0.053 mL/min using a peristaltic pump while the mixture was stirred at a rate of 282 rpm. Two syntheses were performed with temperature and pH held constant at: Route 1, 60 °C and pH 8.5; Route 2, 37 °C and pH 9.0. Conditions were corrected every 5 min by adding small aliquots of 24% ammonium hydroxide solution as necessary. For both syntheses, after the phosphate solution was added, the temperature was set to 60 °C and the mixture stirred overnight. The solid precipitate was filtered and rinsed with double distilled water, 18.2 MΩ (DDW). Excess water was removed from sediment by leaving the samples at 37 °C for 60 h. Precipitated products were pulverized in a porcelain mortar and stored as is. Content of ON29 in ON29·HAP prepared in route 2 was 1.30(±0.02) mol % and in ON29·HAP prepared in route 1 was negligible as inferred from thermogravimetric analysis (TGA) results (data not shown). The hydrated solid peptide−mineral complexes were stable for periods of months. Reference mineral (without peptide) was prepared under the same conditions in the two routes. Part of the mineral prepared by route 1 was calcined at 900 °C for 24 h and used for comparative analysis. Powder X-ray Diffraction. The crystallographic phase analysis was carried out recording powder XRD data on a Bruker AXS D8 Advance diffractometer. The experiment was performed within a 2θ range of 5°−80° with Cu Kα radiation (1.54 Å). All measurements were carried out using the same programs, similar quantities of mineral, and recorded for 0.05 s. Inductively Coupled Plasma Analysis. To quantify Ca and P elemental content in the materials, ICP analysis was performed. For each measurement, 6.9 mg of product powder was suspended in 10 mL of 0.1 M HCl. ICP was carried out using ULTIMA2 device from Jovin-Yvon-Horiba. Errors in Ca/P ratio were determined based on the relative error in Ca and P measurements. Relative error in the values of the two elements is between 0.35% and 1.25%. ICP measurements of apatite mineral precipitated with other NCPs such as B

DOI: 10.1021/acs.biomac.5b00465 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules osteocalcin were done resulting in Ca/P ratios lower than 1.6 (data not shown) to verify that there is no bias the device used for the measurement in this work. Elemental Analysis. Elemental analysis was performed using CHNS Analyzer, FlashEA 1112 Series by Thermo. The analysis was performed using 8 mg of sample. Brunauer−Emmett−Teller Isotherm. Surface area of 65 mg of HAP and HAP·ON29 prepared in each of the routes was determined using NOVA 3200E surface area and pore size analyzer from Quantachrome Instruments. Samples were measured under heating at 120 °C following 60 min of preheating at this temperature. Transmission and Scanning Electron Microscopy (TEM and SEM). Samples were suspended in ethanol, sonicated for 10 min, and placed on copper grids for TEM until complete solvent evaporation was achieved. High-resolution TEM measurements were carried out using JEM 2100, JEOL instrument with LaB6 at accelerating voltage of 200 kV. Samples in water were deposited on a double-sided 12 mm thick carbon tape for HRSEM. Morphology and topography analyses were carried out on a FEI, Magellan 400L high-resolution SEM instrument. NMR Experiments. HAP·ON29 and HAP samples were packed into 4 mm MAS NMR rotors. Solid-state NMR experiments were carried out at room temperature on an 11.74T Bruker AvanceIII spectrometer equipped with a 4 mm MAS VTN probe. 31P direct polarization (DP) experiments were performed using spinning rate of 10 kHz, 3.9 μs 31P 90° pulse, and recycle delay of 10 s. 31P CPMAS experiments were carried out using a spinning rate of 10 kHz, a 2.5 μs 1H 90° pulse, 3 ms contact time, SPINAL64 pulse scheme for proton decoupling,52,53 and recycle delay of 3 s. Slow spinning 31P CPMAS experiments were performed at a spinning rate of 2 kHz to extract CSA parameters. 1 H−31P heteronuclear correlation (HETCOR) experiments were done using spinning rate of 10 kHz, a 2.55 μs 1H 90° pulse, CP contact time of 0.8 ms, and the PMLG5 scheme for 1H homonuclear decoupling at an effective field of 120 kHz during t1.54,55 The two-pulse phase modulation (TPPM) pulse scheme was used for proton decoupling, and the field used was 98 kHz.56 All HETCOR experiments employed a recycle delay of 1 s. 31P chemical shifts were calibrated relative to H3PO4 (85%). Signal to noise values in 2D HETCOR experiments were used as the basis for error bar calculations in CP build-up curves. NMR Simulations and Data Fitting. Chemical shift anisotropy and asymmetry of 14.0 and 1.0, respectively, were deduced from slow spinning 31P cross-polarization (CP)MAS measurements by fitting the sideband patterns to simulated patterns obtained using SIMPSON.57 These chemical shift parameters are similar to values measured for HAP before.57 Deconvolutions of 31P lines were carried out using the DMFIT program developed by Massiot and co-workers.58 CP buildup curves were fitted and kinetic parameters were extracted by minimizing simulated curves against experimental data using a MATLAB’s code employing the numerical minimization function “fminsearch”.

