Roles of Amorphous Calcium Phosphate and Biological Additives in

Potential mechanisms for formation of highly organized biomineralized structures include oriented crystal growth on templates, the aggregation of nano...
4 downloads 0 Views 1MB Size
Download PDF
13410

J. Phys. Chem. B 2007, 111, 13410-13418

Roles of Amorphous Calcium Phosphate and Biological Additives in the Assembly of Hydroxyapatite Nanoparticles Jinhui Tao,† Haihua Pan,† Yaowu Zeng,‡ Xurong Xu,† and Ruikang Tang*,† Department of Chemistry and Center for Biomaterials and Biopathways, Zhejiang UniVersity, Hangzhou, Zhejiang, 310027, People’s Republic of China, and Centers of Analysis and Measurement, Zhejiang UniVersity, Hangzhou, Zhejiang, 310028, People’s Republic of China ReceiVed: April 29, 2007; In Final Form: September 19, 2007

Potential mechanisms for formation of highly organized biomineralized structures include oriented crystal growth on templates, the aggregation of nanocrystals by oriented attachment, and the assembly of inorganic nanoparticles mediated by organic molecules into aggregated structures. In the present study, the potential role of amorphous calcium phosphate (ACP) in facilitating the assembly of hydroxyapatite (HAP) nanoparticles into highly ordered structures was evaluated. The physical characteristics of HAP nanoparticles prepared by three different methods were analyzed after extended exposure to additives in solution. Higher order HAP architecture was detected only when the starting particles were aggregates of nanospheres with HAP cores and ACP shells. Enamel-like HAP architecture was produced when the biologic additive was 10 mM glycine or 1.25 µM amelogenin. Large platelike crystals of the type present in bone were induced when the additive was 10 mM glutamic acid. Surface ACP initially links the HAP nanoparticles in a way that allows parallel orientation of the HAP nanoparticles and then is incorporated into HAP by phase transformation to produce a more highly ordered architecture with features that are characteristic for HAP in biologic structures. These studies provide evidence for a new mechanism for assembly of biominerals in which ACP functions by linking HAP nanocrystals while they assume parallel orientations and then is incorporated by phase transformation into HAP molecules that rigidly link HAP nanocrystals in larger fused crystallites. Biologic molecules present during this process of biomineral assembly specifically regulate the assembly kinetics and determine the structural characteristics of the final HAP architecture.

1. Introduction The classical model of biomineralization considers mineral formation as an amplification process in which individual atoms or molecules add to existing nuclei or templates,1 while living organisms may make use of proteins and peptides to deterministically modify nucleation, growth, and facet stability.2 However, this conventional concept for crystal growth has been recently challenged by a model involving aggregation-based growth. Banfield2d,3-6 and co-workers have demonstrated that inorganic nanocrystals can aggregate into ordered solid phases via oriented attachment to control the reactivity of nanophase materials in nature. The arrangement of nanocrystallites can be realized with the “fusion” of the blocks since they share a common crystallographic orientation at interface, which has been successfully used to synthesize many nano materials.7 Besides, another concept of mesocrystal, which consists of crystallographic oriented crystallines connected by polymers or surfactants,8-10 has been suggested in the studies of biomineralization.2b,11 The formations of mesocrytals have been previously observed in the growth process of CaCO3,12 CdS,13 BaSO4,14 and CuO15 under the regulation of various organic molecules. In these mesocrystals, the fully crystallized inorganic phase and the organic molecules connect the adjacent crystallites, resulting in the superstructures. The interaction between crystallites and * To whom correspondence should be addressed. Phone: + 86-5718795-3736. Fax: + 86-571-8795-3736. E-mail: [email protected]. † Department of Chemistry and Center for Biomaterials and Biopathways. ‡ Centers of Analysis and Measurement.

