Hydroxycitrate is a Unique Inhibitor of Pathologically-Relevant Brushite Crystallization

Abstract

The unique interactions between (macro)molecules and the surfaces of crystalline materials is a common mechanism for controlling the formation of natural, biological, and synthetic crystals. These species, generally referred to as modifiers, can either inhibit or promote crystallization in ways that achieve properties unattainable by alternative processes, or alter nucleation to produce metastable polymorphs or solvates. In the field of pathological crystallization, modifiers can be native species or rationally designed molecules for the therapeutic purpose of suppressing crystal nucleation and/or growth. Here, we assess the effects of hydroxycitrate as a modifier of brushite (calcium phosphate dihydrate) crystallization and demonstrate its superior efficacy as an inhibitor compared to citrate, its molecular analogue present in urine naturally and commonly administered as a preventative therapy for calcium-based kidney stones. Using a combination of bulk crystallization, microfluidics, and atomic force microscopy, we show that hydroxycitrate is capable of completely suppressing brushite nucleation and operates as an effective growth inhibitor at concentrations nearly one order of magnitude lower than citrate. Our findings reveal a concentration-dependent mode of action for hydroxycitrate which at low supersaturation induces the formation of defects (i.e., surface voids) far from nucleation centers that are not observed at high supersaturation. Kinetic studies reveal that hydroxycitrate fully suppresses two of the three principal growth directions, whereas citrate only suppresses one direction. For crystallographic growth directions where modifiers are moderately effective, both citrate and hydroxycitrate inhibit crystallization by as much as 60%, leading to markedly reduced rates of growth and concomitant changes in crystal habit. Collectively, the results of this study open new avenues into the design of potent inhibitors of pathological crystallization with potential implications for other biominerals.

Graphical Abstract

1. Introduction

Brushite (CaHPO₄·2H₂O) is a key calcium phosphate phase involved in kidney stone formation. While calcium oxalate (CaOx) is the most common constituent of renal calculi, brushite plays a crucial role as a primary precursor in pathological soft tissue mineralization in addition to renal calculi.^{1, 2} It has been identified in early-stage mineral deposits within the renal papillae, particularly in Randall’s plaques,³ where it can initiate and facilitate heterogeneous nucleation of both hydroxyapatite (HAP) and CaOx crystals^{4, 5} contributing to stone formation and growth. Furthermore, brushite stones exhibit a higher recurrence rate than other kidney stone types,^{6, 7} posing a greater challenge for the long-term treatment of urolithiasis. Thus, understanding the mechanisms of brushite formation and identifying growth inhibitors is essential for developing targeted strategies against kidney stone formation.^{2, 8}

High urine pH and low urine citrate (hypocitraturia) are two clinical parameters that distinguish the urine chemistry phenotype of brushite stone formers.⁹ At higher urine pH, dibasic phosphate (HPO₄^2-) is the predominant phosphate species, and more readily complexes calcium to form insoluble brushite. Citrate in urine inhibits brushite stone formation through multiple mechanisms, including the formation of soluble citrate-calcium complexes, and potentially through inhibition of brushite crystal formation and growth.¹⁰ Hypocitraturia is present in 20 to 60% of patients with calcium kidney stones.^{11, 12} A retrospective analysis of metabolic profiles from 1988–1994 and then 2007–2010 found an increase in prevalence of hypocitraturia among stone formers, from 46 to 60%.¹¹ In a large cohort with a history of kidney stones, Ganesan et al. reported that 20% of patients had hypocitraturia (<1.04 mmol/day), highlighting the burden of this abnormality in contemporary stone formers.¹³ In the case of calcium phosphate stones, both urinary pH and citrate are predictors of stone formation.¹⁴

One approach to kidney stone prevention involves the use of exogenous substances, which incorporate into the stone’s crystal structure and inhibit growth via various pathways.^{15, 16} An extensive range of natural and synthetic inhibitors for biological mineralization has been studied for systems involving CaOx,^17–19 struvite,^{20, 21} and urate kidney stones.^{22, 23} Furthermore, over the last 30 years, there has been an increase in the prevalence of calcium phosphate kidney stones in humans,^{6, 14, 24} which has led to a rising interest in identifying inhibitors of brushite crystallization, including metal ions,^{25, 26} small molecules,^27–29 and protein-based inhibitors.^{30, 31} For an inhibitor to work as a therapeutic agent, it must be excreted in adequate amounts in urine. Ionized anionic citrate (CA) from citric acid is a naturally occurring component of human urine, and is known to inhibit the growth of calcium oxalate monohydrate (COM)³² and dihydrate (COD),³³ two prevalent constituents of human kidney stones. Provision of CA in the form of oral potassium citrate therapy significantly increases urinary citrate excretion, but also increases urine pH, which may negate the beneficial effects of citrate on brushite saturation.³⁴ Moreover, the usefulness of potassium citrate in treating calcium phosphate stones is not entirely clear.³⁵

