Direct observation of cholesterol monohydrate crystallization

Significance

Cholesterol-related diseases affect a large fraction of the world’s population with significant associated health care costs, yet relatively few studies have explored mechanistic aspects of cholesterol crystallization. Here, we use alcohols as analogues of lipids to create facile environments for in situ characterization of cholesterol crystallization. Our findings reveal a classical mechanism of surface growth where interstep interactions enable a unique self-inhibition mode from merging layers that create macrosteps with much slower growth rates than those of elementary steps. These insights provide a foundation for future design of modifiers that selectively interact with crystal surfaces to cooperatively enhance growth inhibition, thus generating new opportunities to explore therapeutics that improve human health by counteracting the deleterious effects associated with cholesterol precipitation.

Abstract

Cholesterol crystallization is integral to the pathology of diseases such as atherosclerosis and gallstones, yet the relevant mechanisms of crystal growth have remained elusive. Here, we use a variety of in situ techniques to examine cholesterol monohydrate crystallization over multiple length scales. In this study, we first identified a biomimetic solvent to generate triclinic monohydrate crystals, while avoiding the formation of nonphysiological solvates and enabling crystallization at rates where the dynamics of surface growth could be captured in real time. Using a binary mixture of water and isopropanol, with the latter serving as a surrogate for lipids in physiological environments, we show that cholesterol monohydrate crystals grow classically by the nucleation and spreading of crystal layers. Time-resolved imaging confirms that layers are generated by dislocations and monomers incorporate into advancing steps after diffusion along the crystal surface and not directly from the solution. In situ atomic force microscopy (AFM) and microfluidics measurements concertedly reveal abundant macrosteps, which engender a self-inhibition mechanism that reduces the rate of crystal growth. This finding stands in contrast to numerous other systems, in which classical mechanisms lead to unhindered growth by spreading of single layers.

Cholesterol is a ubiquitous molecule in living systems, and it has critical roles in human physiology. The insolubility of cholesterol in water leads to its association with hydrophobic surfaces (1–5), wherein its accretion often leads to crystallization linked to diseases such as atherosclerosis and gallstones (6, 7). Cholesterol crystals, accompanying low-density lipoproteins, are a major structural element of plaque that amasses in arterial walls and is a major contributor to heart disease (8, 9). Cholesterol monohydrate is the physiologically relevant form that accumulates in cell membranes and can lead to inflammatory responses with severe damage to tissues (10). Cholesterol monohydrate crystals in plaque can also lead to heart attacks and sudden cardiac death, accompanied by other side effects (e.g., stroke, chest/leg pain, weakness, and dizziness) (11) with cholesterol-related diseases estimated to cause 2.6 million deaths per year (12). Cholesterol precipitation in the form of gallstones occurs in approximately 20% of adults (13), making this disease one of the most common biliary tract maladies caused by a combination of genetic and environmental factors. Gallstones form as a result of an imbalance of chemical constituents in bile, leading to cholesterol supersaturation. Gallstones can cause various problems that include (but are not limited to) biliary colic/cholecystitis, jaundice, ascending cholangitis, pancreatitis, Bouveret’s syndrome, and gallbladder cancer (14). The most common treatments for gallstone disease are open surgery and drug dissolution therapy (15). Despite numerous studies into the formation of cholesterol, in vitro assays have been unable to elucidate fundamental pathways of cholesterol nucleation and growth with a level of detail required to quantitatively assess molecular modifiers commonly used as therapeutics for pathological crystallization.

