Biomimetically Triggered Inorganic Crystal Transformation by

10838

J. Phys. Chem. B 2009, 113, 10838–10844

Biomimetically Triggered Inorganic Crystal Transformation by Biomolecules: A New Understanding of Biomineralization Wenge Jiang, Xiaobin Chu, Ben Wang, Haihua Pan, Xurong Xu, and Ruikang Tang* Center for Biomaterials and Biopathways and Department of Chemistry, Zhejiang UniVersity, Hangzhou, Zhejiang, 310027, China ReceiVed: January 7, 2009

Phase transformation is an important strategy in biomineralization. However, the role of biomolecules in the mineral transition is poorly understood despite the fact that the biomineralization society greatly highlights the organic controls in the formation of the inorganic phase. Here, we report an induced biomimetic phase transformation from brushite (a widely used calcium phosphate precursor in biological cement) to hydroxyapatite (main inorganic composition of skeletal mineral) by citrate (a rich organic component in bone tissue). The transformation in the absence of the organic additive cannot be spontaneously initiated in an aqueous solution with a pH of 8.45 (no phase transition is detected in 4 days), which is explained by a high interfacial energy barrier between brushite-solution and hydroxyapatite-solution interfaces. Citrate can oppositely regulate these two interfaces, which decreases and increases the stabilities of brushite and hydroxyapatite surfaces in the solution, respectively. Thus, the interfacial energy barrier can be greatly reduced in the presence of citrate and the reaction is triggered; e.g., at 1 mM citrate, the total transformation from brushite to hydroxyapatite can be completed within 3 days. The relationship between the transition kinetics and citrate concentration is also studied. The work reveals how the organic components direct solid-solid phase transformation, which can be understood by an energetic control of the interfacial barrier. It is emphasized that the terms of interfacial energy must be taken into account in the studies of phase transformation. We suggest that this biomimetic approach may provide an in-depth understanding of biomineralization. Introduction Biomineralization is a biological process by which living organisms make use of peptides and proteins to control the formation of functional minerals.1,2 The direct solution crystallization is generally used as a simulation model in laboratories during the previous explorations.2-4 It has been concluded that a constructive interaction of specific faces of inorganic crystals and organic matrixes determines the kinetics of nucleation and crystal growth, leading to a modulation of crystal size, morphology, and orientation.5-7 However, such a biomimetic study based upon the conventional solution model is critically challenged by the latest developments of biomineralization.8-10 It has been observed that mineral crystallites such as hydroxyapatite (HAP) and calcite are not one-step-precipitated in the biological generation of skeletal minerals and mollusk shells.10-12 There is increasing evidence that these biominerals are transformed from their precursor/interim solid states in living organisms.13 Recently, phase transformation is proposed as the most important pathway in biomineralization.10,11 Unfortunately, the mechanism of crystal transition in biomineralization is poorly understood.9,14-16 Although the enrollment of organic matrix is highlighted as a characteristic of biomineralization, scientists have not identified how biomolecules mediate phase transformations of inorganic crystals. The mineralization Via intermediate states involves a minimum energy barrier, which is significantly lower than that for the direct crystallization.17 This feature is highly advantageous in biomineralization to produce the hierarchically higher-order mineral structure under the mild condi* Corresponding author. E-mail: [email protected]. Phone/Fax: +86-57187953736.