Figure 1. Powder XRD patterns of HAP (orange) and HAP·ON29 (red) made by route 1 and HAP (blue) and HAP·ON29 (pink) prepared in route 2. Also shown is the pattern of HAP made in route 1 and calcined at 900 °C (green).

the standard pattern of hexagonal HAP crystals.59 The diffractograms of HAP_1 and HAP·ON29_1 exhibit broad reflections with peak positions and relative intensities that match calcined HAP_1. The wide peaks reflect small crystallites and possibly some lattice imperfections. In HAP·ON29_1 (red), sharper peaks are observed indicating formation of larger crystallites. XRD patterns for HAP_2 and HAP·ON29_2 are almost identical having broader reflections and lower S/N ratio compared to route 1. The broader peaks indicate further smaller crystallites in the samples made at body temperature. XRD data of nanocrystalline HAP were previously shown to have a similar diffraction pattern.60 For HAP grown in the presence of the full ON at 25 °C, a much broader XRD diffractogram was reported.49 HRTEM images of HAP_1, HAP·ON29_1, HAP_2, and HAP·ON29_2 are shown in Figure 2, panels a−d, respectively. The morphology of the crystallites in HAP_1 and HAP_2 is platelet-like with average plate sizes of 30−40 nm and 10−20 nm, respectively. With ON29, elongated needle-like crystals are formed with average size of 45 nm × 9 nm for HAP_1 and 60 nm × 5 nm for HAP_2, as further demonstrated in Figures S1− S3 in the Supporting Information. For HAP·ON29_1, order is higher, and crystal planes are clearly observed in the micrograph. Under conditions of route 2, the peptide’s effect on crystal morphology is slightly modified, producing thinner and longer nanometer-sized needles on average. HRSEM measurements of HAP prepared in the two routes (see Figure S4 in the Supporting Information) show more elongated in HAP·ON29_2 than in HAP_2 in consistency with the TEM images. Needle-shaped apatite crystals are found in dentin and in enamel as basic building blocks of the mineral phase.22 Nitrogen adsorption measurements (BET method) carried out on the four samples are summarized in Table 1. Specific surface areas (SSAs) of samples in all preparations are relatively high (>110 m2/g) and resemble values obtained in previous preparations of the mineral with other mineral binding peptides.44 For ON29, these values remain persistent over



RESULTS The peptide-apatite products from mineral precipitation with the ON29 peptide in route 1 and in route 2 syntheses are hereafter referred to as HAP·ON29_1 and HAP·ON29_2, respectively. The reference minerals prepared without ON29 in route 1 and route 2 are accordingly termed here HAP_1 and HAP_2. All four samples were characterized by powder XRD, EM, surface characterization, and composition analysis techniques. 1D 31P NMR and 2D 1H−31P HETCOR NMR measurements on ON29 and HAP·ON29 were used to investigate the disordered and crystalline phases in the mineral formed. Powder X-ray diffractograms of HAP_1 (orange) and HAP· ON29_1 (red) and diffractograms of HAP_2 (blue) and HAP· ON29_2 (pink) are shown in Figure 1. For comparison, a HAP_1 sample, calcined at 900 °C for 24 h, is also shown (green). The high-temperature annealing produced a highly crystalline mineral with sharp reflections that are identical to C