organic molecules is a crucial condition in the architecture of mesocrystal. Here, we suggest a new model concerning the biological aggregation of apatite nanoparticles. It is an inorganic phase, amorphous calcium phosphate (ACP), that cements the crystallized, hydroxyapatite(HAP). Meanwhile, the biological molecules play as the modifier during the nanoarchitecture. By using HAP nanospheres and ACP, the highly ordered enamellike and bonelike apatites are hierarchically constructed in the presence of glycine (Gly) and glutamate (Glu), respectively. Bones and teeth are made from calcium phosphates in the form of apatitic mineral phases (e.g., HAP and carbonated apatite, CAP) with a large number of proteins. However, this formation of apatitic phase is not a one-step process but probably proceeds through a complex phase transformation.16-19 The involvement of several other calcium phosphate phases, especially ACP, has been identified in the early stage of bone and tooth mineralization,20,21 and it is widely accepted that ACP is an important intermediate. Recent studies also reveal that, despite the complicated hierarchical structures of bones and teeth, their basic building blocks are nanoparticles of calcium phosphate.22-25 Constant composition experiments show that the biodemineralization of bone and tooth enamel reveals stabilized nanoapatitic particles in physiological-like fluids.23,25 Robinson et al. confirm that the enamel crystals appear to be comprised of stacks of distinct nanometer-sized apatite subunits.22 The mineralization of apatite by oriented aggregation has been also reported,26 but there is little detailed information

10.1021/jp0732918 CCC: $37.00 © 2007 American Chemical Society Published on Web 11/03/2007

Assembly of Hydroxyapatite Nanoparticles about the initial state and evolution mechanisms as the apatite mineral formation proceeds. In the present study, the role of the amorphous phase, ACP, in facilitating the assembly of hydroxyapatite (HAP) nanoparticles is evaluated. The experiments provide evidence for a mechanism for assembly of biominerals in which ACP functions by linking HAP nanocrystals to construct the highly ordered structures. It is also interesting that ACP may be incorporated by phase transformation into HAP, resulting in larger fused crystallites. With the involvement of the amorphous phase, the architecture becomes moldable so that biologic molecules present during this process of biomineral assembly can specifically determine the structural characteristics of the HAP architecture and regulate the assembly kinetics. 2. Experimental Section 2.1. Analytic Methods. Solid samples were examined by transmission electron microscopy (TEM) (JEM-200CX, JEOL, Japan) and high-resolution transmission electron microscopy (HRTEM), (JEM-2010HR, JEOL). Elemental analysis was conducted by coupled energy-dispersive X-ray spectroscopy (EDX), (INCA System, Oxford Instruments, United Kingdom). Scanning electron microscopy (SEM) was performed using a SLR10N field-emission instrument (FEI, Netherlands) after solids were coated with a thin Au film. Sample size was measured using a Nanoscope IVa multimode scanning probe microscopy (SPM) (Veeco, United States). The phase of nanospheres was examined by X-ray diffraction (XRD) (D/max2550pc, Rigaku, Japan) with monochromatized Cu KR radiation at the working voltage of 40 kV (scanning step of 0.02°). Fourier transform infrared (FT-IR) spectra were characterized using a Nexus-670 spectrometer (Nicolet, United States). Particle size was determined by dynamic light scattering (DLS) measurements using a 90 Plus Particle Size Analyzer (Brookhaven Instruments Corporation, United States). Crystallinity of HAP nanoparticles was confirmed by TEM, thin film X-ray diffraction (TFXRD), and selected area electron diffraction (SAED). 2.2. Synthesis of ACP-Coated HAP Nanospheres and Nanoaggregates. The 11.0 mM calcium solutions and 66 mM phosphate solutions were prepared using analytic grade CaCl2 and Na2HPO4 (A.R., Sino Chemical Co), respectively. The pH of a 11.0 mM calcium solution was adjusted to 9.0 with 4.0 M NaOH after adding 1.5 mM of poly(acrylic acid) (PAA) (Mw 2000 D from Sigma). A 66 mM phosphate solution (10 mL) was added to 100 mL of the 11 mM calcium solution at 0.5 mL/min while vigorously stirred. The pH of the solution during the resulting reaction was maintained at 9.0 ( 0.1 by the slow addition of 0.01 M NaOH. The resulting suspension was filtered using 0.22 µm Millipore and Nuclepore N003 filter membranes. Fifty microliters of filtered suspension was centrifuged at 2800g for 30 min. The upper 30 mL of suspension was carefully withdrawn and then was centrifuged at 36 000g for 30 min to obtain nanoparticles. To reduce the PAA level, the particles were washed three times with 5 mL of a 2.0 M Na2HPO4 solution and were further washed three times with 5 mL of distilled water alternating with 5 mL of ethanol. The particles were centrifuged at 36 000g after each wash. The nanospheres obtained were not very stable in solution without a stabilizer. Aggregates were obtained by dispersion of the nanospheres in water (0.22 wt %) at pH ) 9.0. 2.3. Synthesis of Uncoated Crystalline HAP Nanoparticles. A 0.025 M calcium solution was prepared by adding 0.2 g of calcium acetate monohydrate to 50 mL ethylene glycol (Sigma). The 0.03 M phosphate solution was prepared by adding