tions,18 e.g., the formation of seven hierarchical levels of the organizations of apatite and collagen composites in bone.19,20 The in ViVo transformation from amorphous calcium phosphate (ACP), brushite (dicalcium phosphate dehydrate, DCPD), tricalcium phosphate (TCP), and octacalcium phosphate (OCP) to HAP has been widely observed.13,21 Due to the precursor effect, these calcium phosphate interim phases are frequently used in biomedical cements to repair biological hard tissue such as bone.22,23 In order to improve the bone regeneration, organic additives are additionally applied in association with the inorganic components, in which the promotion effect of citrate has been confirmed clinically.24,25 It is suggested that the organic matrix may selectively adsorb on precursor crystals to regulate crystal morphology and orientation.26 However, such a discussion is insufficient for an in-depth understanding of the reaction mechanism. For example, there are two contrary effects of biomolecules on mineralization kinetics: inducing nucleation but inhibiting crystal growth.27,28 Both cannot be used to explain the kinetic controls of phase transformation in biomineralization. We reveal a trigger and promotion effect of biomolecules on crystal phase transition. Different from the previous opinions, the transformation from the metastable phase into the stable one is not always spontaneous. We found that this phenomenon can be interpreted by an interfacial energy barrier. The presence of organic components can greatly reduce this barrier. The current study highlights the organic-induced modification of the interfacial energy barrier during the inorganic solid phase transformation. This idea can provide a new strategy of biological controls in biomineralization. In this paper, the biomimetic evolutions from DCPD to HAP in the absence and presence of citrate are studied. In ViVo

10.1021/jp904633f CCC: $40.75  2009 American Chemical Society Published on Web 07/10/2009

Biomimetically Triggered Mineral Transformation

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10839

observation has confirmed the precursor effect of DCPD in the HAP formation.13 It is also important that DCPD has a smooth (010) surface which allows the in situ investigation using atomic force microscopy (AFM).29 It is generally agreed that the proteins most active in mediating mineralization often contain acidic amino acid residues, specifically carboxylic-acid-rich regions.30,31 Osteocalcin, the effective protein in formation of bone mineral, contains three adjacent Gla residues which regulate the birth and growth of hydroxyapatite in bone.26,32 Also, the carboxyl groups of Gla residues share the similar chemical structures with citrate. Actually, citrate is frequently detected in biological organic-mineral composites and its content is relatively high in bone.33 As we mentioned above, citrate has been widely used as a positive factor in clinical treatments of bone.25 We expect the selected DCPD-HAP-citrate model can hold for a fundamental understanding of both bone generation and biomineralization.

TABLE 1: Surface Tension Components Used in the Calculations of Interfacial Energy between Calcium Phosphate Solids and Reaction Solutionsa

Experimental Section

where the subscripts S and L represent the solid surface and test liquids, respectively. γ+ is the Lewis acid (electron-acceptor) parameter and γ- the Lewis base (electron-donor) parameter. The interfacial energy between the crystal and solutions, γSL, was then calculated using eq 235

Materials. DCPD crystals were synthesized in our laboratory according to the Marshall method.34 Sodium citrate (g99%) was purchased from Sigma-Aldrich. Triple-distilled water was used in the experiments. Crystal Transition. The phase transformations from DCPD to HAP were performed in a solution containing 75 mM sodium chloride and 10 mM sodium phosphate at room temperature. Different citrate concentrations (0-5 mM) were applied in the study. The pH value of the solutions was adjusted to 8.45 ( 0.01 using 0.01 M sodium hydroxide or hydrochloric acid. DCPD seeds were added into the reaction solution at a finial concentration in the reaction solution of 0.5 g/L. The suspension was magnetically stirred at 300 rpm. The solids during the transition processes were withdrawn periodically by filtration, washed by water, and dried in a vacuum. Characterizations. The solid phase was measured by X-ray diffraction (XRD, D8-Advance, Bruker, Germany) with Cu KR radiation (λ ) 1.540 Å, Ni filter) to follow the transformation. In the scanning electron microscopy (SEM) studies, the solid samples, under vacuum, were sputter-coated by platinum and then examined by a field-emission SEM (S-4000, Hitachi, Japan), typically at 20 or 30 keV. Transmission electron microscopy (TEM) was performed by using a JEM-200CX instrument (JEOL, Japan). The acceleration voltage in the selected area electron diffraction (SAED) study was 160 kV. AFM. Images were collected using a tapping mode using a DI Nano IVa instrument (Veeco, Santa Barbara, CA). The tip-surface interaction was reduced by using the lowest tip force to ensure that the images were authentic representations of the surfaces. We measured adhesion forces by recording the cantilever deflection in the reaction solution. The data were obtained from >200 individual force curves at >50 different points on the faces. Contact Angle. The sessile drop method was used for the contact angle measurements with an OCA15+ instrument (Dataphysics, Germany). The interfacial energy, γ, was calculated by Young’s equation using the parameters of solids and liquids. The unknown components of the calcium phosphates and the reaction solutions were obtained when the Young equation was solved by using the known parameters of the probing liquids and the probing substrates (Table 1) and the measured contact angle, θ35