Hydroxycitrate (HCA) has a structure similar to CA with an additional hydroxyl group on the main carbon chain. Our previous study has shown that significant urinary excretion of HCA can be achieved in humans by administering relatively low doses of the drug.³⁶ HCA was found to be a more effective inhibitor of COM crystallization than CA via a non-conventional mechanism involving strain-induced layer retraction and etch pit formation on COM crystal surfaces. Moreover, HCA does not seem to have a significant effect on urine pH, furthering its potential as a kidney stone therapeutic, particularly in patients with existing alkaline urine.²⁴ Li et al. studied the effect of CA and HCA on the brushite (010) surface, showing an inhibitory effect on step velocity and a decrease in step density.²⁸ A more detailed examination on the effect of CA, HCA, and their combination on bulk brushite crystallization and surface dynamics over a wider range of supersaturation may provide useful insights on the potential of HCA as a therapy for kidney stones.

Here, we investigate the effects of CA vs HCA on brushite crystallization in a range of equimolar calcium phosphate concentrations (CaP) using a combination of bulk crystallization, nucleation studies, in situ microfluidics, and atomic force microscopy (AFM). Bulk studies reveal that both CA and HCA modify crystal morphology, with HCA demonstrating a much higher inhibition efficacy that leads to full suppression of brushite nucleation. Microfluidic experiments on single brushite crystals show that CA acts as a weak, direction-dependent inhibitor, whereas HCA exhibits strong inhibition across multiple crystallographic directions. Furthermore, in situ AFM imaging of the brushite (010) surface reveals that at high supersaturation, CA and HCA exhibit similar inhibitory effects, whereas at low supersaturation, HCA uniquely suppresses step growth through defect formation. Studies combining both inhibitors revealed antagonistic cooperativity, with HCA dictating inhibitory effects irrespective of the CA content. Our collective findings provide mechanistic insights into the inhibition of brushite crystallization by CA and HCA and highlight potential strategies for controlling pathological mineralization.

2. Experimental Methods

2.1. Materials.

The following compounds were purchased from Sigma Aldrich (St. Louis, MO): Calcium chloride dihydrate (CaCl₂∙2H₂O), potassium dihydrogen phosphate (KH₂PO₄), sodium chloride salt (NaCl), sodium citrate tribasic dihydrate (C₆H₅Na₃O₇∙2H₂O), potassium hydroxycitrate tribasic monohydrate (C₆H₅K₃O₈∙H₂O), and sodium hydroxide (NaOH) pellets. All reagents were used as received without further purification. Deionized (DI) water was produced by a Millipore reverse osmosis – ion exchange system (Rios-8 Proguard 2 – MilliQ Q-guard).

2.2. Bulk crystal precipitation and monitoring time evolution of pH.

Bulk brushite crystals were precipitated by reactive crystallization methods where CaCl₂ and KH₂PO₄ were allowed to react to form brushite (CaHPO₄∙2H₂O) crystals. Solution pH was adjusted to 5.6 using a 1 M NaOH stock solution and NaCl was used to maintain the ionic strength within a physiological range. Bulk brushite crystallization in batch conditions were monitored (i) without or (ii) with stirring to understand the effect of the modifiers on crystal morphology, nucleation and crystallization kinetics. All the solutions used for these experiments were prepared with 125 mM NaCl at pH 5.6 in the presence and absence of modifiers using equimolar calcium phosphate concentrations (CaP), where CaP refers to the total concentration of calcium ions (equal to that of the phosphate ions). Without stirring, crystals were precipitated in 20 mL glass vials at a control solution of 20 mM CaP. The crystal morphology was assessed at varying modifier concentrations after incubating growth solutions over 48 hours. In another assay, solution pH was continuously monitored to understand the effect of the modifiers on the crystallization kinetics in a continuously stirred system in a 20 mL glass vial using 16 mM CaP along with different modifier concentrations. The stirring rate and the control concentration were optimized based on the time required for the solution to nucleate and grow at a reasonable rate for real time measurements. The pH decreased during the extent of crystallization, from its starting value of pH 5.6 until reaching equilibrium at pH 4.9 (control solution). We calculated the extent of reaction based on the instantaneous pH drop in the presence of modifiers compared to the control solution in the range of pH 5.6 to 4.9. These rates were compared to that of the control to calculate the precent inhibition for each modifier. We also studied the cooperative effects of using binary combinations of HCA and CA through an isobolograph analysis wherein we plotted the concentrations of each individual modifier for a given inhibitory effect. Lines are drawn between each individual modifier concentration to establish the regions of cooperativity. Any combination of binary modifiers having similar effects of individual modifiers exhibit points along the line, while those with synergistic or antagonistic behavior fall below and above the line, respectively.