J. Phys. Chem. B, Vol. 111, No. 47, 2007 13411 2.27 mL of 0.3 M Na2HPO4 and 0.20 mL of 1.3 M NaOH solutions to 20 mL of ethylene glycol in a refluxing system at 120 °C. The phosphate solution was added into the calcium solution. The mixture was then refluxed at 150 °C for 24 h. The HAP nanocrystals formed were harvested by centrifuging at 36 000g for 30 min and then were washed at least three times with water alternating with ethanol with centrifugation at 36 000g for 30 min after each wash and were vacuum-dried at 30 °C. 2.4. Preparation of HAP from Metastable Supersaturated Solution. One hundred milliliters of of 2.0 mM CaCl2 solution was slowly mixed over 60 min with 100 mL of 1.2 mM Na2HPO4. The pH was adjusted to and maintained at 9.0 at 37.0 ( 0.5 °C using 0.01 M NaOH during the reaction. After aging for 24 h, the solids obtained by filtration (0.22 µm Millipore) were washed with water and were dried in a vacuum at 30 °C. 2.5. Assembly of Nanoparticles. One milliliter of a 0.22 wt % slurry of nanoaggregates prepared in section 2.2 was added to 10 mL of pH 9.0 ( 0.1 solutions of either 11.0 mM glycine (Gly) or 11.0 mM glutamic acid (Glu) (Sigma) or no amino acid. After dilution of these nanoaggregates to 0.04 wt %, 1 mL of the slurry was mixed with 1 mL of a pH 9.0 ( 0.1 solution containing 2.5 µM amelogenin (kindly provided by Bone and Joint Research Center, Second Affiliated Hospital, Zhejiang University College of Medicine). All experiments were performed at 37.0 ( 0.5 °C. During experiments, 0.1 mL aliquots were withdrawn periodically for analysis. To preclude a role for PAA in the construction process, experiments were repeated in a 0.05 mM PAA solution. However, presence of PAA significantly stabilized the nanospheres in the suspension preventing nanoaggregation or assembly for g1 month. In control experiments, one mL of a 0.22% slurry of uncoated crystalline HAP nanoparticles was added to 10 mL of pH 9.0 ( 0.1 solutions of either 11.0 mM Gly or 11.0 mM Glu or 1.375 µM amelogenin. 2.6. Computer Simulation. The apatite model proposed by Hauptmann et al.27 was used in the computer simulation, and the SPC and OPLS-AA force fields were applied for water molecules and amino acids, respectively. The molecular dynamics simulations were performed using the GROMACS package. The simulations were done in the NpT ensemble at atmospheric pressure and in 310 K temperatures. Initial Maxwell distribution velocities were used as the starting configuration. To ensure thermodynamic equilibrium, the convergence of the total energy, temperature, pressure, and simulation volume was carefully monitored during the equilibration period. 3. Results and Discussion When the calcium and phosphate solutions were mixed to prepare the supersaturated HAP solutions, a few 4-6 nm sized calcium phosphate nanospheres were nucleated in the metastable aqueous solution (Figure 1A). However, their existence did not induce the significant precipitation of HAP, and the system still kept to a metastable state. This phenomenon could be understood by a critical nucleation condition of the classical crystal growth theories. The sizes of these nuclei, 4-6 nm, were smaller than the estimated critical sizes of HAP under the experimental condition, tens of nanometers.25 Thus, the large-scaled spontaneous precipitation was not triggered. The relative stability of these nanospheres could be explained by our nanodissolution model of HAP,25 which demonstrates that nanoparticles can be dynamically protected against dissolution by their size effects. A large amount of ∼5 nm HAP spheres were synthesized by using 1.5 mM PAA as the stabilizing agent in the supersaturated

13412 J. Phys. Chem. B, Vol. 111, No. 47, 2007

Tao et al.