(1 + cos θ)γL ) 2(√γSLWγLLW + √γS+γL- + √γS-γL+)

(1)

probing liquid/solid

γ

γLW

γ+

γ-

octane R-bromonaphthalene diiodomethane formamide glycerol ethylene glycol water silicon tetrafluoroethylene methyl methacrylate vinyl chloride

21.62 44.4 50.8 58.0 64.0 48.0 72.8 55.62 19.60 43.20 43.79

21.62 44.4 50.8 39.0 34.0 29.0 21.8 38.07 19.60 43.20 43.0

0 0 0 2.28 3.92 1.92 25.5 3.67 0 0 0.07

0 0 0 39.6 57.4 47.0 25.5 20.98 0 14.24 2.21

a

The parameters (mJ m-2) are cited from refs 36-38.

γSL ) (√γSLW - √γLLW)2 + 2(√γS+γS- + √γL+γL- -

√γS+γL- - √γS-γL+)

(2)

Results and Discussion It is well-known that DCPD is stable under acidic aqueous conditions. When DCPD is exposed to a basic solution, HAP should be preferentially transformed, as HAP is the most thermodynamically stable one among the calcium phosphates. Actually, the crystal transformation from DCPD to HAP was extremely slow without any organic additive. In a solution of pH 8.45 ( 0.01 (75 mM sodium chloride was used to adjust the ionic strength; 10 mM sodium phosphate was added to provide a buffer effect), the XRD result showed that HAP could not be detected even at 96 h (Figure 1a) and all of the solids in the solution were still pure DCPD. Figure 1b showed a SEM image of the original DCPD crystallites. After 96 h exposure to the basic solution, only partial dissolution of the crystals could be observed on their edges and the characteristic DCPD morphology remained (Figure 1c). These repeatable phenomena implied that the phase transformation did not occur despite the fact that such a transition was thermodynamically preferred. We suggested that this transformation was kinetically suppressed. A previous study also reported that the spontaneous phase transformation form DCPD to HAP could not be observed in the absence of protein.26 Therefore, the transformation of DCPD to HAP was not readily the expectation. In contrast, the in ViVo transformation from DCPD to HAP is easy under biological conditions, which makes DCPD a perfect hard tissue replacement material.39,40 We noted that the biological milieu contains abundant organic compounds such as proteins. When 1 mM citrate was added into the reaction solution (the pH was kept at 8.45 ( 0.01), the signal of HAP appeared in the XRD spectrum within 14 h (Figure 1a). SEM also confirmed the dramatic influence of citrate on the transformation from DCPD to HAP. Numerous nano flake-like crystals were newly formed on the original crystal surfaces (Figure 1d). The core-shell structure of the DCPD-HAP complex could be seen under TEM: a layer of HAP tiny crystallites covered onto the DCPD mother crystal (Figure 2). Such a coexistence of DCPD

10840

J. Phys. Chem. B, Vol. 113, No. 31, 2009

Jiang et al.

