Molecular modulation of calcium oxalate crystallization by osteopontin and citrate

Abstract

Calcium oxalate monohydrate (COM), which plays a functional role in plant physiology, is a source of chronic human disease, forming the major inorganic component of kidney stones. Understanding molecular mechanisms of biological control over COM crystallization is central to development of effective stone disease therapies and can help define general strategies for synthesizing biologically inspired materials. To date, research on COM modification by proteins and small molecules has not resolved the molecular-scale control mechanisms. Moreover, because proteins directing COM inhibition have been identified and sequenced, they provide a basis for general physiochemical investigations of biomineralization. Here, we report molecular-scale views of COM modulation by two urinary constituents, the protein osteopontin and citrate, a common therapeutic agent. Combining force microscopy with molecular modeling, we show that each controls growth habit and kinetics by pinning step motion on different faces through specific interactions in which both size and structure determine the effectiveness. Moreover, the results suggest potential for additive effects of simultaneous action by both modifiers to inhibit the overall growth of the crystal and demonstrate the utility of combining molecular imaging and modeling tools for understanding events underlying aberrant crystallization in disease.

Kidney stone disease is a common chronic disorder in humans with the majority of stones being primarily composed of calcium oxalate monohydrate (COM) crystals. COM is the most thermodynamically stable form of calcium oxalate and is the predominant calcium crystallite in many plants (1). Because normal urine is frequently supersaturated with respect to calcium oxalate, urinary crystals are often formed. In most humans, progression from crystalluria to stone disease is prevented by biologic control mechanisms. Normal urine contains inhibitors that decrease the formation, growth, and aggregation of COM crystals (2–4). However, the ways that these normal urinary proteins and small molecules modify COM have not been previously defined at a molecular level. In this article, our investigations using in situ atomic force microscopy (AFM) and molecular modeling provide molecular-scale views of COM modulation by two urinary constituents, a small organic anion, citrate, and a protein, osteopontin (OPN).

Citrate and OPN were chosen for study because urinary levels of both citrate and OPN have been shown to inhibit the growth and change the gross morphology of calcium oxalate crystals (5–7). Citrate is a short-chain nonplanar molecule with three carboxylic acid groups. It is normally secreted into urine and is also widely used in the therapy of renal stone disease (8). OPN is a single-chain protein with a peptide molecular mass of ≈33 kDa. Normal human urine contains levels of OPN (>100 nM) that markedly inhibit several aspects of COM crystallization (9). OPN has an abundance of sequence domains rich in dicarboxylic acids. Because the acidic residues of organic molecules have been implicated in the control of mineralization in a wide range of organisms and mineral systems (10, 11), analysis of modulation of COM by citrate and OPN at a molecular level may also offer general insights concerning the molecular control of biomineralization.

Methods

Seed Crystals and Solution Preparation. The COM crystals used for these studies were grown in vitro by using a gel method. Details of the synthesis are given in ref. 12. For in situ AFM images experiments, supersaturated CaC₂O₄·H₂O solutions (calcium-to-oxalate ratio, 1) were prepared by using reagent grade K₂C₂O₄ and CaCl₂·2H₂O. Relative supersaturation (S, defined below) ranged from 0.2 to 1.2, which are within the physiological range. The pH of all solutions was adjusted to 7.0 before each experiment. Additives (citrate or OPN) in aqueous phase were added to the supersaturated salt solutions for the impurity-doped experiments. Citrate was introduced at concentrations in the range of 0.01 to 0.1 mM, and OPN was used at 1–25 nM, both of which are below the physiological range. , where a(Ca²⁺) and are the activities of the calcium ion and oxalate ion, respectively, and K_sp is the solubility product for COM at room temperature.

Protein Isolation and Purification. OPN was isolated by immunoaffinity purification as described (5). Aliquots of OPN solutions were stored at –80°C before use.

In Situ AFM. In situ images were collected in contact mode (Nanoscope III, Digital Instruments, Santa Barbara, CA) on surfaces of the COM crystal that was anchored inside of the enclosure of a commercially available fluid cell. Supersaturated solution was flowed through the system while the images were taken. Flow rate was adjusted to ensure the surface kinetics was not limited by diffusion. That is, for a given supersaturation, the step speed did not change when flow rate was increased.

To ensure AFM images were authentic representation of the surface morphology, several precautions were taken. First, the imaging force was reduced to the minimum possible value that allowed the tip to remain in contact with the surface so that there was no measurable effect of imaging on the growth kinetics. We verified this by gradually increasing the force until effects on the morphology and step speed were observed or by zooming out to a larger scan box and looking for the signature of the smaller scan area. A consequence of imaging in this regime is that sometimes the tip can momentarily pull off of the surface. This procedure reduces image quality but ensures that the morphology and kinetics of the steps are unaffected. Next, images were also collected at different scan angles, and trace and retrace images were regularly compared, to eliminate the possibility of imaging artifacts induced by tip contamination or tip sticking. Furthermore, step speed was estimated from both images collected at a larger and smaller scale to ensure there was no effect of scanning.