There are relatively few investigations of cholesterol crystallization mechanisms and methods to control its formation to facilitate the design of effective therapeutics (16). In order to design new therapies that selectively interact with and reduce rates of cholesterol crystallization, it is essential to better understand at a molecular level the mechanism(s) by which cholesterol crystals grow under physiologically relevant conditions. Prior studies have largely focused on the effects of phospholipids and bile salts on the formation of cholesterol crystals (17–21) without resolving mechanisms at the near molecular level. Addadi and Leiserowitz published a series of seminal papers examining the role in which lipid domains impact the growth and structure of thin film cholesterol crystals (22), using techniques such as grazing incidence X-ray diffraction to characterize film structure (20, 21, 23). They recently used computational methods to examine how physiological environments promote the nucleation of different cholesterol crystal structures (24), recognizing that the nature of these states impacts either its physiological or pathological activity (25, 26). Swift and coworkers have examined conditions promoting cholesterol crystal dissolution (27) and the epitaxial relationships between cholesterol crystals and mineral substrates (28). In this study, we use a binary alcohol–water solvent to examine cholesterol monohydrate crystallization by a combination of optical and scanning probe microscopies to elucidate cholesterol monohydrate crystallization in a biomimetic medium. Our findings reveal that cholesterol crystals grow classically by generation and spreading of new crystal layers. We show that solute monomers reach the edges of unfinished crystal layers (i.e., steps) after adsorption on and diffusion along crystal surfaces. This incorporation pathway invokes strong competition for supply between adjacent steps which leads to a unique self-inhibition mode of action that can induce the cessation of surface growth.

Results and Discussion

Identifying a Biomimetic Solvent for Cholesterol Monohydrate Crystallization.

Cholesterol is an amphiphilic molecule (Fig. 1A) that is highly soluble in organic solvents and nearly insoluble in aqueous media. Cholesterol monohydrate can be prepared as one of two crystal structures (SI Appendix, Table S1): triclinic (Fig. 1A) and monoclinic (26). Triclinic monohydrate crystals (P1 space group) are the most pathologically relevant form of cholesterol in humans (25). In this crystal lattice, opposing cholesterol molecules align with their hydroxyl groups in close proximity to form hydrogen bonds with water molecules (Fig. 1A). This packing arrangement leads to anisotropic growth, which generates platelets exposing large basal (001) surfaces. To mimic physiological environments for monitoring cholesterol crystallization, in which water from the biological fluids (29) inevitably penetrates and saturates the organic environments (30), we used a mixed organic–water solvent for all analyses. Prior studies have largely used lipids as a medium for assessing bulk crystallization of cholesterol (18–20). Lipids are an ideal biomimetic environment for assessing pathological cholesterol formation; however, these systems are highly susceptible to phase separation, which limits characterization of cholesterol crystallization using in situ methods. To this end, we explored alcohols as surrogate polar organic solvents that possess both hydrophilic and hydrophobic moieties, similar to lipids but without the large groups of distinct polarity that drive their self-assembly into micelles or bilayers. Bulk crystallization assays in the presence of alcohols with greater hydrophobicity (e.g., n-butanol to n-octanol) resulted in rapid crystallization (i.e., order of seconds), which made time-resolved characterization of crystal growth infeasible; therefore, we focused on the analysis of less hydrophobic alcohols.

Fig. 1.

(A) Structures of cholesterol and triclinic cholesterol monohydrate crystals (P1 space group with a unit cell containing 8 cholesterol molecules). Color code: carbon (gray), oxygen of cholesterol (red), and oxygen of water (blue). (Bottom image) Optical micrograph of a cholesterol monohydrate crystal prepared in isopropanol and water (50% volume). (Scale bar, 10 µm.) (B) Powder XRD patterns of solids extracted from alcohol (65%) and water (35%) solutions, which are compared with reference patterns for cholesterol monohydrate (Top) (31) and cholesterol ethanol solvate (Bottom) (32). (C) Equilibrium concentration of soluble cholesterol in different alcohol–water mixtures as a function of water content at 37 °C. Similar trends were observed at different temperatures for select cases (e.g., IPA/water mixtures, SI Appendix, Fig. S2). Data for methanol and ethanol correspond to cholesterol alcohol solvate crystals, whereas those in IPA and PA are cholesterol monohydrate crystals (SI Appendix, Table S1). (Inset) Solubility (C_e) of cholesterol monohydrate crystals in 50% water-isopropanol mixtures at different temperatures. (D) Enthalpy, entropy, and free energy of crystallization of cholesterol monohydrate crystallization as a function of the water content in an IPA/water mixture at 37 °C (see also SI Appendix, Fig. S2 for changes in free energy of crystallization with temperature).