Figure 1. (a) XRD pattern of the solid samples at different experimental stages. Bottom: standard pattern of DCPD from JCPDS 09-0077. In the absence of citrate, the solid was still DCPD without phase transformation. In the presence of 1 mM citrate, the formed HAP could be detected at ∼14 h and the complete transformation from DCPD to HAP was achieved within ∼72 h. The peaks of HAP were marked with an asterisk. SEM images of (b) original DCPD, (c) crystallite in a citrate-free solution at 96 h, (d) transforming crystallite in 1 mM citrate at 14 h, and (e) transformed HAP crystallite in 1 mM citrate at 72 h.

and HAP could not be observed in the absence of citrate. Clearly, the crystal transition was triggered by citrate. It was interesting that no HAP precipitation could be detected in the bulk solution; all of the resultant HAP crystallites adsorbed strongly onto the DCPD surfaces. At a time of 72 h, the DCPD phase disappeared from the XRD pattern and all of the solids were HAP (Figure 1a), indicating a complete solid-solid transformation. Another interesting phenomenon was that the resultant pure HAP crystals were well aggregated; the original structure of DCPD could be kept by these aggregates (Figure 1b and e). The transformed HAP crystals were flake-like; however, the conventionally precipitated HAP crystals in

supersaturated solutions were typically rod-like with random and loose aggregation. It should be mentioned that the nano apatite crystallites in biological bone are also flake-like.41 The promotion effect of citrate on the transition kinetics depended upon the concentrations. The DCPD-HAP transformation rate increased with an increasing citrate concentration (Figure 3); e.g., at 0.4 mM citrate, the complete transformation from DCPD to HAP required an extended time period of 120 ( 10 h. In the range of 0-1 mM, the time for complete transformation decreased with an increasing citrate concentration. The reaction rates were approximately proportional to the additive amounts. The maximum rate was achieved at a citrate

Biomimetically Triggered Mineral Transformation

Figure 2. (a) SEM images of a fracture product at early stage of phase transformation (at 14 h). (b) The SAED pattern of the inside structure implied a phase of DCPD single crystal. (c) The diffraction dots were different from those in part b, which was attributed to HAP crystallite aggregates.

Figure 3. Plot of the time period for complete transformation from DCPD to HAP against citrate concentration.

concentration of 1 mM (complete transformation within only 72 ( 6 h), which was only one-third of that at 0.1 mM citrate (216 ( 12 h). However, further increase of the citrate concentration (>1 mM) could not affect the reaction any more; the complete transformation period was kept unchanged, around 64-70 h. The plot (Figure 3) showed that a small amount of citrate could modulate the calcium phosphate transformation effectively. A combination of ex situ SEM and in situ AFM observations was used to follow the crystal evolutions. At the early state, only the dissolution of DCPD was observed (Figure 4a), which mainly occurred at the crystal edges. It was in agreement with an understanding that the crystal edges are always active in dissolution. Numerous steps were formed by the dissolution around the edge; the DCPD crystals became sawtoothed with rough surfaces. The dissolution of DCPD was terminated when an equilibrium state was reached in the solution, and no change was observed since then. Therefore, in the absence of citrate, the SEM images of the samples were not changed significantly with the experimental time (Figure 1c). However, nano plate-