DOI: 10.1021/acs.biomac.5b00465 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

against commercial nanoHAP from Sigma-Aldrich, which gave a value of 1.66(±0.01) in accordance with the stoichiometric value for hydroxyapatite (1.67). For HAP_1 and HAP_2, respective values of 1.62 and 1.76 were obtained, indicating 3% higher and 5% lower calcium content than in stoichiometric hydroxyapatite. HAP·ON29 samples, made by both routes, exhibit Ca/P ratios that are higher than the stoichiometric value. These results indicate an overall depletion of phosphates from the mineral prepared with ON29. The content changes in the basic mineral ions cannot be explained by the presence of a different crystal phase as XRD data show and since known calcium phosphate crystalline phases have lower Ca/P stoichiometries.57,60,61 It is plausible, then, to relate phosphate deficiency to shuffled ion composition in the surface layer. A disordered coat with a markedly higher Ca/P ratio can explain the overall changes observed. For HAP and HAP·ON29 made by route 2, for example, surface layers that constitute 33% of the total mineral content with respective Ca/P ratios of 1.95 and 2.1 (meaning a phosphate ion for every two Ca ions) are adequate to explain the ratio measured. The higher depletion of phosphates in HAP·ON29_2 versus HAP·ON29_1 can be linked to higher peptide content in the former, as seen in the Elemental Analysis. At body temperature and slightly increased basic conditions, more negatively charged ON29 molecules are found in the surface layer lowering further the phosphate concentration. Since neither the relative mineral content nor the Ca/P ratio in the surface layer are known, it is not possible to delineate the exact partitioning between the two phases. Evidence for phosphate depletion is rare and was reported only once, to the best of our knowledge, by George and coworkers in preparation of apatite with a dentin NCP and collagen.62 Most preparations of bone or dental mineral with other NCPs have reported Ca/P ratios lower than 1.67, that is, calcium deficiency as well as ex vivo measurements of bone. This is not in contradiction to current results since ON may not be the most abundant NCP in bone, and different NCPs have disparate effects on Ca/P ratios. In fact, the relative concentration of different NCPs in the hard tissue is modulated at different mineralization stages and can be used to modulate ion content and structure of the ordered and disordered phases of the mineral. 31 P DP MAS NMR measurements of the samples are shown in Figure S5 in the Supporting Information. Spectra of HAP· ON29_1 and HAP·ON29_2 exhibit a broad resonance at 2.8 ppm (fwhm 1.4 ppm) and 2.9 ppm (fwhm 2.5 ppm), respectively, related to the mineral phosphates. In similarity to reflections in the powder X-ray diffractograms, the phosphate line in HAP·ON29_2 is much broader than in HAP·ON29_1. The 31P spectrum of calcined HAP (also shown in the figure) exhibits a markedly narrower peak at 2.9 ppm (fwhm 1.0 ppm) reflecting the high order of phosphates in the crystals after calcination and removal of any disordered surface layer. 31 P CPMAS spectra of HAP·ON29_1 and HAP·ON29_2, also shown in Figure S5 in the Supporting Information, exhibit a broad resonance at similar chemical shifts, with line widths of 1.3 and 2.8 ppm (fwhm), respectively. The trends in-line widths are similar to the ones observed in the DP spectra, whereby HAP·ON29_2 with its higher disorder exhibits broader phosphate line than HAP·ON29_1. 31P chemical shift anisotropy measured on HAP and HAP·ON29 at low spinning gives anisotropy and asymmetry values of 14.0 ppm and 1.0,

Figure 2. Transmission electron micrographs of (a) HAP prepared in route 1, (b) HAP·ON29 prepared in route 1, (c) HAP prepared in route 2, and (d) HAP·ON29 prepared in route 2. Scale bars (white) in insets are 5 nm each.

Table 1. BET Measurements of HAP and HAP·ON29 in Two Syntheses

HAP HAP·ON29

surface area

(m2/g)

Synthesis 1

Synthesis 2

145(±14) 111(±11)

163(±16) 170(±17)

months reflecting the stabilizing effect of peptide molecules. In HAP_1, the SSA is slightly lower than that of HAP_2. In the presence of ON29, the SSA drops by 30% in route 1 and increases negligibly in route 2. HAP·ON29_2 shows a 1.5-times higher SSA than that of HAP·ON29_1. This result is consistent with its larger aspect ratio crystallites, as determined from the HRTEM micrographs, and its smaller size crystallite, as seen in the broader XRD pattern. Similarity of SSA for HAP_1 and HAP·ON29_1 is also in accordance with the HRTEM and XRD results, though some aggregation may be induced by water loss as the measurement is done at 120 °C. It is noted that the differences in SSA may also be affected by adsorbent properties of the disordered surface phase on apatite crystals. Elemental analysis (Table S1 in Supporting Information) of the HAP·ON29 products shows very low (