To assess the suitability of different alcohols for in situ growth studies, cholesterol crystals were prepared in two ways and tested by a combination of PXRD, single crystal X-ray diffraction (SCXRD), and 3-dimensional electron diffraction (3D ED). In the first method, as-received commercial anhydrous cholesterol powder was incubated in organic-water solutions for more than two weeks. In the other method, crystals prepared in an organic-water mixture were obtained by the slow cooling of supersaturated cholesterol solutions. All crystals prepared by this method yielded a platelet morphology. Prior studies have used ethanol (EtOH) in water as a solvent for assessing cholesterol dissolution and bulk crystallization (27). Based on PXRD our findings reveal that cholesterol crystallization in ethanol/water mixtures leads to the formation of a cholesterol ethanol solvate (Fig. 1B and SI Appendix, Table S1). A similar outcome was observed when exposing anhydrous cholesterol powder to EtOH (SI Appendix, Fig. S1). Replacing ethanol with methanol (MeOH) leads to a cholesterol methanol solvate crystal structure, as confirmed by SCXRD (SI Appendix, Fig. S1 and Table S1). Experiments using mixtures of either propanol (PA) or isopropanol (IPA) with water both result in cholesterol monohydrate (triclinic) crystals (Fig. 1B and SI Appendix, Fig. S1), which was verified by SCXRD and 3D ED (SI Appendix, Table S1). In prior literature, a mixture of acetone and water has been used to form cholesterol monohydrate (triclinic, P1) crystals (31); however, the high volatility of acetone poses difficulties for in situ characterization.

To determine the effect of organic/water ratio on potential organic solvate formation, we prepared crystals in both acetone/H₂O and IPA/H₂O media with varying water volume percent and observed identical PXRD patterns. SCXRD analysis of samples extracted from acetone/H₂O and IPA/H₂O solvents revealed the formation of cholesterol monohydrate (triclinic, P1) crystals. In contrast, cholesterol crystallization in EtOH/H₂O mixtures results in both cholesterol ethanol solvate and monohydrate crystals where the latter is more prevalent in PXRD patterns with increasing water volume percent (SI Appendix, Fig. S1). We also measured the equilibrium concentration of soluble cholesterol as a function of water volume percent in different alcohols at 37 °C (Fig. 1C). For IPA and PA, each concentration corresponds to the solubility of cholesterol monohydrate; however, we refrain from using this terminology for EtOH and MeOH mixtures with water, which lead to a combination of cholesterol monohydrate and cholesterol alcohol solvate crystals. For both IPA/water and PA/water mixtures, the solubility of cholesterol decreases with increasing water volume percentage. Collectively, our findings identified the most suitable conditions for obtaining physiologically relevant cholesterol monohydrate crystals as mixtures of either IPA or PA with water, which exhibit high organic content while maintaining a moderate equilibrium concentration. IPA emerged as the most optimal alcohol (Fig. 1C, Inset) for studying cholesterol monohydrate crystallization at timescales that are amenable to in situ measurements of crystal growth.