2.3. In situ monitoring of the bulk crystal growth.

We used a microfluidic device described in previous studies^{20, 37} for in situ analysis of bulk crystal growth in a continuous flow system using an inverted optical microscope (Leica DMi8 instrument). All solutions used for these experiments contained equimolar CaP and 125 mM NaCl (at pH 5.6), and were supplied to the device with a syringe pump operated under a continuous flowrate of 6 mL h⁻¹. We prepared two different syringes with one containing calcium and the other containing phosphate, NaCl, and modifier. During in-line mixing, the solutions reached the desired concentrations. To seed the microchannels, we used a higher supersaturation (25 mM CaP) and observed plate-like crystals after 10–15 minutes. For crystallization assays, the growth solution was switched to a lower supersaturation (20 mM CaP) to avoid nucleation and only monitor seed growth. In situ experiments were performed over a 90-minute period to track the linear growth rate of [101] and [$10\stackrel{-}{1}$] directions (in-plane growth). The control solution was replaced after 20 minutes of imaging with the same solution containing a modifier, wherein the experiment was continued for another 60 minutes. We calculated the relative growth rate with modifiers for two in-plane directions compared to the control growth rate at the initial times. For statistical sampling, growth rates were measured for 10 to 15 crystals for each population at different modifier concentrations.

2.4. In situ monitoring of surface dynamics.

We used a Multimode 8 (Bruker) for all atomic force microscopy (AFM) experiments. Images were taken in contact mode over 20 minutes using a ScanAsyst-Fluid probe (silicon nitride, 150 kHz, 0.7 N/m). Solutions prepared for AFM were similar to those of bulk assays (equimolar CaP, 125 mM NaCl, pH 5.6) using a syringe pump to supply growth solutions at a continuous flowrate of 6 mL h⁻¹ and a similar dual syringe in-flow mixing setup like microfluidics described above. We prepared samples on AFM disks (Ted Pella Inc.) with clear quickset Gorilla epoxy and kept it overnight in 20 mL glass vials to seed brushite crystals at 20 mM CaP. We measured step velocities in different directions for the control and compared the results to solutions with modifier to elucidate their effect(s) on surface growth dynamics.

3. Results and Discussion

3.1. Macroscopic effects of citrate-class inhibitors on brushite crystallization.

In the literature there are several reported morphologies of brushite (CaHPO₄∙2H₂O) crystals. In this study we selected a nominal growth condition similar to the one reported by Li et al.²⁸ using aqueous solutions of equimolar calcium and phosphate concentrations (16 mM CaP) at pH 5.6 and 125 mM NaCl. These conditions mimic two properties of physiological environments (i.e., tubular lumen in kidneys) where the pH ranges from 5 to 8 and the ionic strength is around 150 mM.^38–40 The supersaturation of calcium phosphate used in these studies, however, is intentionally higher than physiological ranges (2 – 3.5 mM calcium and 10 – 30 mM phosphate)¹⁴ to enable measurements of crystal growth at reasonable timescales. We used an even higher supersaturation (25 mM CaP) to obtain large, plate-like brushite crystals (Figure 1A) with basal (010) surfaces as substances or seeds for in situ measurements (vide infra). Phosphates exist in one of four species having either neutral (H₃PO₄) or negative charges (H₂PO₄^–, HPO₄^2–, PO₄^3–) based on three dissociation pKa values of 2.1, 7.2 and 12, respectively.⁴¹ The population of phosphate species as a function of pH (Figure 1B) reveals that the native form of brushite (HPO₄^2–) is more prevalent at the upper range of physiological pH, whereas the condition selected for these studies (Figure 1B, dashed line) predominantly contains H₂PO₄^–. Attempts to increase the pH of the growth medium result in dramatic changes to the final product. Notably, we observe precipitates with ultrasmall sizes (5–10 μm) with spheroidal morphology (Figure S1). In order to produce crystals with sufficiently large sizes for both optical and atomic force microscopy measurements, we chose to work with a pH 5.6 growth medium.

Figure 1.

(A) Representative scanning electron microscopy (SEM) image of a brushite crystal prepared at room temperature from a 25 mM CaP solution with 125 mM NaCl at pH 5.6. (B) Speciation profiles for phosphate as a function of solution pH. The fraction of each species was calculated using a combination of mass balances and speciation reaction equations, taking into account the ionic strength (see Supporting Information, SI, for details). The vertical dashed line highlights the condition selected for all experimental measurements in this study (pH 5.6). (C) Bulk assays using pH to track the extent of brushite crystallization in the absence (control) and presence of modifiers. The pH was monitored for 20 hours under continuous stirring for solutions prepared with an initial composition of 16 mM CaP and 125 mM NaCl. (D) Structures of the two crystal growth modifiers tested in this study: citrate (CA) and hydroxycitrate (HCA).