Figure 1. HRTEM of the well-dispersed calcium phosphate nanospheres (A) formed at the very initial stage of the calcium and phosphate solutions mixing. The inserted images show that the sizes of spheres were about 5 nm. Their cores had the structures of crystallized HAP. However, a layer of ACP coexisted. These nanospheres were not stable in their aqueous solutions and could be spontaneously aggregated to form uniform colloidal spheres (B). The inserted selected area electron diffraction (SAED) pattern shows that the colloidal aggregates were polycrystalline HAP, and the result of thin film XRD is shown in Figure S1. The subunits of the aggregates, ∼5 nm sized spheres, could be observed (C) and the measured lattice space in the core, 0.236 nm, was consistent with the (220) face of HAP (D). However, these nanospheres in the colloidal aggregates were not crystallographically organized. The arrows show the existence of amorphous calcium phosphate phases in the aggregates. The surfaces of the aggregates were covered by the thin layers of ACP too. These ACP were confirmed by SAED. No diffraction ring or dot patterns were detected, but the element analysis showed that they consisted of calcium and phosphate.

HAP solutions. The XRD examination confirmed the formation of HAP in the presence of PAA. Under HRTEM, it was revealed that the internal structures of these 5 nm sphere-shaped particles were not homogeneous as each of them consisted of a crystallized core of HAP and an amorphous layer. The lattice structure of the well-crystallized HAP core was clearly shown in Figure 1A. However, no lattice structure was detected in the shell. Except for Ca, P, and O, no other element was found using EDX microanalysis (Figure S1), and the molar ratio of calcium: phosphorus:oxygen of the shell, 1:0.68 ( 0.03:2.72 ( 0.05, was close to the stoichiometric one of Ca3(PO4)2. Therefore, it was suggested that the noncrystallized outlayer was ACP. When the protected agent, PAA, was removed, the nanospheres could form the colloidal aggregates (30 ( 5 nm) spontaneously (Figure 1B). The aggregation processes were confirmed by using in situ DLS (Figure S2). The SAED lattice fringes in Figure 1B corresponded to the (220), (400), and (132) faces of HAP (their interplanar distances were 0.204, 0.189, and 0.236 nm, respectively). The incontinuous diffraction rings or the dot rings implied that these aggregates were polycrystallites. By using HRTEM, the individual units, ∼5 nm spheres, were observed in the aggregates. Besides the amorphous phase, the crystallized HAP cores remained and their lattice structure could also be observed. However, these crystallized units were surrounded and linked by ACP (Figure 1). The SAED intermittent rings also showed no preferential crystallographic directions of these HAP during the aggregation. This phenomenon implied that the 5 nm nanospheres were randomly packed to result in the ∼30 nm aggregates. Analogous to the core-shell structure

of the ∼5 nm spheres, the surface of the aggregate was a thin layer of ACP (thickness was around 1.4 nm, Figure 1D). A model of crystallized HAP spheres locating in a moldable ACP atmosphere could be used to describe this stage. However, these ∼30 nm sized colloidal aggregates were unstable, and they acted as a transition state during the HAP formation. The nanoparticles would be deaggregated and reorganized with the moldable amorphous phases, and this process might be controlled by biological components. In our experiments, the colloidal aggregates were redispersed in water under a condition of T ) 37 °C and pH ) 9.0 in the absence and presence of the biological additives. It should be emphasized that no additional calcium or phosphate ions were introduced (so the crystallites were built from the particles and not from the ions in the supersaturated solutions). By the controls of Gly and Glu, highly ordered enamel- and bonelike apatites could be finally evolved, respectively. In this study, Gly and Glu were used since their roles in the conventional solution growth of HAP have been revealed. In the presence of Gly (10 mM), the reorganization of aggregates preferred one-dimensional assemblies as the 5 nm nanospheres were lined up (Figure 2A). It was noted that there was no direct attachment of the crystallized HAP phases in the linear assembly; however, they were connected by the ACP phase. Although 10 mM Gly was introduced, the EDX analysis indicated that the main composition between the crystallized HAP spheres was still ACP. The existence of amino acid was not detected, which might be explained by its low percentage in the amorphous phase. Therefore, the connecting role of Gly

Assembly of Hydroxyapatite Nanoparticles

J. Phys. Chem. B, Vol. 111, No. 47, 2007 13413

Figure 2. Reorganization of HAP nanospheres in the presence of Gly. The previously formed colloidal aggregates were dissociated and the detached nanospheres were aligned into the linear chains (A). ACP (white arrows, B) played the cementing role in the connections of the HAP nanoparticles. At this stage, the HAP nanospheres were not homogeneously oriented since their crystallographic directions were disordered. The evolution was associated with the phase transformation of calcium phosphates, and the newly crystallized HAP domains (circles, C) in the ACP phase are illustrated between the two HAP nanocrystallites. The final linear assemblies could be an integral and complete single HAP crystal (D and E). The anisotropic diffraction dots are shown under SAED. The crystal lattice spacing along the long axis of crystal was 0.342 nm, which corresponded to the (002) face of HAP.