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10841 like HAP formed on the DCPD (Figure 4b) as early as 10 h in the presence of 1 mM citrate. These HAP crystallites first appeared on the step-rich regions such as the DCPD crystal edges. AFM revealed the relationship between the surface steps and phase transformation in details. The phases of DCPD and HAP could be distinguished using a combination of phase imaging by tapping mode and force curve measurement under AFM. The adhesion force curves of the AFM tip on HAP and DCPD in citrate solution were different (Figure 4e). The force on HAP was significantly higher than that on DCPD, which might be attributed to the distinct adsorption profiles of citrate onto these minerals. The examined adhesion force on the dark domains in Figure 4c and d was 0.38 ( 0.06 nN, which was close to that of 1 mM citrate treated DCPD (0.40 ( 0.06 nN). The force on the bright domains was 0.70 ( 0.07 nN, and it was similar with that of HAP treated by 1 mM citrate solution (0.69 ( 0.07 nN). These data implied that the dark and bright domains were DCPD and HAP (Figure 4c and d), respectively. The nucleation of HAP was in association with the dissolution steps. Therefore, the bright lines under the AFM phase image could be considered as the step outlines. Their height was the same, 1.60 ( 0.25 nm (Figure 4d); this value was consistent with the lattice parameter of DCPD at the (010) dimension, 1.52 nm.13 It was interesting that the new transformed HAP phase was often first detected on the dissolution steps of DCPD and the HAP phase then developed along the steps on the DCPD surfaces. The density of HAP was proportional to the density of steps on the DCPD surface. For an example, the dissolution step density on the edge of DCPD was relatively high at the early stage and, then, most HAP phases were observed there (Figure 4a-d). This result indicated that the dissolution steps of DCPD provided the active sites for the phase transformation. Pits were the generated dissolution sites on the crystal surfaces during the dissolution, and they could dominate the reaction. As a result, most transformation actually occurred on the large crystal faces later to form a close layer on the crystal plane. It was well-known that calcium ions of biominerals can interact with carboxyl groups of additives (e.g., citrate) to induce the precipitation of calcium minerals such as HAP. However, the hydrated layer was the terminating surface on the top/bottom faces (010) of DCPD in aqueous solutions.26 The steps could make calcium ions exposed to the solution. We suggested that these exposed calcium ions offered the active sites for the phase formation with the additional enrollment of citrate. It was noteworthy that citrate dramatically promoted the crystal transformation of calcium phosphate crystals. This result could be extended to understand the roles of citrate additives in bone therapies. An arising question was why citrate could affect the kinetics of the inorganic phase transition. A common understanding was that the citrate altered the dissolution behavior of the precursor phase, DCPD. It was suggested that, in a liquid mediated solid-solid phase transition, the size and shape of crystallites could determine the dissolution rate of the metastable phase, thus the rate of transformation from the metastable phase to the stable phase. However, our previous work had already demonstrated that citrate had no significant effect on DCPD dissolution.42 Besides, it was also reported that citrate inhibited HAP formation in solution.28 Therefore, these previous understandings could not interpret the promotion effect of citrate on the biomimetic phase transformation of calcium phosphate. Under our experimental conditions of pH 8.45 ( 0.01, HAP should have the lowest Gibbs free energy in aqueous media

10842

J. Phys. Chem. B, Vol. 113, No. 31, 2009

Jiang et al.

Figure 4. Ex situ SEM images of the crystal transformation in the presence of 1 mM citrate. (a) At 3 h, the sites of dissolution (arrows) were rich at the crystal edge. (b) At 10 h, the flake-like HAP crystallites (arrows) were first observed at the edges of DCPD. In situ AFM phase images were obtained by tapping mode, which was extremely sensitive to surface inhomogeneities and compositions. It was found that the formation of the HAP phase was always in association with the DCPD dissolution steps. In the phase images, the DCPD dissolution steps could also be reflected by the HAP phase. (c) At 5 h, the HAP phase (bright domains, white arrows) was detected on the dissolution steps of DCPD (dark domains, black arrows). (d) At 7 h, the HAP phase was spread along the steps. AFM adhesion study: (e) Force curves on the standard calcium phosphate crystals and probing domains in the reaction solution (in the presence of 1 mM citrate).

among the calcium phosphates. Thus, the transition from the metastable phase, DCPD, to the stable phase, HAP, should be spontaneous. However, the experimental result was against this hypothesis. It was noted that the above-mentioned thermodynamic understanding only described the states of bulk solids during the transformation. It was important to emphasize that almost biomineralization reactions occur in solution environments. Actually, it had been experimentally confirmed that calcium phosphate precursor phases including ACP, DCPD, TCP, and HAP can be stable in some solvents. It implied that the crystal-solvent interaction actually played a key role in the solid transition. Analogous to other chemical reactions, the phase transformation from DCPD to HAP was also driven at the solid-liquid interface. The interfacial structure of the solid-

solution must be taken into account during the discussions. Besides the change of total volume Gibbs free energy between the two crystals (∆GV), the interfacial Gibbs free energy change (∆GS) of solid-solution was also altered during the reaction. In the previous discussion, the bulk term of ∆GV was considered only, which decided the stability of the solids thermodynamically. The interface term of ∆GS was unfortunately ignored. Actually, the total Gibbs free energy change in the crystal transformation (∆GT) could be described as eq 3.