We analyzed the thermodynamics of cholesterol crystallization in IPA/water mixtures using an established protocol (33) for calculating the crystallization enthalpy ΔH°_cryst, Gibbs free energy ΔG°_cryst, and crystallization entropy ΔS°_cryst (Fig. 1D). We evaluate these thermodynamic variables from the temperature dependencies of the cholesterol solubility in solvents with varying water/IPA ratios. The solubility relates to ΔG°_cryst as $\mathrm{\Delta }{G}_{\mathit{cryst}}^o=RT\ln C_e$ and to ΔH°_cryst by the van’t Hoff law $\frac{\partial \ln C_e}{\partial \left(1/T\right)}=\frac{\mathrm{\Delta }{H}_{\mathit{cryst}}^o}{R}$ (34). To evaluate ΔS°_cryst we use the Gibbs–Helmholtz relation $\mathrm{\Delta }{S}_{\mathit{cryst}}^o=(\mathrm{\Delta }{H}_{\mathit{cryst}}^o-\mathrm{\Delta }{G}_{\mathit{cryst}}^o)/T$. Since cholesterol crystalizes as identical triclinic monohydrate crystals from solvents with all tested water/IPA ratios, we use the variations of ΔG°_cryst, ΔH°_cryst, and ΔS°_cryst to elucidate the molecular interactions between cholesterol and solvent. The crystallization free energy, $\mathrm{\Delta }{G}_{\mathit{cryst}}^o=G_{\mathit{crystal}}-G_{\mathit{solution}}$ (where $G_{\mathit{crystal}}$ and $G_{\mathit{solution}}$ are the free energies of a molecule in a kink and in the solution, respectively), mildly decreases with increasing water content in the solvent from 40 to 60 vol.%, corresponding to molar fractions 0.74 and 0.86 (Fig. 1D). The decreasing $\mathrm{\Delta }{G}_{\mathit{cryst}}^o$ indicates that cholesterol monohydrate crystallization is favored at higher solvent hydrophilicity. Surprisingly, the respective crystallization enthalpy and entropy, $\mathrm{\Delta }{H}_{\mathit{cryst}}^o$ and $\mathrm{\Delta }{S}_{\mathit{cryst}}^o$, decrease substantially, reflected in the greater sensitivity of the solubility to temperature in the more hydrophilic solvents (SI Appendix, Fig. S2). Since the final state of the molecule, in the monohydrate crystal, is independent of the water content, the lower $\mathrm{\Delta }{H}_{\mathit{cryst}}^o$ and $\mathrm{\Delta }{S}_{\mathit{cryst}}^o$ indicate that the enthalpy and entropy of the initial solution state of cholesterol increase as the water concentration increases. Greater $H_{\mathit{solution}}$ implies increasing repulsion between solvent and solute, consistent with expectations for a hydrophobic solute in an increasingly hydrophilic solvent. Higher $S_{\mathit{solution}}$ indicates that elevated water concentrations destabilize solution structures that may be due to the ordering of IPA around cholesterol molecules, similar to what has been seen for octanol molecules around hematin, another organic solute, in mixed octanol-water solvents (35, 36). The observed strong correlation between cholesterol solubility and water content suggest that care must be taken to select an IPA/water ratio for in situ measurements such that crystal growth dynamics can be tracked in real time; therefore, all data reported herein were obtained at 23 °C using a 50:50 (by volume) mixture of IPA and water.

Macroscopic Analysis of Cholesterol Monohydrate Crystallization.

We used a microfluidic system (37, 38) to visualize growth of cholesterol monohydrate platelets at a macroscopic length scale. The anisotropic morphology of the platelets restricts analysis of growth rates along two directions, [100] and [010], which are referred to as the length (L) and width (W), respectively. Channels of the microfluidic device were first seeded with cholesterol monohydrate crystals by introducing a 1.6 mM cholesterol solution, which has a supersaturation ratio S = 1.21 (S = C/C_e where C_e = 1.32 mM, Fig. 1C). Platelets with (001) basal surfaces were grown under steady supersaturation. For this, a syringe pump was used to supply a constant flow of growth solution at lower supersaturation (S = 1.02 to 1.14) to monitor the evolution of crystal length and width in time. We observed that the seeding protocol resulted in two types of cholesterol platelets, referred to generically as “thin” (Fig. 2A and Movie S1) and “thick” (Fig. 2B and Movie S2) crystals. A quantitative analysis of crystal dimensions and growth rates in the vertical direction (out of plane) was difficult owing to the highly anisotropic shape of crystals and the challenges associated with measuring rates of layer nucleation (vide infra). Sequential images from in situ microfluidic measurements were used to assess changes in crystal dimensions in plane (i.e., L and W) by statistically sampling multiple crystals (Fig. 2C). These studies revealed that thin crystals (solid symbols) grow at much faster rates; however, as the crystals become thicker with longer crystallization time (>12 min), we observe a change in the growth rate (i.e., slope of curves, SI Appendix, Fig. S3). Time-elapsed images of crystal growth (SI Appendix, Fig. S3 and Movie S3) show the eventual formation and advancement of macrosteps on the (001) surface, which can explain their slower rate of growth (vide infra).

Fig. 2.

