Article pubs.acs.org/Biomac
Different Kinetic Pathways of Early Stage Calcium-Phosphate Cluster Aggregation Induced by Carboxylate-Containing Polymers Jing Ye, Dongbo Wang, Diana N. Zeiger,† William C. Miles,‡ and Sheng Lin-Gibson* Biomaterials Group, Material Measurement Laboratory, NIST, 100 Bureau Drive, Gaithersburg, Maryland 20899-8543, United States S Supporting Information *
ABSTRACT: Acidic proteins are critical to biomineral formation, although their precise mechanistic function remains poorly understood. A number of recent studies have suggested a nonclassical mineralization model that emphasizes the importance of the formation of polymer-stabilized mineral clusters or particles; however, it has been difficult to characterize the precursors experimentally due to their transient nature. Here, we successfully captured stepwise evolution of transient CaP clusters in mineralizing solutions and studied the roles of functional polymers with laser light scattering (LLS) to determine how these polymers influence the stability of nanoclusters. We found that the polymer structure can alter CaP aggregation mechanisms, whereas the polymer concentration strongly influences the rate of CaP aggregation. Our results indicate that the ability of acidic biomolecules to control the formation of relatively stable nanoclusters in the early stages may be critical for intrafibrillar mineralization. More importantly, LLS provided information about the size and the structural evolution of CaP aggregates, which will help define the process of controlled biomineralization.
1. INTRODUCTION Mineralized tissues (such as bones and teeth) are composed of hierarchically organized apatite within an organic matrix, which in the case of bone and dentin is mostly type I collagen.1,2 Such organized structures have been shown to form by nonclassical mineralization processes directed by noncollagenous proteins (NCP) and other polyelectrolytes rich in acidic moieties. Substantial effort has been made to understand the interactions between charged molecules and calcium phosphate (CaP) to better control this polymer-mediated mineralization process. Indeed, NCPs and acidic polymers have been shown to stabilize transient nanometer-sized CaP clusters,3−8 as they have the capacity to infiltrate into collagen fibers and transform into crystalline apatite. Recent research suggests that such clusters are pervasive in mineralizing systems,7,9−12 and the ability to form higher-order structures from such clusters is a hallmark of the biomineralization process.13,14 Understanding CaP biomineralization provides new opportunities to treat diseased or injured hard tissues and provides new design cues for the controlled assembly of advanced materials. The highly transient nature of CaP clusters in terms of their size, phase, and shape has made their experimental observation difficult. Much of what is known was determined by cryogenic transmission electron microscopy (cryoTEM) measurements of hydrated particle size, phase, and aggregate structure.4,7 Although these measurements have fundamentally changed our understanding of early mineralization processes, cryoTEM is time- and labor-intensive and is sometimes techniquesensitive, thus, rendering a full characterization of the effects of different NCPs on CaP aggregation impractical. By contrast, © 2013 American Chemical Society
more commonly used methods such as ion-specific electrodes and colorimetric assays simply do not have the capacity to produce quantitative information regarding the structure in nucleation events. In phosphate biominerals, a number of NCPs such as dentin sialoprotein 115,16 and fetuin,7,17 in addition to model acidic molecules such as polyaspartic acid7,18 (pAsp) and poly(acrylic acid) (pAA) used in conjunction with other additives,3,19 have been shown to induce intrafibrillar mineralization. These molecules share acidic characteristics but differ in some properties, including specific composition and charge density. As a result, they exhibit differences in their ability to direct the mineralization process. Specifically, pAsp has been found to be effective at generating biomimetic structures through a wide compositional range, whereas pAA, a molecule that also has pendent carboxyl groups but a different backbone structure, shows limited ability to induce intrafibrillar mineralization.8 The interaction between charged polymers and CaP clearly underpins biomineralization, but its precise nature has remained elusive. Here, we have used laser light scattering (LLS) to obtain kinetic information and observe structural evolution with considerably higher throughput than imaging techniques. Both dynamic light scattering (DLS) and static light scattering (SLS) are well suited for determining the structure and aggregation kinetics of polymer solutions and colloids20−22 and have been applied to study the size evolution of polymerReceived: May 8, 2013 Revised: August 20, 2013 Published: August 22, 2013 3417
dx.doi.org/10.1021/bm400660a | Biomacromolecules 2013, 14, 3417−3422
Biomacromolecules
Article
mediated mineralization processes.6 In this study, we have used light scattering methods to determine the size and structure evolution of CaP clusters, the formation of which was mediated by either pAsp or pAA, at constant physiological pH (7.4) prior to the formation of an insoluble crystalline phase. These results demonstrate fundamental differences in the mechanism and kinetics of CaP aggregation in the presence of different acidic macromolecules.