∆GT ) ∆GV + ∆GS and the interfacial term in this study was expressed by

(3)

Biomimetically Triggered Mineral Transformation

∆GS ) SHAPγHAP - SDCPDγDCPD

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10843

(4)

where S was the total surface area of the crystal phase and γ the interfacial energy between the solid and reaction solution. In the current study, the DCPD-solution interface was turned into the HAP-solution interface gradually when the transformation occurred. The interfacial energies between the solid and reaction solution were calculated using Young’s equation. The data showed that the interface of HAP-solution was significantly different from that of DCPD-solution under citrate-free conditions. It could be seen from Figure 5 visually that the examined contact angles of the reaction solution on HAP and DCPD were 42 ( 2 and 14 ( 1°, respectively. The estimated γHAP, 9.00 mJ/m2, was greater than γDCPD, 0.20 mJ/m2. These two values were in agreement with the previous studies.17,29,37 The difference of γ indicated that the DCPD surface was more stable than the HAP surface in the reaction solution. Extra energy was required to turn the stable DCPD-solution interface into the unstable HAP-solution interface, and the interface transformation was thermodynamically unfavored. Thus, the change of interfacial energy, ∆GS, during the crystal transition was positive. This energy barrier must be overcome prior to the transformation reaction (Figure 5). It was emphasized that the dimensions of HAP crystallites were always at least 1 order smaller than those of DCPD. It implied that SHAP was greatly larger than SDCPD after the transformation. Our SEM results showed that the formed HAP crystallites were tiny (size of hundreds of nanometers) and the mother DCPD single crystals were in the size scale of tens of macrons. As the significantly increased values of γ and S, the term of ∆GS might not be compensated by ∆GV. Thus, the interfacial energy gaps resulted in the suppressed transformation from DCPD to HAP. However, citrate could change the interfacial properties of HAP and DCPD in the reaction solution. It was interesting to note that the surfaces of HAP and DCPD became close with each other in the presence of citrate. A visual representation was demonstrated lively by the contact angle measurements: in the presence of 1 mM citrate, the contact angles of the reaction solutions on HAP and DCPD were 24 ( 1 and 23 ( 1°, respectively (Figure 5), which became similar. The γHAP, 4.90 mJ/m2, and γDCPD, 4.12 mJ/m2, values were calculated by using Young’s equation. It was indicated that citrate increased the solid-solution energy of DCPD but decreased that of HAP, which meant that the DCPD surface became more active (unstable) and the HAP surface was stabilized in the reaction solution after the citrate treatment. These two opposite modification effects greatly reduced the interfacial barrier for the phase transformation from DCPD to HAP (Figure 5). Thus, we could understand that citrate triggered and promoted the transformation reaction due to the narrowed interfacial barrier. The regulation effects of citrate on γHAP and γDCPD were concentration-dependent within the citrate concentration range 0-1 mM (Figure 6). The values of γHAP and γDCPD kept on decreasing and increasing, respectively, with the increase of citrate. Thus, the interfacial barrier was reduced accordingly in this concentration range. At a higher concentration level (>1 mM), the alternation degrees remained unchanged. This tendency of the reduced interfacial energy gap between DCPD and HAP vs citrate concentration matched that of the transformation kinetics well. Conclusions Nowadays, there are various understandings about proteins and biological matrix in biomineralization. It is widely agreed

Figure 5. Influence of citrate on contact angles and interfacial free energy change (∆GS) of DCPD and HAP in the reaction solutions. Parts a-d show visually the contact angles of the reaction solution on HAP and DCPD surfaces in the absence and presence of 1 mM citrate.