Note that step-angle distortion exists in the images because the step front advances during the scan time. Images reported here are not corrected for this effect.

Molecular Modeling. Energy minimizations were performed to the citrate/COM system by using the Universal Force Field 1.02 with the dielectric constant equal to 80 to account for the water solvation effect. Binding energies of citrate were obtained by subtracting the optimized energy of the adsorbate-surface system from the energies of the surface and the adsorbate, when separated beyond the interaction distance. Flat and stepped surfaces of COM were built by using cerius² (Accelrys, San Diego) surface builder (version 4.2MS) from the unit cell derived from the crystallography data (13). Because hydrogen atoms were not included in the unit cell in ref. 13, their positions were determined computationally by castep (Accelrys) at the GGA-RPBE⁴ level of theory. 3D periodic boundary condition was applied to the COM unit cell. The geometrical dimensions of the cell were fixed at its crystallographically determined values, and all atoms within the unit cell were frozen except the hydrogen atoms, which were free to move. The geometry of the trivalent ion of citrate was optimized by using gaussian 98 software at the Hartree–Fock level of theory (14) using a 6–31+G(d,p) basis set. The electrostatic charges on both the citrate ion and the COM system were assigned by using the method of fitting to molecular electrostatic potentials (15–17) with spartan 5.1.1 software for electronic calculations (Wave-function, Irvine, CA). Optimization calculations were run until convergence criteria, typical for this class of inorganic crystals, were met.

Results and Discussion

The equilibrium shape of pure COM crystals reflects their monoclinic symmetry (13) and is comprised of three face types (Fig. 1 a and b). Their relative sizes show that the {120} faces are the fastest growing, whereas the {–101} faces are the slowest. The miller indices (hkl) denote a single plane and {hkl} represent plane equivalent by symmetry. [UVW] specify a unique vector direction, and 〈UVW〉 represent a set of vectors equivalent by symmetry. Growth of COM crystals in pure solutions occurs on atomic steps generated at dislocation hillocks in a manner comparable to many other crystals growing near equilibrium (18, 19) (Fig. 1 c and d). Hillocks on the (–101) face exhibit three step directions, the crystallographically identical [–120] and [–120] steps, which are parallel to the (–120) apical faces, and the [101] step. The step kinetics on this face are highly anisotropic with the [101] step moving at 12 times the rate of the other two. Expression of the [101] step is not expected from the equilibrium crystal form, which would predict its elimination in favor of the [1-20] and [120] steps corresponding to the other two apical faces (to the left in Fig. 1b). Apparently, these two steps have considerably higher speeds and are therefore unexpressed. This anisotropy reflects the fact that, although all {120} faces are crystallographically identical, the steps on the (–101) face moving toward the two opposed apices are not. Similarly, as Fig. 1f shows for the [101] and [–10–1] steps, the steric presentation of oxalates causes the steps to form acute angles with the underlying terrace in one direction, whereas they form obtuse angles in the other. The geometry of growth hillocks on the (010) face more closely resembles the crystal habit (Fig. 1 b and d), and all step speeds are nearly equal.

Fig. 1.

Images showing the crystal habit and growth hillock morphology of COM. Shown are a scanning electron microscopy image of an ≈100-μm-long COM crystal (a) and the corresponding geometric model (b). The unit cell convention for COM is that of Deganello and Piro (13). (c and d) AFM images of COM growth hillocks in pure solution (supersaturation, S = 0.8) showing a 1.8-μm × 1-μm area for the (–101) face (c) and a 1-μm × 0.8-μm area on the (010) face (d). Step directions are denoted by the arrows. Hillocks in c do not reflect the crystal habit of (–101) face caused by the unexpected [101] step. In contrast, hillocks in d resemble crystal habit of the (010) face. (e and f) Molecular structure of (–101) and (010) crystal faces. The green insert in e shows the relationship between crystal habit and molecular structure when viewed down the [–101] direction. The green insert in f shows the molecular structure of the [101] step viewed down the [010] axis. Dashed lines in e highlight AABBAA stacking perpendicular to the [010] direction. The orientation of the oxalate groups and calcium pairs in layers AA is different from that of layers BB. Although slight, this structural change leads to small differences in step speed between the A and B layers. Because faster steps will always catch up to slower steps, after only a few turns of the dislocation spiral, step bunches displaying the periodicity of the AABB stacking are generated. Thus the unique double-stacking structure along with the change in screw axis orientation at the A/B interface leads to growth of the (010) face on quadruple height steps. Blue, Ca; red, O; black, C; gray, H₂O.