to the induction of a crystalline phase, as indicated by a nearconstant calcium concentration in this regime (Figure S1).23 A clear indication of CaP aggregate formation is the increased scattering intensity ⟨I⟩ as a function of time, t (Figure 1), where
2. MATERIALS AND METHODS 2.1. Materials. Poly-L-aspartic acid, sodium salt (pAsp, 5000− 15000 g mol−1, MP Biomedicals), polyacrylamide (pAM, 1500 g mol−1, Aldrich), and poly(acrylic acid) (PAA, 1800 and 5000 g mol−1, Polysciences) were used as received. Stock solutions of calcium and phosphate were prepared by adding calcium chloride dihydrate (CaCl2·2H2O, ≥99.5%, Sigma) and potassium phosphate monobasic (KH2PO4, ≥99.5%, Sigma) into HEPES-buffered saline (HBS buffer; for control) or polymer solutions with predetermined concentrations in HBS buffer, respectively. 2.2. Mineralizing Solutions. All mineralizing solutions were made in HBS (150 mM NaCl and 20 mM HEPES at pH 7.4). A calcium containing HBS solution ([Ca2+] = 10 mM) was added to a phosphate containing HBS solution ([PO43‑] = 5 mM) to reach a final calcium concentration of 5 mM and a final phosphate concentration of 2.5 mM. In experiments with polymers, equal concentrations were prepared in both calcium and phosphate solutions. Data on poly-Laspartic acid was collected at concentrations from 37.4 to 93.75 μg/ mL, polyacrylamide was collected at 38.87 μg/mL, and poly(acrylic acid) was collected at concentrations from 19.71 to 49.275 μg/mL. 2.3. Laser Light Scattering Instrumentation. A Brookhaven laser light scattering (LLS) spectrometer (BI-200SM) equipped with a digital time correlator and a compact solid-state diode-pumped Nd:Vanadate (Nd:YVO4) laser (λ0 = 532 nm) was used to study the mineralization solution with and without polymeric additive in HBS buffer at pH 7.4 and 37 °C. The standard uncertainty associated with the measurement is less than 5%. In dynamic light scattering (DLS), the Laplace inversion of each measured intensity−intensity time correlation function G(2)(q,t) in the self-beating mode leads to a line-width distribution G(Γ), where q is the scattering vector. For dilute solutions, Γ is related to the translational diffusion coefficient D by (Γ/q2)q→0,C→0 → D. Therefore, G(Γ) can be converted to a translational diffusion coefficient distribution G(D) or further to a hydrodynamic radius distribution f(Rh) via the Stokes−Einstein equation, Rh = (kBT/6πη)/D, where kB, T, and η are the Boltzmann constant, the absolute temperature, and the solvent viscosity, respectively. In static light scattering (SLS), the angular dependence of the excess absolute time-averaged scattered light intensity ⟨I(q)⟩ provides the information of the z-average root-mean square radius of gyration ⟨Rg⟩, when q⟨Rg⟩ < 1, on the basis of the equation known as the Zimm plot. Furthermore, the ratio of ⟨Rg⟩/⟨Rh⟩ reflects the structural or conformational change of a scattering object because ⟨Rg⟩ reflects the density distribution of the chain in real physical space, whereas ⟨Rh⟩ is an equivalence of the radius of a hard sphere with the same translational diffusion coefficient under the same conditions. For example, for a random coil in a good solvent and a uniform hard sphere, ⟨Rg⟩ and ⟨Rh⟩ ≈ 1.5 and 0.774, respectively. 2.4. TEM Imaging of CaP Aggregates. A total of 5 μL of LLS mineralization solution was pipetted onto a carbon-coated TEM grid. After 5 min of incubation, the grid was blotted from the back side and then rinsed in ethanol. Images were collected using a Philips EM400 transmission electron microscope operated at 120 kV.
Figure 1. Time-dependent scattered light intensities ⟨I⟩ of mineralization solutions (5.0 mM CaCl2·2H2O and 2.5 mM KH2PO4): control (no additives), polyacrylamide (pAM), polyaspartic acid (pAsp), or poly(acrylic acid) (pAA). The molar concentrations of functional groups are identical. The addition of pAM had no impact on CaP aggregation, while both pAA and pAsp inhibited CaP aggregation. Inset shows the control and pAM data over a larger intensity range.