Figure 6. Interfacial energies of DCPD-solution and HAP-solution as a function of citrate concentration in the reaction solution.

that these biomolecules with the functional groups (e.g., carboxyl groups) can induce the nucleation of calcium phosphates.27 In contrast, much experimental evidence also shows that the adsorbed functional organic compounds can inhibit the crystal growth of apatite or other minerals.28,29 It is previously reported that citrate can inhibit HAP growth and it cannot affect DCPD dissolution significantly. Obviously, such an understanding cannot explain many biomineralization phenomena in nature, especially the promoted phase transformation kinetics. It is important that the medical trials have clearly confirmed the acceleration effect of citrate on in ViVo formation of apatite from bone cements.25,43 These conflictions imply an unrevealed mechanism of the biological controls in biomineralization. In this study, we demonstrate the trigger and promotion effects of citrate on the phase transformation from DCPD to HAP, supporting the positive effect of citrate in the evolution of apatite in living organisms. The interfacial barrier is revealed, which suppresses the spontaneous transformation from DCPD to HAP. However, the presence of citrate decreases the stability of DCPD in solution and increases that of HAP to reduce the barrier, triggering the transition. The regulated interfacial controls can be considered as another effective pathway for biomolecules to regulate mineralization and other solid phase transformations in solutions. Acknowledgment. We thank Dr. Liping Xiao and Ms. Jieru Wang for their assistances on the XRD and SEM studies. This

10844

J. Phys. Chem. B, Vol. 113, No. 31, 2009

work was supported by National Natural Science Foundation of China (20601023 and 20871102), Zhejiang Provincial Natural Science Foundation (R407087), and Daming Biomineralization Foundation. References and Notes (1) Mann, S.; Didymus, J. M.; Sanderson, N. P.; Heywood, B. R.; Samper, E. J. A. J. Chem. Soc., Faraday Trans. 1990, 86, 1873. (2) Addadi, L.; Moradian, J.; Shay, E.; Maroudas, N. G.; Weiner, S. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 2732. (3) Lowenstam, H. A.; Weiner, S. On Biomineralization, Oxford Univ. Press: Oxford, U.K., 1989. (4) Ngankam, P. A.; Lavalle, Ph.; Voegel, J. C.; Szyk, L.; Decher, G.; Schaaf, P.; Cuisinier, F. J. G. J. Am. Chem. Soc. 2000, 122, 8998. (5) Mann, S.; Archibald, D. D.; Didymus, J. M.; Douglas, T.; Heywood, B. R.; Meldrum, F. C.; Reeves, N. J. Science 1993, 261, 1286. (6) Teng, H. H.; Dove, P. M.; Orme, C. A.; De Yoreo, J. J. Science 1998, 282, 724. (7) Alivisatos, A. P. Science 2000, 289, 736. (8) Douglas, T. Science 2003, 299, 1192. (9) Weiner, S.; Sagi, I.; Addadi, L. Science 2005, 309, 1027. (10) Tseng, Y. H.; Mou, C. Y.; Chan, J. C. C. J. Am. Chem. Soc. 2006, 128, 6909. (11) Belcher, A. M.; Wu, X. H.; Christensen, R. J.; Hansma, P. K.; Stucky, G. D.; Morse, D. E. Nature 1996, 381, 56. (12) Politi, Y.; Arad, T.; Klein, E.; Weiner, S.; Addadi, L. Science 2004, 306, 1161. (13) Dorozhkin, S. V.; Epple, M. Angew. Chem., Int. Ed. 2002, 41, 3130. (14) Tolbert, S. H.; Landry, C. C.; Stucky, G. D.; Chmelka, B. F.; Norby, P.; Hanson, J. C.; Monnier, A. Chem. Mater. 2001, 13, 2247. (15) Furuichi, K.; Oaki, Y.; Imai, H. Chem. Mater. 2006, 18, 229. (16) Yamamoto, K.; Kumagai, H.; Fujii, T.; Yano, T. Biosci. Biotechnol. Biochem. 1993, 57, 1788. (17) Liu, Y.; Nancollas, G. H. J. Phys. Chem. B 1997, 101, 3464. (18) Colfen, H.; Mann, S. Angew. Chem., Int. Ed. 2003, 42, 2350. (19) Weiner, S.; Wagner, H. D. Annu. ReV. Mater. Sci. 1998, 28, 271. (20) Weiner, S.; Traub, W. FASEB J. 1992, 6, 879. (21) Lowenstam, H. A.; Weiner, S. Science 1985, 227, 51. (22) Ambrosio, A. M. A.; Sahota, J. S.; Khan, Y.; Laurencin, C. T. J. Biomed. Mater. Res. 2001, 58, 295.