Addition of citrate markedly altered the morphology and kinetics on the (–101) face but had little or no effect on the (010) face. In general for all levels of citrate concentration used, we found that as the concentration increased step-pinning occurred more rapidly and the effect on morphology increased. Complete cessation only took place at the highest citrate level. Fig. 2 a–d shows the temporal evolution of steps at a dislocation hillock on the (–101) face during growth in a solution containing 1.2 × 10^–5 M citrate. Images collected at different scan angle display the same morphological features as shown in Fig. 2 a–d. After introduction of citrate, the [101] step slowed and roughened dramatically because of step pinning, eventually losing lateral stability and reducing speed by a factor of 25. Although the [120] steps also roughened and slowed, this effect was relatively minor. These steps became rounded and the speed dropped only by a factor of 2. As a consequence, at this citrate level, the step speed became a nearly isotropic function of orientation, resulting in a disk-shaped hillock. In contrast, no significant changes in either morphology or step speed were seen on the (010) face, indicating that the citrate-step interactions are much weaker on this face.

Fig. 2.

Representative images (deflection mode) showing the effect of citrate on COM morphology. (a–d) Sequential AFM images during growth in citrate-bearing solution (S = 0.7; accounted for the citrate effect, citrate-to-calcium ratio, 0.1) at t = 3.5 min (a), t = 11 min (b), t = 79 min (c), and t = 103 min (d). Shape of hillock in pure solution is shown in Fig. 1c. (e and f) Scanning electron microscopy images of COM crystals grown in the absence (d) and presence (e) of citrate in solution. Citrate-to-calcium ratio is 0.05. Image horizontal dimensions are: 1.5 μm (a and d), 1 μm (b and c), 100 μm (e), and 25 μm (f).

The macroscopic growth habit of COM, which, in the presence of citrate, becomes disk-shaped (6, 20) (Fig. 2f), parallels the modified shape of the elementary steps on the (–101) face demonstrated here (Fig. 2d). This result suggests that citrate modifies the shape and inhibits the growth of COM as a direct result of step-specific pinning on the (–101) face.

To understand the source of this highly specific interaction, we used molecular modeling with energy minimization to calculate binding energies of citrate docking to steps and faces of COM for several possible configurations. Although strong binding of the carboxyl moieties to the (–101) face has been reported recently (21), our calculations predict that binding energies of citrate to any of the steps expressed during growth are considerably higher than to any face investigated here. Most important, these calculations predict that citrate has the highest binding energy (–166.5 kJ/mol) for the single acute [101] step, i.e., precisely the one that is observed to be most strongly affected by addition of citrate. The predicted geometry of citrate binding is shown in Fig. 3a. These calculations also estimate the average binding energy to the obtuse [–120] step to be –143.6 kJ/mol. There are three possible ways to build the [–120] step. Binding energies at all three configurations were calculated and the arithmetic average is reported here.

Fig. 3.

Geometry of citrate binding to steps in the minimum energy configuration. (a) Citrate molecule bound to an acute single [101] step on (–101) face. (b) Citrate molecule bound to a [–100] step on the (010) face.

When compared to steps on the (–101) face, binding of citrate to steps on the (010) face is much less favorable. For example, the binding energy to the [–100] step is only –90.2 kJ/mol. These results are consistent with the experimental observations that citrate has very little effect on the (010) face. The predicted geometry of citrate binding to the [–100] step is shown in Fig. 3b.

The stereochemistry of citrate binding determined by modeling reveals two important geometric factors. The first is the orientation of the oxalate groups. The dicarboxylic acids at both ends of oxalate molecules present domains that repel the carboxylic acids of the citrate molecule. Thus, steps and terraces that expose flat oxalate configurations are favored. The second factor is the configuration of Ca sites on the crystal surface. The conformation of the nonplanar citrate molecule is relatively rigid so that its three carboxylic acids cannot be easily rotated. Thus a Ca configuration that mimics the carboxylate geometry maximizes the binding energy. The acute geometry of the [101] step on the (–101) face provides a favorable steric configuration for citrate to bind with minimal strain because its geometry optimizes both factors. Specifically, the flat orientation of the oxalate groups on the basal plane avoids electrostatic repulsion of carboxylates, whereas the configuration of Ca sites accommodates all three carboxylic acids, as shown in Fig. 3a. In contrast, on the (010) face, dicarboxylic acids of the oxalates are exposed and extend beyond the (010) plane as demonstrated in Fig. 3b, making it difficult for citrate molecules to bind either to the steps or the face because of the electrostatic repulsion. Moreover, the 90° angle between the basal plane and the step riser results in a poor geometric match between the Ca ions and the citrate carboxylates, a steric condition that can be accommodated only by strain associated with distortion of the citrate molecule.