We first performed bulk assays to monitor the extent of crystallization under continuous stirring wherein we used a pH meter to indirectly track the temporal depletion of solute (e.g., phosphate) from initially supersaturated solutions (16 mM CaP). Our studies revealed that this approach is more sensitive than conventional methods of tracking the depletion of calcium⁴² using an ion selective electrode (ISE). During brushite crystallization in the initially acidic medium, the first step in crystallization is the deprotonation of H₂PO₄^– to HPO₄^2– (Figure 1C, inset), which releases hydronium ions that lower the pH of the solution. Over the course of crystallization, we observe a monotonic reduction in pH (Figure 1C, control) until reaching saturation (plateau) around pH 4.9 within 1 hour. These changes reflect the temporal consumption of solute during collective processes of crystal nucleation and growth rather than an explicit rate of crystallization. This procedure was repeated for the same solutions containing different concentrations of the two growth modifiers assessed in this study (Figure 1D): citrate (CA) and hydroxycitrate (HCA). The introduction of modifiers has three major effects on brushite crystallization: (i) extended induction time, (ii) reduced rate of crystallization; and (iii) shifts to higher pH in the plateau region at later crystallization times. The latter is a common thermodynamic effect^43–45 associated with modifier – solute complexation that hinders the system from reaching solubility. Comparison of CA (at 340 μM) and HCA (at 29 μM) reveals that the latter is a significantly more effective and potent inhibitor of brushite crystallization (Figure 1C).

The kinetics of brushite crystallization was evaluated using the extent of bulk crystallization calculated from differences between the instantaneous and final pH. A comparison of the extent of crystallization for HCA (Figure 2A) and CA (Figure 2B) over a range of modifier concentrations reveals that HCA is effective at quantities an order of magnitude less than that of CA. In the absence of modifier (control), crystallization is complete within 50 min. This steady-state condition (at pH 4.9) is defined as complete (100%) crystallization. The addition of a modifier reduces the rate of crystallization and extends the induction time (i.e., period before the onset of nucleation). The latter is estimated as the time for the initial pH of 5.6 to reduce to a value of 5.55, which was selected as the approximate limit of experimental error for the pH probe to ensure statistical relevance. As shown in Figure 2C, both modifiers increase the induction time; however, HCA extends this period by nearly 6-fold longer than CA, indicating that HCA is a more effective inhibitor of brushite nucleation. When we increase the concentration of HCA above 50 μM, we do not observe any detectable reduction in pH for more than 24 hours (Figure S2). We recently reported a similar observation for COM crystallization using phosphate-based modifiers⁴² where the mechanism for complete suppression of COM formation was attributed to the modifier’s ability to disrupt a two-step (nonclassical) nucleation pathway – arguing that modifiers are theoretically incapable of fully suppressing nucleation via a classical pathway. Here, we propose a similar line of reasoning to rationalize the effect of HCA on brushite nucleation, although its exact mechanism is not fully understood and remains a subject of ongoing investigation.

Figure 2.

Kinetic studies of brushite crystallization at varying modifier concentrations in continuously stirred solutions prepared with 16 mM CaP, 125 mM NaCl, and an initial pH of 5.6. The induction period was set equal to the time when the solution reached a pH of 5.55, which also marked the initial time for measuring the extent of crystallization. (A, B) Time evolution of the extent of crystallization at varying concentrations of (A) HCA and (B) CA (see the supplementary methods for additional details). (C) Nucleation induction times of brushite and (D) percent inhibition of brushite crystallization as a function of modifier concentration for HCA (red triangles) and CA (blue squares). Data are the averages of 3 independent measurements; error bars span two standard deviations; and dashed lines are interpolated to guide visualization.

The extent of crystallization was used to calculate the percent inhibition of brushite crystallization as a function of modifier concentration. For these studies we used a solute concentration of 16 mM CaP, which corresponds to a supersaturation ratio S = 2.13 (where S = C/C_e with solubility C_e = 7.5 mM CaP at 21°C). As shown in Figure 2D, HCA is a more potent and effective inhibitor compared to CA. At a concentration of 3.5 μM, HCA can achieve near maximum inhibition with full suppression of nucleation (i.e., 100% inhibition) occurring around 30 μM HCA. On the contrary, CA cannot fully suppress crystallization and reaches its maximum inhibitory effect (ca. 90%) at a concentration that is nearly an order of magnitude higher than that of HCA. Bulk crystallization assays were performed using similar concentrations of HCA and CA to determine their effects on crystal morphology; however, in order to obtain a sufficient number of crystals for statistical analysis, the concentration of solute was increased to 25 mM (S = 3.33). In the presence of both modifiers, the morphology shifts from large platelets (Figure 1A) to rod-like crystals (Figure 3). There are clear differences in the habit of each rod-like crystal, with those prepared by HCA having more distinct facets (Figure 3A) and those prepared by CA exhibiting tapered ends (Figure 3B). One unique aspect of CA is its propensity to induce aggregation over a broad range of modifier concentrations (Figure S3), with many arranged in a radial (flower-like) pattern.