Jiang et al. (23) Crane, N. J.; Popescu, V.; Morris, M. D.; Steenhuis, P.; Ignelzi, M. A., Jr. Bone 2006, 39, 434. (24) Tsuruoka, S.; Schwartz, G. J.; Ioka, T.; Yamamoto, H.; Ando, H.; Fujimura, A. Am. J. Nephrol. 2005, 25, 233. (25) Schneiders, W.; Reinstorf, A.; Pompe, W.; Grass, R.; Biewener, A.; Holch, M.; Zwipp, H.; Rammelt, S. Bone 2007, 40, 1048. (26) Flade, K.; Lau, C.; Mertig, M.; Pompe, W. Chem. Mater. 2001, 13, 3596. (27) Macipe, A. L.; Morales, J. G.; Clemente, R. R. AdV. Mater. 1998, 10, 49. (28) Johnsson, M.; Richardson, C. F.; Sallis, J. D.; Nancollas, G. H. Calcif. Tissue Int. 1991, 49, 134. (29) Tang, R.; Darragh, M.; Orme, C. A.; Guan, X.; Hoyer, J. R.; Nancollas, G. H. Angew. Chem., Int. Ed. 2005, 44, 3698. (30) Qiu, S. R.; Wierzbicki, A.; Orme, C. A.; Cody, A. M.; Hoyer, J. R.; Nancollas, G. H.; Zepeda, S.; De Yoreo, J. J. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 1811. (31) Stephenson, A. E.; De Yoreo, J. J.; Wu, L.; Wu, K. J.; Hoyer, J.; Dove, P. M. Science 2008, 322, 724. (32) Hoang, Q. Q.; Sicheri, F.; Howard, A. J.; Yang, D. S. C. Nature 2003, 425, 977. (33) Dixon, T. F.; Perkins, H. R. Citric acid in bone. In The Biochemistry and Physiology of Bone; Bourne, G. H., Ed.; Academic Press: New York, 1956. (34) Marshall, R. W.; Nancollas, G. H. J. Phys. Chem. 1969, 73, 3838. (35) Van Oss, C. J. Interfacial Forces in Aqueous Media; Marcel Dekker: New York, 1994. (36) Van Oss, C. J.; Chaudhury, M. K.; Good, R. J. Chem. ReV. 1988, 88, 927. (37) Wu, W.; Nancollas, G. H. AdV. Colloid Interface Sci. 1999, 79, 229. (38) Shen, Q. Langmuir 2000, 16, 4394. (39) Frayssinet, P.; Gineste, L.; Conte, P.; Fages, J.; Rouquet, N. Biomaterials 1998, 19, 971. (40) Marino, F. T.; Torres, J.; Tresguerres, I.; Jerez, L. B.; Cabarcos, L. J. Biomed. Mater. Res. 2007, 81A, 93. (41) Tao, J.; Pan, H.; Zeng, Y.; Xu, X.; Tang, R. J. Phys. Chem. B 2007, 111, 13410. (42) Tang, R.; Hass, M.; Wu, W.; Gulde, S.; Nancollas, G. H. J. Colloid Interface Sci. 2003, 260, 379. (43) Yokoyamaa, A.; Yamamotoa, S.; Kawasakia, T.; Kohgob, T.; Nakasu, M. Biomaterials 2002, 23, 1091.

JP904633F