⟨I⟩ is a function of the cluster size and concentration (number of clusters).20,22 In the absence of a polymer additive or in the presence of polyamide (pAM, a control polymer not known to influence CaP mineralization) ⟨I⟩ increased rapidly with t, indicating that large CaP aggregates formed within several minutes. The aggregation rate of CaP decreased substantially with the addition of pAsp and pAA, and pAA was more effective at delaying CaP aggregation at the same carboxylate molar concentration. As the amide functionality in pAM had no effects on CaP aggregation and the carboxylate concentration was constant, the role of acidic polymers is likely more complex than simple complexation with Ca2+ ions. Furthermore, these results suggest that the mechanisms by which pAA and pAsp mediate CaP aggregation must be different. We measured the structural evolution and kinetics of CaP aggregation at varying concentrations of pAA and pAsp to investigate how the two polymers controlled CaP aggregation. Polymer concentration was normalized with respect to the number of carboxylate moieties to facilitate a direct comparison between pAsp and pAA. We defined 0.547 mM carboxylate as 1×, corresponding to 75.00 and 39.42 μg/mL for pAsp and pAA, respectively. The 1× concentration was also selected for its ability to lead to intrafibrillar mineralization.24 All other reported polymer concentrations are proportional to the 1× concentration. Figure 2 shows the scattering intensities (A,C) and average hydrodynamic radius (B,D) measured by DLS as a function of time in the presence of pAsp (A,B) or pAA (C,D). ⟨I⟩ obtained from DLS shows a decrease in aggregation rate as the polymer concentration increased for both polymers. All pAsp concentrations examined (0.5−1.25×) resulted in notable increases in ⟨I⟩ after 15 min, corresponding to the formation of CaP clusters with hydrodynamic radius ⟨Rh⟩ of 7−8 nm. The clusters gradually increased in size at a rate that was pAsp concentration-dependent. Upon reaching ≈15 nm, aggregation of clusters ensued as ⟨Rh⟩ increased more rapidly. Within the same carboxylate molar concentration range examined, only low concentrations (0.5−0.6×) of pAA led to the formation of
3. RESULTS AND DISCUSSION In this study, we examined the stability and aggregation behavior of CaP clusters formed in the presence of pAsp or pAA, as these properties may affect biomineralization processes. All measurements were made on the incipient CaP phase prior 3418
dx.doi.org/10.1021/bm400660a | Biomacromolecules 2013, 14, 3417−3422
Biomacromolecules
Article
Figure 2. Scattering intensities ⟨I⟩ in the presence of pAsp (A) and pAA (C) as a function of time, and average hydrodynamic radius ⟨Rh⟩ and aggregate structure (shape) in the presence of pAsp (B) and pAA (D) measured as a function of time. The CaP aggregation rate is highly concentration-dependent, which decreases with increasing polymer concentration. pAsp promotes the formation of pseudostable CaP clusters (relatively stable Rh shown in B) irrespective of concentration, whereas stabilized CaP clusters are notably missing in pAA-mediated systems (D). The structure evolution for the pAsp-mediated system determined by combined SLS and DLS is hyperbranched, HB; extended structure, ES; rod, R; and fractal, F (B). The structure evolution for the pAA-mediated system consists of extended structure, ES; random coil, RC; and fractal, F (D).
for detailed monitoring of structural evaluation. In addition to the size, shape evolution was determined from the widely used shape factor ρ,25−27 a dimensionless quantity defined as ρ = ⟨Rg⟩/⟨Rh⟩, where ⟨Rg⟩ is a geometrical average over the physical size of the particle and ⟨Rh⟩ is a hydrodynamically effective radius. This well-defined quantity depends strongly on the structure and shape of the object instead of molecular dimensions.25,28,29 For example, for a linear polymer coil in good solvent and for a uniform hard sphere, ρ = 1.5 and 0.775, respectively.25,28−30 In our studies, CaP aggregates underwent the following structural evolution for the 0.75× pAsp solution:
large structures, and the time required to reach a detectable aggregation size was concentration-dependent. Higher pAA concentrations (0.75−1.25×) drastically suppressed CaP cluster formation such that no significant ⟨I⟩ increase was observed over 5 h (data not shown). These results clearly indicate that CaP cluster formation was highly sensitive to changes in pAA concentration. The structure evolution (size and shape) of CaP aggregates was determined by combined SLS and DLS analysis (Figure 2B,D). For pAsp-containing solutions, the two intermediate concentrations (0.75× and 1.0×) aggregated at a rate suitable 3419
dx.doi.org/10.1021/bm400660a | Biomacromolecules 2013, 14, 3417−3422
Biomacromolecules
Article