Figure 3.

Optical micrographs of brushite crystals formed in the presence of (A) 15 μM HCA and (B) 18 μM CA (see Figure S3 for micrographs at different modifier concentrations). Images were taken 24 hours after incubation in 20 mL vials at quiescent conditions. Supersaturated solutions were prepared with 25 mM CaP and 125 mM NaCl at initial pH 5.6.

3.2. Impact of inhibitors on macroscopic crystallization.

Here we used a previously described microfluidics device³⁷ to track the dynamics of seeded brushite growth while flowing supersaturated calcium phosphate solutions containing each modifier at a fixed volumetric flow rate (6 mL h⁻¹, within the reaction-limited regime). Optical micrographs in real time were used to extract growth rates along in-plane [101] and [$10\stackrel{-}{1}$] directions (Figure 1A), while out-of-plane growth along the [010] direction was measured by an alternative technique (vide infra). Microchannels were first seeded using a supersaturated solution (S = 3.33, 25 mM CaP). Once a sufficient number of crystals formed, the solution was switched to one with lower supersaturation (S = 2.67, 20 mM CaP) to avoid homogeneous nucleation and only promote seeded growth. Plate-like crystals with an average width of 100 μm ([101] direction) form within 15 min of seeding. Growth measurements at lower supersaturation were then performed for 20 min prior to each measurement with a modifier(s). Crystallization in the presence or absence of modifiers (control) was monitored over a 60 min period and revealed continuous growth of each in-plane direction (Figures 4A and S4). The same experiment performed in the presence of CA reveals an asymmetrical effect (Figure 4B) wherein growth along the [101] direction is fully suppressed while growth along the [$10\stackrel{-}{1}$] occurs at a slower rate than the control. In the case of HCA, we observe full suppression of growth in both in-plane directions (Figure 4C); however, its effect is more pronounced on the [101] direction.

Figure 4.

(A – C) Time-resolved images of brushite crystals growing in a microfluidic channel under constant flow (6 mL h⁻¹) of supersaturated growth solution in the (A) absence of modifier (control) and in the presence of (B) 850 μM CA and (C) 734 μM HCA. Solutions were prepared with 20 mM CaP and 125 mM NaCl at pH 5.6. Images at times t = 0 and 60 min were extracted from Movies S1 – S3. (D) Growth inhibition of brushite crystals in the [101] direction (closed symbols) and [$10\stackrel{-}{1}$] direction (open symbols) as a function of modifier concentration. Symbols represent the average of at least 10 measurements on different crystals in a single experiment; error bars span two standard deviations; and dashed lines are interpolated to guide the eye.

Bulk crystallization studies presented in Figures 1 and 2 cannot uniquely distinguish between the effects of modifiers on nucleation versus growth. Although the extent of crystallization is measured after the induction period, nucleation can still occur at later times. In microfluidics measurements, these events are decoupled by seeding wherein we focus solely on the modifier’s effect on brushite growth. Under these conditions, we find that the maximum inhibitory effects require much higher modifier concentrations (Figure 4D) than those of bulk assays. The 100% inhibitory effects of HCA along the [101] and [$10\stackrel{-}{1}$] directions occur at concentrations of 20 and 730 μM, respectively. Full suppression of growth along the [101] direction is also achieved by CA, but at a slightly higher concentration of 35 μM. The effect of CA on brushite growth along the [$10\stackrel{-}{1}$] direction is significantly less, reaching a maximum of 57% inhibition at a concentration that is well within physiological limits of CA in humans (990 – 1830 μM).¹⁴ The disparate efficacies of modifiers on each in-plane growth direction also results in anisotropic changes to the crystal morphology.

3.3. Impact of inhibitors on microscopic surface growth.

Here we used atomic force microscopy (AFM) to measure the dynamics of layered growth on basal (010) surfaces of brushite crystals where the nucleation and advancement of layers provide information for growth in the [010] direction, normal to the imaging area (i.e., out-of-plane growth). For these experiments we tested two different supersaturations (S = 1.33 and 1.67), both of which are less than the calcium phosphate concentration used in microfluidics to achieve slower rates of growth, which was necessary to track the dynamics of surface growth in real time. At the lower supersaturation (10 mM CaP), the (010) surface of brushite grows classically by monomer addition with layers advancing in a spiral growth pattern from screw dislocations. We first monitored the velocity of step advancement v_o for a control solution in the [101] and [$10\stackrel{-}{1}$] directions (Figure 5A) using sequential AFM images from Movie S4. Under these growth conditions, we observe a high density of step bunches that prevent measurement of step velocity in the [$\stackrel{-}{1}00$] direction (Figure 5A). After initial periods of growth in the absence of modifier, the solution was switched to one containing either CA (Figure 5B) or HCA (Figure S5) to observe their effect on step velocities v in both directions. In situ AFM measurements revealed no change in step growth along the [$10\stackrel{-}{1}$] direction; however, we did observe that both CA and HCA inhibit growth along the [101] direction, leading to a higher density of steps in close proximity to the screw dislocation (Figure 5B and Movie S5).