should be excluded in this case. The lattice structures of the HAP blocks are illustrated in Figure 2B, and the interplanar spacing of (002) HAP faces, 0.342 nm, was determined. Thus, the crystallographic directions of these HAP particles could be indexed. Figure 2B showed that the directions of these particles were disordered and that the crystal faces of the neighboring HAP nanoparticles did not match with each other. It could be concluded that at the beginning, the HAP spheres in the linear assembly were random and the presence of ACP made such mismatched attachments possible. However, ACP was not thermodynamically stable under the solution conditions, and it could transform into HAP. Such a transformation was detected in the linear assembly. Figure 2C shows an intermediate state: the ACP, linking the HAP nanoparticles, started to partially crystallize (shown in white circles), indicating a phase transformation process. During the phase transformation, the “hard-soft-hard” (HAP-ACPHAP) connections disappeared and the “hard-hard” (HAPtransferred HAP-HAP) attachment began to dominate in the linear assembly. As a suggestion of previous literature, the amorphous phase in solution had a fluidlike character.28,29 Thus, the adjustments of crystallographic orientation of the nanoparticles by jiggling in the moldable ACP phase were possible. Since the phase transformation of ACP into HAP phase occurred concurrently with the direction adjustment of the adjacent HAP nanoparticles, the existing HAP and the newly formed HAP could rotate to find the lowest energy configuration, reducing the mismatching of the crystal faces. Thus, these subunits would like to be attached by sharing the same crystal face, and the evolved hard-hard attachment became oriented. Finally, the

adjustments and phase transformations resulted in the fusion of the nano HAP and ACP to form the complete HAP single crystals. Figure 2D and E demonstrates the final state of the “evolved” linear assemblies, which were the needle-shaped single crystals of HAP. The formation of the single crystals could be confirmed by their continuous and complete lattice structures. At this stage, no ACP or Gly domain was detected under HRTEM in the interior of the resulting single crystals. The uniform lattice spacing was 0.346 nm, indicating that the long axis of the needlelike HAP crystal was its c direction (or [001] direction). The finally formed HAP single had a complete and uniform structure, which implied that the foreign compounds such as Gly were expelled from the interior structures during the evolution. A typical time period for the formation of such single crystals from the reorganization of nanoparticles in the presence of Gly was about 10 days. Actually, the surface of the resulting HAP single crystal (∼5 nm in width and ∼80 nm in length) was also covered by a thin layer of ACP (Figure 2B), and it could again aggregate to form a hierarchical structure of HAP under a unique control of Gly. The micrometer-sized and enamel-like apatites could be achieved with an evolution period of 2 months, and their morphologies were similar to the single crystals (Figure 3). The SAED pattern also showed that the formed large-sized apatites had good crystallinities and uniform structures. However, they were the typical mesocrystals as TEM revealed that these apatites were well-organized assemblies of HAP nanocrystals. In this stage, numerous HAP single crystals assembled side by side along their c-axes and the ACP cemented them together (Figure 3B). The slightly elongated diffraction spots corresponding to (002)

13414 J. Phys. Chem. B, Vol. 111, No. 47, 2007

Tao et al.

Figure 3. By using the previously resulted HAP single crystals as the building blocks, secondary assembly could be induced by Gly. The mesocrystals of enamel-like apatite could be finally achieved, which were formed by the side-side assembly of the HAP needlelike crystals along their c axes (A, their SEM results are shown in Figure 8). The inserted SEAD shows the strong diffraction dots at the [002] and [004] HAP directions, confirming the orientation. (B) The detailed heterogeneous structures show that ACP phase (white arrows) cemented the single crystals (black arrows). (C) SEM of the formed enamel-like HAP in the presence of amelogenin protein.