Time-resolved images of (A) thin and (B) thick crystals growing in a microfluidic device at a fixed cholesterol concentration of 1.35 mM in 50:50 IPA/water at 23 °C. The supersaturation ratio, S, of these solutions equals 1.02 (i.e., C_e = 1.32 mM; Fig. 1C, Inset). Experiments were performed at a constant flow rate of 6 mL/h for over 100 min of continuous imaging. All scale bars equal 20 µm and the units of listed times are minutes. (C) Displacement of crystal length (L) and width (W) as a function of crystallization time for thin and thick crystals (closed and open symbols, respectively). (D) Growth rate of crystals in the [100] and [010] directions (L and W, respectively) evaluated from the slopes of displacement curves in panel C (ΔL/2 and ΔW/2 vs. time). Analyses are made within the first 10 to 12 min of imaging at different supersaturation, C–C_e. Solid lines are linear regression with slopes listed in the legend (ranging from 20 to 130 nm s⁻¹ mM⁻¹). The line/symbol colors in panel D are identical to those in panel C. Symbols are the average of 10 measurements on different crystals and error bars span two SD.

We measured the growth rates of cholesterol monohydrate crystals at different supersaturations (Fig. 2D). Comparison of thin and thick crystals revealed approximate twofold and threefold differences in the rate of growth at the lowest and highest supersaturation, respectively. There is an expected linear relationship between the growth rate and supersaturation for both thin and thick crystals. To understand why thick crystals grow at reduced rates we used in situ atomic force microscopy (AFM) to assess the dynamics of cholesterol monohydrate surface growth at a near molecular resolution under growth conditions similar to those in the microfluidic experiments.

Microscopic Analysis of Cholesterol Monohydrate Surface Growth.

AFM measurements were performed in growth solutions of varying cholesterol concentration. We first determined the minimum flow rate to ensure surface growth occurs in a reaction-limited regime (38) and found that growth rates were constant above 50 mL h⁻¹ (SI Appendix, Figs. S4 and S6). For all experiments reported herein, we used a flow rate at 60 mL h⁻¹ to supply growth solution to the AFM liquid cell. Based on solubility measurements of cholesterol (Fig. 1C: C_e = 1.32 mM for 50:50 IPA:H₂O at 23 °C), we prepared a series of growth solutions spanning from slightly undersaturated to moderately supersaturated. In situ AFM measurements of cholesterol monohydrate (001) surfaces in undersaturated media resulted in layer-by-layer dissolution and the formation of etch pits (Fig. 3A and Movie S4). Single layers with step heights (ca. 3.4 nm) equivalent to the unit cell dimension of triclinic crystals (Fig. 1A) dissolve at much faster velocities than those of multisteps or step bunches (Fig. 3A, arrow). When the growth solution is mildly supersaturated (1.35 mM cholesterol) there is shift from dissolution to layer-by-layer growth (SI Appendix, Figs. S7–S11). As the cholesterol concentration is further increased to 1.4 mM, we observe predominantly single layers generated from screw dislocations (Fig. 3B and Movie S5) that advance faster along the [100] direction; however, we do not observe two-dimensional islands and also very few dislocation centers in AFM images, which makes it difficult to evaluate the effect of supersaturation on the vertical growth rate (i.e., out of plane growth). When the cholesterol concentration is increased to 1.45 mM, we begin to see a higher percentage of multiheight steps. Time-elapsed images extracted from Movie S6 show that single layers (Fig. 3C, white arrow) advance with step velocities threefold larger than those of double layers (Fig. 3C, yellow arrow), which is a signature feature of growth by surface diffusion (39). Further evidence is gleaned from observations at fixed supersaturation where we observe a monotonic reduction in single-step velocity with decreasing interstep distance (SI Appendix, Fig. S5), consistent with trends previously reported for other crystals, for which solute is supplied to the steps after adsorption on the crystal surface and two-dimensional diffusion toward the steps (40–44). By contrast, if solute reaches the steps directly from the solution step growth is unaffected by the step separation or the heights of the individual steps owing to the abundant supply of molecules from the bulk solution (45, 46).

Fig. 3.

(A–C) Time-resolved in situ AFM images showing layers on a (001) face at cholesterol concentrations of (A) 1.3 mM, (B) 1.4 mM, and (C) 1.45 mM (C_e = 1.32 mM) in 50:50 (by volume) isopropanol:water at 23 °C and a constant flow rate of 60 mL h⁻¹. Images in panel (A) of an undersaturated solution reveal dissolution by layer retraction (white arrow) and etch pit formation (yellow dotted line). The height profiles (Right plot) along the yellow dashed lines show a single step at time t = 0 s and an etch pit at t = 150 s. (B) Single layers advancing from a screw dislocation with a new layer generated at t = 150 s. (C) Growth of single (white arrow) and double (yellow arrow) steps. All AFM scale bars equal 3 µm and listed times are in seconds. (D) Step velocity as a function of supersaturation for single steps in the [100] direction (red circles) and [010] direction (black squares) as well as double steps in the [100] direction (blue triangles). Data represent the averages of 6 to 8 measurements, error bars span two SD from the average, and lines are linear regressions where the slopes were used to calculate the kinetic coefficient β.