Figure 5.

(A, B) AFM deflection mode images of a triangular hillock emanating from a screw dislocation on a brushite (010) surface with steps advancing in the [101] and [10–1] directions. A supersaturated growth solution (10 mM CaP, 125 mM NaCl, pH 5.6) was continuously supplied to the liquid sample holder at a flow rate of 6 mL h⁻¹. Images extracted from Movies S4 and S5 show surface growth (A) without modifier (control) and (B) in presence of 340 μM citrate, respectively. Scale bars are identical for both images. (C) Relative step velocities in the [101] and [$10\stackrel{-}{1}$] directions (open and closed symbols, respectively) for solutions with CA (blue) and HCA (red) as a function of modifier concentration. (D) Idealized schematic of surface growth inhibition by a kink blocking mechanism.

In Figure 5C we plot the relative step velocity (v/v₀) as a function of modifier concentration. The maximum level of growth inhibition in the [101] direction is comparable for HCA (75%) and CA (70%). The observation of a plateau in v/v₀ is characteristic of kink blocking (Figure 5D), which is a common mechanism of surface growth inhibition.^{22, 46} The advancement of layers on basal (010) surfaces dictates the rate of growth normal to the imaging surface, or [010] direction, which was not assessed in microfluidics (Figure 4D). The combined set of AFM and microfluidics data indicate that HCA can fully suppress growth in the [101] and [$10\stackrel{-}{1}$] directions, whereas CA only blocks growth in the [101] direction. The inability of either modifier to fully suppress growth in the b-direction reveals that brushite platelets become thicker, but at significantly reduced rates of growth.

The v/v₀ data was evaluated in close proximity to the nucleation center (Figure 6A, dashed box); however, when expanding the AFM imaging area we observe a unique effect of HCA on layer advancement farther from the screw dislocation. The appearance of voids reveals a different mechanism of inhibition not previously observed for brushite or related minerals. As steps advance in the [$10\stackrel{-}{1}$] direction, we show in Figure 5C that their rate of growth near the screw dislocation is unaffected by HCA, but farther from the center where interstep distances are larger, leading to reductions in step velocity, we observe the formation of step bunches or macrosteps (Figure 6B, box). Layers continue to grow around these features, creating voids that increase in depth with continued imaging. This is evident in time-resolved height profiles (Figure 6C) where measurements of step velocity in this region reveal a distribution of v/v₀ values that seemingly depend on the density of voids, which impose a maximum of around 50% inhibition based on measurements of step velocities in close proximity to the voids. It is interesting to note that voids were not observed during AFM experiments using CA as the modifier.

Figure 6.

(A) AFM deflection mode image extracted from Movie S6 capturing a brushite (010) surface growing in a low supersaturation solution (10 mM CaP, 125 mM NaCl, pH 5.6) containing 15 μM HCA. The dashed yellow box highlights the region in close proximity to a screw dislocation. (B) Images extracted from Movie S7 at 0 min (left) and 12 min (right) show the emergence of voids. (C) Time-resolved height profiles of the void highlighted in panel B (white box) where the profiles are offset using the minimum height as a reference point.

The exact mechanism of void generation is not well understood. Step bunches are often created by a step pinning mode of action⁴⁷ that leads to the complete suppression of layer advancement for materials that grow by surface diffusion. Here we do not observe direct evidence of step pinning. Interestingly, when the growth solution is replaced with one having a higher supersaturation (i.e., switching from 10 to 12.5 mM CaP), we no longer observe voids during AFM measurements (Figure 7A and B). At higher supersaturation, we also observe growth in all three directions without the appearance of step bunches that were shown to impede growth in the [$\stackrel{-}{1}00$] direction at lower supersaturation (Figure 5A). The performance of HCA and CA on brushite (010) layer growth is comparable at higher supersaturation, even for AFM measurements at 20 mM calcium phosphate (Figure S6), which matches the conditions for microfluidics assays. Steps advancement is unhindered in the [$\stackrel{-}{1}00$] and [$10\stackrel{-}{1}$] directions, while growth along the [101] direction is reduced by a maximum of 60% (Figure 7C). This value is slightly less than the efficacy of both modifiers assessed at lower supersaturation (Figure 5C). Notably, both conditions examined in this study are considerably more supersaturated than typical human urine, in which calcium concentrations are approximately 2 mM in non-stone formers and 3.5 mM in stone formers.¹⁴ This suggests that the mechanism observed at lower supersaturation in this study (Figure 6) is more reflective of what could be expected in physiological environments. It should also be noted that the urine of stone formers exhibits a phosphate-to-calcium ratio of 1.48, which is lower than the 2.25 ratio observed in non-stone formers and more closely aligns with our equimolar experimental conditions.