and (004) faces revealed the presence of multiple crystallized needlelike HAP with a small deviation from the perfect alignment, which was a feature of human enamel.30 The bunch structures of the oriented HAP crystals were analogous to the biological enamel rods. The formation of enamel-like apatites was also repeated by using bovine amelogenin. At a concentration of 1.25 µM of amelogenin, the large enamel-like HAP mesocrystals appeared within only 3 days (Figure 3C), which was about 20 times faster than 10 mM Gly. It showed that this protein, a well-known effective modifier during the in vivo tooth enamel formations, could dramatically accelerate the kinetics of nanoassembly. This result was also in agreement with the previous understanding that amelogenin promoted the formation of elongated apatite microstructures.31 Another amino acid, Glu, could result in the formation of plate-shaped, or bonelike, apatite crystals by a similar pathway. Figure 4 showed that the nano HAP crystallites and ACP were assembled into plates while the initial 5 nm spheres detached from unstable calcium phosphate colloidal aggregates. The thickness of the plate, ∼5 nm, confirmed by TEM and AFM (Figures S3 and S4), was in agreement with the dimension of the basic building blocks. Similar to the initial stage of the needle-shaped apatites, the HAP nanoparticles were linked disorderly by ACP in the plates, and their final assemblies evolved into the complete single crystals with large (001) crystal faces. Six-fold symmetry in the SAED pattern and lattice fringes of (200) face (Figure 4C and 4D) of the evolved crystals confirmed that the oriented crystallographic directions for the “evolution” were parallel to the ab planes (or perpendicular to c axes) of HAP. The partial formation of (210) faces of HAP phase in the ACP phase shown by black arrows and the lattice structures were also detected in an intermediate state (Figure 4B). At this stage, although the homogeneous orientations of the subunits in the platelike assemblies of numerous crystallized particles were not approached, the features of the incontinuous

and nonuniformity SAED dots (Figure 4A) indicated that the assemblies were somehow close to a well-organized structure. When the evolution was completed, the platelike single crystals of HAP were achieved by the nanoassembly. In this case, the addition of Glu resulted in a different architecture mode, which could not be obtained in the control experiments at all. Thus, it was realized that one of the roles of biological molecules in the nanoassembly was the modification of the mode of nanoconstruction to achieve the desired superstructure. Actually, these modification effects of amino acids with HAP agreed with the previous results of the HAP preparations in the supersaturated solutions in the presence of amino acids.32 Here, the linking role of ACP was emphasized during the nanoconstruction of HAP. In the previous understanding of the aggregation-based crystal growth,3-15,33 the crystallized blocks always had high shape anisotropy, for example, the blocks such as nanorods and nanodisks spontaneously aligned to produce crystals with analogous morphologies. The enrollment of the foreign organic phase was required in the formation of mesocrystals, for example, the neighboring crystal faces should be connected by polymer. However, with the involvement of ACP, the nanoassembly of apatite could confer much more moldable morphologies or structures of the assemblies than the aggregation-based crystal growth. It was demonstrated that the different facets of nano HAP crystallites could be connected by ACP so that the assemblies could be more flexible. However, the assemblies could also maintain a better chemical continuity of the formed mineral than the mesocrystals since the ACP could transfer into HAP. It was noted that the structural continuity of crystal was also a characteristic of biominerals. To confirm the role of ACP during the nanoassembly of HAP, the highly crystallized nanocrystals were used to repeat the above experiments. The crystallinity, phase purity, and morphology of the well-crystallized HAP nanoparticles are shown in Figures 5 and S1. It is shown that the ACP phase did not coexist

Assembly of Hydroxyapatite Nanoparticles

J. Phys. Chem. B, Vol. 111, No. 47, 2007 13415

Figure 4. Intermediate state of the plate-shaped assembly under the control of Glu (A). The existence of intermittent rings in SAED indicated that the plate consisted of crystallized HAP particles with the different crystallographic orientations. Analogously, the HAP nanospheres were cemented by the ACP phase, which is shown by the black arrows (B). The newly formed HAP structures from the ACP are also clearly labeled. (C) A final state of the evolved plate-shaped assembly in the presence of Glu. The SAED pattern, having typical 6-fold axis symmetry, shows the formation of well-crystallized single HAP crystals, and their large crystal face is (001). The complete and integral lattice structure can be observed on the (001) crystal face. The measured lattice space is inconsistent with interplanar distance of (200) (D). The inserted image shows the result of in situ inversed fast Fourier transformation.