In striking contrast to citrate, OPN produced major morphological modifications and strong inhibition on the (010) face, but had little effect on the (–101) face. These changes are caused by step-specific OPN interactions leading to pinning. Qualitatively similar effects were seen at all OPN levels (1–25 nM) but were more rapid and quantitatively greater at higher OPN levels. Fig. 4 a–d shows a sequence of AFM images collected on the (010) face during growth in solution containing 5 nM OPN. After OPN was introduced, although the [–101] step was most strongly affected, both the [100] and [–101] steps became strongly pinned and lost lateral stability, with step speeds dropping by an order of magnitude in both directions.

Fig. 4.

AFM images showing the effect of OPN on COM growth hillock morphology. (a–d) Sequential images (horizontal dimension, 1.7 μm) showing OPN modification of the growth hillocks on the (010) face by 5 nM OPN. (e and f) Sequential images (horizontal dimension, 6.0 and 4.5 μm) showing the effect of OPN on growth hillocks morphology of the (–101) face. (g–i) Temporal evolution of step train and protein adsorbates on the (–101) face (horizontal dimension, 2 μm).

However, OPN did not alter either step speeds or morphology on the (–101) face (Fig. 4 e and f). Although interactions of OPN with steps on this face were apparently weak, discrete adsorbates appeared on the (–101) terraces at all OPN concentrations investigated (1–25 nM). The number of adsorbates increased over time without changes in their dimensions, showing that they were not simply 2D nuclei. Although the lateral dimensions among these adsorbates had a wide distribution, their heights were fairly constant at ≈10 Å. We believe these findings show that OPN molecules interact with the terraces strongly enough to form adsorbates, but these bound OPN molecules and those in solution interact weakly with the steps on this face and thus cause no changes in step kinetics or morphology. In fact, the steps propagated freely without interference from the adsorbates on the terraces (Fig. 4 g–i). The adsorbates in these figures remained intact after the passage of growing steps, which apparently move beneath them.

As with citrate, divergent effects of OPN on the (–101) and (010) faces must reflect differences in geometric relationships of functional domains with crystal steps and terraces. Dicarboxylic acid residues aspartic acid-rich proteins are known to be responsible for their strong interaction with crystal faces (11). Phosphorylation of OPN has been also linked to inhibition of COM growth (22). However, protein binding to heterogeneous interfaces reflects both charge density of functional groups and spacing of these groups with relation to the local geometry of the mineral face. Although OPN molecules are highly flexible in 10 mM phosphate solution (23), a fixed conformation may be induced by Ca²⁺ ions (5) [or by binding to a mineral surface, as suggested by the apparently inert nature of the OPN adsorbates on (–101) terraces].

Among the local step characteristics that should impact OPN binding is the height. On the (–101) face, step heights are only 6.0 Å. However, a screw-axis symmetry element of COM perpendicular to (010) creates an AABBAA stacking sequence that leads to periodic bunching of steps through a phenomenon known as step interlacing (24, 25). This results in growth on quadruple steps with heights of 16 Å (Fig. 1e), suggesting that, in contrast to the single step heights seen on the (–101) face, the much greater height of the quadruple steps on the (010) face facilitates binding of a large number of carboxylic acid and phosphate groups of OPN to the step riser and the basal plane, leading to a strong OPN–step interaction that pins the steps. On the (–101) face, where the step risers may not be tall enough to satisfy the steric requirement for the binding of OPN anionic groups, a weak binding of OPN molecules to the planar terrace may well predominate because of the larger availability of binding sites.

The field of molecular analysis of biologic processes is evolving into a predictive science. The combination of nanoscale imaging and computational modeling helps to drive this evolution by providing a quantitative basis for new insights. In this work, in situ AFM and molecular modeling were used to develop a physical model for the molecular-scale mechanism by which two urinary constituents modulate the morphology and growth kinetics of calcium oxalate crystals. Specifically, we have shown that strong interactions between these control molecules and individual steps on existing crystal faces result in step pinning, modification of step kinetics, and consequent growth inhibition and shape modification. Moreover, the effects of OPN and citrate on the steps of different faces suggest the potential for additive effects when these modulators are combined. This picture of biomineral control extends the principles of biomineralization beyond those developed on the basis of bulk crystallization experiments from which the modification of crystal habit was attributed to differential binding of growth modulators to the various faces of COM. In the case of citrate-based compounds, strong binding to the (–101) plane was thought to stabilize that face by reducing the surface energy (6), a thermodynamic picture that has also been used to explain shape modification in many other systems (10, 26). Our results show that, for COM, a purely kinetic picture of modification is more consistent with experimental observations.