Figure 7.

(A) AFM deflection mode image extracted from Movie S8 capturing a brushite (010) surface growing in a higher supersaturation solution (12.5 mM CaP, 125 mM NaCl, pH 5.6). Images are shown in the (A) absence of modifier and (B) presence of 440 μM HCA. All scale bars equal 1 μm. Similar measurements with CA are reported in Figure S7 of the SI. (C) Relative step velocities v/v₀ in all three directions of step advancement in the presence of 510 μM CA (blue) and 440 μM HCA (red). Data are the average of five measurements in a single experiment and error bars span two standard deviations.

3.4. Cooperative effects of binary growth modifiers.

In physiological conditions, CA is a naturally occurring molecule in human urine whereas HCA is not.^{48, 49} This leads to questions regarding the cooperative effects of both modifiers. In prior studies we used an isobologram analysis^{17, 50} to determine whether binary combinations of modifiers exhibit antagonistic, additive, or synergistic cooperativity. This method involves the generation of percent inhibition plots as a function of each singular modifier, which is shown in Figure 2C (and expanded in Figure S8) for HCA and CA. We performed the same measurements using a binary combination of modifiers with different n CA: 1 HCA ratios (Figure S9), with n = 12, 23, and 46. The resulting isobologram in Figure 8 compares the efficacy of each individual modifier as open symbols for three levels of growth inhibition: 30% (green triangles), 50% (blue squares), and 80% (orange circles). Solid lines connecting each set of individual modifiers show an approximate order of magnitude difference in efficacy observed in bulk crystallization assays. The solid symbols are binary combinations for each n CA/HCA ratio (dashed line) where the HCA concentration is fixed and the CA concentration is varied. The inset outlines how to interpret the plot, with solid symbols on the line referring to additive effects, and those below and above the line corresponding to synergistic and antagonistic interactions, respectively. For each set of data (shaded regions), we observe that the cooperative effect is by definition antagonistic; however, the fact that the efficacy is unchanged with increasing CA concentration indicates the mechanism of brushite growth inhibition is controlled entirely by HCA. This indicates that the presence of CA does not enhance crystal growth inhibition, but also does not interfere with the effects of HCA. This is an important consideration for the potential therapeutic translation of HCA. To our knowledge, this is the first time such behavior has been reported for a binary combination of crystal growth modifiers. It is more common in the literature to observe antagonism resulting from one adsorbed species negatively impacting the effects of the other adsorbed species.⁵⁰ However, this is clearly not the case for binary combinations in this study where the presence of the more potent modifier (HCA) nullifies the effect of the other modifier (CA).

Figure 8.

Isobologram analysis of binary combinations of citrate (CA) and hydroxycitrate (HCA) as inhibitors of brushite bulk crystallization. Open symbols indicate the concentrations of individual inhibitors from Figure S8 that elicit a certain percentage of inhibition: 30% (green triangles), 50% (blue squares), and 80% (orange circles). Solid lines connecting individual modifiers demarcate the regions of cooperativity, as outlined in the scheme (lower right inset) illustrating how solid symbols on the solid line correspond to an additive effect, whereas solid symbols below and above the solid line correspond to synergistic and antagonistic effects, respectively. Open symbols represent the concentrations of the paired inhibitors at different n CA: 1 HCA ratios (dashed lines) where n = 12, 23, and 46. Shaded regions correspond to different n CA: 1 HCA ratios at a fixed HCA concentration.