in these HAP particles. In water, the well-crystallized nanoparticles were randomly aggregated but the HAP aggregations had the loose structures (Figure 5A). In the presence of Gly (Figure 5B), Glu (Figure 5C), and amelogenin (Figure 5D), these nano HAP remained in the original stage without any change even for a month and their TEM images were identical. These control experimental results demonstrated that the biological additives and HAP could not induce any nanoassembly without the involvement of ACP, which indicated that the presence of ACP should be an important prerequisite for the biological architecture of apatite. The similar evolution from the random aggregates to the needlelike single crystals of HAP was also observed in the absence of Gly (Figure 6). This phenomenon again showed that the involvement of Gly was not a required condition during this biological architecture. Therefore, it was clear that the link role of Gly could be eliminated. However, in the absence of Gly or amelogenin, the resulting needlelike HAP nanocrystals remained in a state of random aggregations (the same as Figure 6), and there was no higher level construction so that the enamellike superstructure could not be achieved. Thus, the hierarchical architecture of enamel-like apatite could be due to the specific enrollments of Gly and amelogenin. Recently, Wang et al. reported that HAP could be assembled in the presence of amelogenin.31 Their study was performed in controlled supersaturated solutions so that ACP should be involved. Our control experiment demonstrated that the two components, the wellcrystallized HAP nanoparticles and the amelogenin, could not result in any assembly (Figure 4d). Thus, we suggest that a role

of Gly, Glu, and amelogenin in this biological construction was the control of orientation and superstructures of the HAP assemblies. The biomolecules could also significantly promote the kinetics of the biological construction of apatite. For example, it took about 10 and 15 days to form the HAP single crystals from the nanoaggregates in the presence and absence of 10 mM Gly, respectively. The rate increased about 30% with the involvement of Gly. Furthermore, during the construction of the enamellike HAP, 1.25 µM amelogenin could accelerate the formation rate, and the superstructure was achieved within only 3 days, which was about 20 times faster than with 10 mM Gly. These results implied that the promotion effects of the biological additives were obvious during the nanoconstruction of HAP. The previous observed effect on crystallization kinetics strongly suggested that organic compounds modify the growth stage of minerals.34,35 However, most literature reports that the additives inhibit the crystal growth of calcium phosphates in the conventional crystallization process. On the other hand, many biomineralized structures in nature suggest that organic components induce nucleation. In this paper, we demonstrate that organic components might control the nanoassembly of mineral particles and its kinetics. The biomineralization of HAP is usually understood by using the classical growth models.34 The present work demonstrates that the strategy of nanoassembly with amorphous phase could be also used to describe the HAP formation at the nanoscale. A question is raised as to how the amino acids or proteins control the directions and orientations during the nanoconstruction and is thus one of the major challenges to an understanding

13416 J. Phys. Chem. B, Vol. 111, No. 47, 2007

Tao et al.

Figure 5. TEM image of HAP nanocrystals with high crystallinity and their aggregates in the presence of different biological regulators. The average size of these well-crystallized HAP particles were ∼30 nm, and the crystallinity was also confirmed by the inserted SAED image (A). The ACP shell was not detected in these solids. In the solution, they were randomly aggregated with loose structure. This state was unchanged in the solutions after 1 month in the presence of 10 mM Gly (B), 10 mM Glu (C), and 1.25 µM amelogenin.

Figure 6. The needlelike HAP crystallite could be spontaneously constructed by using core-shell structured HAP-ACP nanospheres in the absence of any additives. However, a longer time period (15 days at 37 °C) was needed to reach this stage in the absence of Gly.

of the mechanisms by which this level of control is achieved. A very general and useful construct for approaching this question is that of energy landscape, which highlights the minimum free energy during mineral formation.35,36 All aspects of the crystal formations, including phase and habit, can finally be determined by the landscape, and the geometry and the stereochemistry of the interaction between crystals and modifiers can be also understood by the effect of this interaction on the energy landscape. A model of interfacial energy control is suggested for the studies on mechanisms of biomineralization.34 Figure 7 and Table 1 summarize the results of computer simulations, which show that Gly and Glu anisotropically change the HAPsolution interfacial energies, γ, at the different crystallographic

Figure 7. Molecular dynamic simulation snapshots of the adsorbed Gly and Glu on the HAP (001) or (100)/(010) faces at 1.5 ns. Gly adsorbed on HAP faces over a layer of water; Glu could adsorb directly to the calcium sites on the (001) and (100) faces. These anisotropic adsorptions resulted in the different interfacial energy changes (Table 1).

directions. The apatite crystal belonged to a hexagonal system with the identical (100) and (010) crystal faces. In the control HAP-water system, γ of (001) face was close to that of (100)/ (010) face. The introduction of Gly raised the energy of (001) face greatly and reduced that of (100) slightly. The value ratio