(1) hyperbranched cluster (HB), (2) extended structure (ES), (3) rod-like (R), and (4) fractal (F) (Figure S2). CaP with 1× pAsp aggregated at a slower rate and had an HB structure (Figure S3) with an ⟨Rh⟩ ≈ 25 nm after 3.5 h. As noted earlier, pAA has a greater capacity to suppress CaP aggregation than does pAsp, and applying a similar analysis to determine ⟨Rh⟩ as a function of time (Figure 2D) for pAA-mediated mineralization showed a different path of structural evolution: (1) extended structure (ES), (2) random coil structure (RC), and (3) fractal (F). As the aggregation rate was quite sensitive to pAA concentration, we were able to observe the full range of aggregation structures for the 0.55× concentration (Figure S4). For other concentrations, only early stage (ES; Figure S5) or late stage (RC and F; Figure S6) structures were detected. It is clear that the addition of pAA resulted in a slower aggregation process, and the smallest detectable shape is ES at ≈20 nm. Unlike pAsp, where clusters of comparable size formed after 15 min, the onset of cluster formation is sharply concentrationdependent in pAA-mediated solutions. As a result, no measurable ⟨Rh⟩ could be detected, even after 4 h at high pAA concentrations (greater than 0.6×). Analysis of CaP aggregation kinetics provided additional insights regarding the role of functional polymeric additives (Figure 3), which fall into three different classes (i.e., power law, exponential growth, and complex growth behavior). Control and pAM-mediated solutions produced CaP cluster aggregation that followed power-law growth (⟨Rh⟩ ∼ tα; Figure 3A), further confirming that the presence of amide functional groups alone has no effect on CaP aggregation kinetics. The addition of pAA changed the aggregation to exponential-growth kinetics (⟨Rh⟩ ∼ et) with a strong dependence on polymer concentration (Figure 3B). The pAA-mediated CaP aggregation process can be characterized with the classical reactionlimited colloid aggregation (RLCA) model, with the fractal dimension of later stage aggregates being ≈2.1 (Figures S5 and S6). RLCA aggregation kinetics in calcium phosphate cluster formation has been recently observed elsewhere in CaP without polymer additives,31 suggesting that pAA may not significantly alter the aggregation process. Moreover, ⟨Rh⟩ converged at ≈3 nm when all concentrations were extrapolated to t = 0, further confirming the existence of a single aggregation pathway for pAA-mediated CaP aggregation and possibly suggesting the size of a fundamental CaP cluster. For pAsp-mediated CaP aggregation, no single kinetics model was found to adequately fit. Figure 3C shows two distinct kinetic regimes: particle growth and particle aggregation. In the particle growth regime, particles of ≈7 nm formed after 15 min with all pAsp concentrations and slowly increased in size but did not aggregate. In the aggregation regime, the kinetic profiles deviated from each other and exhibited a strong concentration dependence on pAsp, where increasing pAsp concentration both delayed the start of the aggregation regime and slowed the kinetics of aggregation. The most prominent difference between pAsp- and pAAmediated aggregations is that changing the concentration of pAsp had little effect on the formation of incipient CaP clusters. In LLS, ⟨I⟩ is proportional to particle concentration (c), unit mass (m), and structure factor (P(q)) by the relationship22 I ∼cmP(q)
Figure 3. Measurements of ⟨Rh⟩ reveal aggregation kinetics of CaP particles produced in 5.0 mM CaCl2·2H2O and 2.5 mM KH2PO4. (A) CaP aggregation with no additives or pAM follows power law growth kinetics. (B) pAA-mediated CaP aggregation follows exponential growth kinetics categorized by reaction-limited colloid aggregation (RLCA) that is highly sensitive to pAA concentration. (C) pAspmediated aggregation produces a particle-growth regime and an aggregation regime. The commencement of the first regime is independent of pAsp concentration, with the same incipient particle size of 7−8 nm, whereas the duration of the first regime and the aggregation rate within the second regime depend strongly on the polymer concentration. The dashed-line indicates the transition between the two growth regimes, which is defined as roughly the intercept of the extrapolated growth rate curves within each regime.
conditions, and the unit mass relating to ⟨Rh⟩ and particle structure is largely constant within this regime. Therefore, ⟨I⟩ is roughly proportional to the particle number density (c). ⟨I⟩ and ⟨Rh⟩ values are essentially the same for all pAsp concentrations at 15 min; thus, changes in pAsp concentration had a negligible effect on the particle size and number density of CaP clusters. The same is true at 1.5 h for the three pAsp concentrations remaining in the particle growth regime (0.75×, 1.0×, and 1.25×). Our results support previous reports that the formation of pseudostable clusters stabilized by pAsp allows CaP to be delivered to collagen, thereby facilitating intrafibrillar mineralization.7 By contrast, such behavior does not exist for pAA, where small, stable CaP clusters are notably missing from the scattering data, and particle aggregation occurs as CaP clusters form. To validate LLS’s ability to correctly determine the shape of CaP aggregates, we collected samples for TEM imaging at the same time points as LLS measurements for pAsp-mediated CaP
(1)
where in the first kinetic regime, considering the particle concentration is extremely low (≈10−5 g/mL) and particle size is small (