4. Conclusions

In summary, we have investigated the effects of citrate and its molecular analogue hydroxycitrate as modifiers of brushite crystallization at different growth conditions using various in situ techniques to probe their efficacy over multiple length scales. Bulk crystallization assays reveal that hydroxycitrate is more potent and is also capable of suppressing brushite nucleation using concentrations less than 5 μM. The observed shift in bulk crystal habit from a platelet to needle-like morphology in the presence of both inhibitors indicates their preferential interactions with select brushite crystal surfaces. This was confirmed using a combination of microfluidics and atomic force microscopy to monitor the effect of modifiers on both macroscopic and microscopic growth rates, respectively. Microfluidics confirmed that hydroxycitrate fully suppresses growth along the [101] and [$10\stackrel{-}{1}$] directions, but with preferred inhibition of the former. On the contrary, citrate is only capable of fully suppressing growth in the [101] direction. Time-resolved AFM measurements reveal both modifiers have nearly equal impact on the kinetics of layer advancement wherein their partial suppression of step velocity in the [101] direction indicates citrate and hydroxycitrate are unable to fully suppress growth in the b-direction (out-of-plane). Comparison of layered growth at different calcium phosphate concentrations reveals a unique effect of hydroxycitrate that is only observed at lower supersaturation. AFM measurements in regions near screw dislocations (nucleation centers) indicate both modifiers exhibit a kink blocking mechanism; however, at low supersaturation we observe a unique effect far from the nucleation center in the presence of hydroxycitrate where macrosteps are formed, slowing the rate of growth and creating voids by a mechanism that is not fully understood, but resembles step pinning, which is a common mode of surface growth inhibition. Prior studies of hydroxycitrate interactions with calcium minerals (e.g., calcium oxalate) have demonstrated that the additional hydroxyl group relative to citrate alters the modifier’s conformation and hydrogen bonding capability when bound to crystal surfaces, which can induce strain in the lattice.^{36, 51} It is possible that similar effects may occur for brushite, but a molecular-level understanding of these interactions and resulting inhibition mechanisms will require future computational studies, which is outside the scope of this work.

The presence of citrate in physiological environments for brushite formation poses the question about potential antagonistic cooperativity with hydroxycitrate given their similar inhibitory mechanisms of crystallization. To assess this, we performed bulk crystallization assays using binary combinations of both modifiers at different molar ratios. Our findings reveal that hydroxycitrate dominates the overall performance of growth inhibition, irrespective of the amount of citrate present. This is a promising result for future exploration of hydroxycitrate as a potential therapeutic for pathological crystallization. These collective insights provide a foundation for understanding the role of modifiers and their ability to counter the formation of brushite, one of the most pervasive forms of kidney stones.

Supplementary Material

Supporting Information is linked to the online version of the paper.

Detailed methods for calculating phosphate speciation, table of speciation reactions and dissociation constants for phosphate, additional electron and optical micrographs of amorphous and crystalline products at various synthesis conditions, pH measurements and extent of crystallization analyses, and additional AFM images of in situ studies at various conditions (pdf)

Time-elapsed sequence of optical micrographs depicting the growth of brushite crystals in a microchannel under a constant flow of growth solution (6 mL h⁻¹, 20 mM CaP, 125 mM NaCl, pH 5.6) in the absence of modifier. Snapshots at various times were used for Figure 4A. (MP4)

Time-elapsed sequence of optical micrographs depicting the growth of brushite crystals in a microchannel under a constant flow of growth solution (6 mL h⁻¹, 20 mM CaP, 125 mM NaCl, pH 5.6) in the presence of 850 μM CA. Snapshots at various times were used for Figure 4B. (MP4)

Time-elapsed sequence of optical micrographs depicting the growth of brushite crystals in a microchannel under a constant flow of growth solution (6 mL h⁻¹, 20 mM CaP, 125 mM NaCl, pH 5.6) in the presence of 734 μM HCA. Snapshots at various times were used for Figure 4C. (MP4)

Time-elapsed sequences of in situ AFM deflection mode images of a brushite (010) surface measured in a growth solution containing 10 mM CaP and 125 mM NaCl at pH 5.6, flowing at a continuous rate of 6 mL h⁻¹ in the absence of modifier. A snapshot from this sequence was used for Figure 5A. (MP4)

Time-elapsed sequences of in situ AFM deflection mode images of a brushite (010) surface measured in a growth solution containing 10 mM CaP and 125 mM NaCl at pH 5.6, flowing at a continuous rate of 6 mL h⁻¹ in the presence of 340 μM CA. A snapshot from this sequence was used for Figure 5B. (MP4)

Time-elapsed sequences of in situ AFM deflection mode images of a brushite (010) surface measured in a growth solution containing 10 mM CaP and 125 mM NaCl at pH 5.6, flowing at a continuous rate of 6 mL h⁻¹ in the presence of 15 μM HCA. Snapshots at various times were used for Figure 6A. (MP4)

Time-elapsed sequences of in situ AFM deflection mode images of vacancies formed on a brushite (010) surface measured in a growth solution containing 10 mM CaP and 125 mM NaCl at pH 5.6, flowing at a continuous rate of 6 mL h⁻¹ in the presence of 15 μM HCA. Snapshots at various times were used for Figure 6B. (MP4)

Time-elapsed sequences of in situ AFM deflection mode images of a brushite (010) surface measured in a growth solution containing 12.5 mM CaP and 125 mM NaCl at pH 5.6, flowing at a continuous rate of 6 mL h⁻¹ in the absence of modifier. Snapshots at various times were used for Figure 7A. (MP4)