Assembly of Hydroxyapatite Nanoparticles

J. Phys. Chem. B, Vol. 111, No. 47, 2007 13417 was significantly lowered by Glu but that of (100)/(010) remained unchanged. The ratio of γ(001):γ(100)/(010), 0.47, implied that the (001) face was stabilized by Glu and that the aggregation along the a and b HAP axes was preferred, resulting in the bonelike crystals. Besides, it was found that the nucleation or growth of apatite crystallines was rarely observed at relatively low supersaturations, which could be attributed to the relatively high-energy barrier for HAP spontaneous crystallizations.25 However, the biological assembly of HAP could provide a sequential process with lower energy barrier for the organisms to make biomaterials at mild conditions since it did not require the strictly critical condition for the conventional crystallization. The computer simulation also indicated that the amino acid could interact with the active sites on the HAP crystal with a distance of ∼1.0 nm. For example, the interaction of the amino acids and HAP crystal faces could be over a layer (Figure 7). It has been reported that the aragonite CaCO3 platelets in nacre of Haliotis laeVigata were covered with a continuous layer of disordered amorphous CaCO3 (ACC) and that there was no protein interaction with this layer.37 The charge interaction of the highly polar aragonite face could extend throughout the 3to 5-nm ACC carbonate layer.37 Interestingly, the thickness of the ACP layer in current biological architecture of apatite was at the same scale, and the HAP crystal faces were highly polarized too. Thus, it was reasonable that the amino acids or protein interacted with the HAP over the thin ACP layer. Although the presence of the Gly and Glu between the neighboring HAP nanoparticles (or in ACP phase) was not detected directly in the current experiments, the existence of these biomolecules could not be fully excluded from the amorphous phase. However, as the formations of the needleand bonelike single crystals, it could be concluded that the molecules of Gly and Glu were expelled from the interior of the crystals. In nature, amorphous phases exist extensively with readily moldable isotropic properties and of structure materials. For example, it has been recently discovered that amorphous calcium carbonate exists in organisms and plays an important role in the biomineralization of calcium carbonate.37-40 Our current study emphasizes the important role of amorphous phases in the nanoconstruction of biological apatite. The nanoassembly model is schematically presented in Figure 8, and their kinetic results are summarized in Table 2. It can be seen that the nanospheres and their colloidal aggregates formed initially by particle attachments in the solutions. With the participation of the moldable ACP, the biological additives could induce the rearrangements to achieve the desired structure. The highlights of such a biological apatite formation can be described as follows:

Figure 8. A schematic model of apatite evolution via the conglutination of HAP nanocrystallites (purple) and ACP phase (light blue). HAP nanospheres with ACP layer were nucleated in the metastable supersaturated solution and they could spontaneously aggregate. This process was not influenced by any additives. Under the control of biological components such as Gly and Glu, the HAP subunits in the loose aggregates could be reorganized in the moldable ACP surroundings. The modifiers could determine the different evolutionary forms, e.g., one-dimensional linear assemblies or two-dimensional plates. The crystallized HAP was cemented by the amorphous phase with the flexible structure. The ACP could transform into the thermodynamically stable HAP phases with time, and the individual HAP domains could fuse to form the single HAP crystals (red). These single crystals might be used as the building blocks in the next level of architectures, and the hierarchical structure of apatite was achieved. The SEMs show the corresponding experimental states of the nanoassembly of HAP with ACP.

TABLE 1: HAP-Solution Interfacial Energies Calculated Using Molecular Simulation systems HAP-water HAP-water with Gly HAP-water with Glu

crystal face

interfacial energy J/m2

(001) (100)/(010) (001) (100)/(010) (001) (100)/(010)

0.4675(10) 0.5164(11) 0.7139(14) 0.5096(11) 0.2373(7) 0.5052(10)

of (001) to (100)/(010) was about 1.4. Higher values of γ indicate a greater difficulty in forming such an interface between the solid and aqueous phases. Thus, HAP (001) faces became unstable in the Gly solutions and the ordered assembly along the c-axis was preferred to minimize the total interfacial energy of the assembly in the solutions. In contrast, γ of (001) face

TABLE 2: Summary of the Kinetics of Different HAP Architectures Pathways time required for starting nanoparticles

solution additivea

aggregates with HAP cores & ACP shells (30 nm)

none 10 mM glycine 10 mM glutamic acid 1.25 µM amelogenin

2 days