Step velocities for both single and double-height steps were measured over a range of cholesterol supersaturation (Fig. 3D). We calculated the step kinetic coefficient (β) from the linear relation of step velocity and cholesterol supersaturation as v = βΩ(C−C_e); where Ω is the volume occupied by a cholesterol molecule in the crystal, i.e., 0.64 nm³. The calculated β in the [100] direction for single layers is nearly twice that of the [010] direction, despite the crystal structure having equivalent a and b unit cell dimensions. The anisotropy of step velocity may relate to anisotropic surface diffusion length on the crystal surface and to distinct structures of the steps growing in the [100] and [010] directions and their respective kinks (44, 46–48); however, a comprehensive molecular-level understanding of the step-growth anisotropy is currently lacking.

The measured step velocities are several orders of magnitude greater than those of other organic crystals growing from either mixed aqueous-organic or purely organic solvents at similarly low supersaturations (41, 45, 46, 49, 50). The fast step growth ensures fast sequestration of solute into crystals as cholesterol is delivered to the arterial walls during its physiological homeostasis (51, 52). Importantly, as cholesterol step growth and dissolution are symmetric at the molecular level (Fig. 1D), crystal dissolution is also fast. In this way, cholesterol crystals may serve as a ready storage and supply of cholesterol and a part of the mechanism that maintains its steady body concentration (52).

AFM images of basal (001) surfaces reveal populations of steps with distinct heights (Fig. 4A) that can be generally grouped as follows: i) single steps, ii) step bunches, and iii) macrosteps. The macrosteps represent virtually immobile layer stacks with extremely slow growth rates. Single steps grow with much faster velocity than those of step bunches. In situ imaging reveals that single step and step bunches advance until they join a macrostep where growth ceases (Fig. 4A, arrow). Infrequently we observe recesses (Fig. 4B, dashed box) that are likely generated by the merging of layers to create regions enclosed by macrosteps (see also Movie S7). Steps reaching these sites become pinned by the recesses, which induces localized step bunching. In regions far from dislocation centers the anisotropy of step advancement is difficult to discern given the basal surface is composed of numerous layers generated from many different step sources. For example, the AFM image in Fig. 4C shows a region of the basal surface where multiple dislocations are located in close proximity. Areas surrounding each dislocation (Fig. 4C, dashed box) reveal the anisotropic advancement of layers; however, the interstitial regions between these step sources exhibit irregular patterns as layers merge and generate macrosteps. This phenomenon is consistent with our observation from microfluidics (Fig. 2) where thicker crystals composed of a higher percentage of multi- and macrosteps exhibit ultraslow rates of crystal growth. These findings collectively indicate that during cholesterol crystallization natural step bunching, with the step bunches evolving into macrosteps, acts as a mode of self-inhibition, comparable to reported cases in the literature where foreign species (i.e., growth modifiers) are required to achieve similar outcomes (53–56). The evolution of single steps into step bunches and macrosteps is more frequent at high supersaturations, at which the overall step density is higher. Importantly, step bunching is a consequence of strong step–step interaction (57), which, in this case, is associated with the overlap of surface diffusion fields of adjacent steps (58–62). In turn, solute supply to the steps via surface diffusion, a precondition for step bunching, is more likely for crystals growing from aqueous or partially aqueous solvents (40–42), whereas purely organic solvents appear to favor the direct incorporation pathway (46, 63). Thus, the identified self-inhibition mechanism may be more common in water-containing solvents. From another point of view, the slow growth of surfaces populated with macrosteps serves as a potential regulation mechanism, which prevents ultrafast cholesterol crystallization at elevated cholesterol concentrations, similar to the enzyme turnover rate in the Michaelis–Menten mechanism (64).

Fig. 4.

(A–C) Time-resolved in situ AFM images showing the growth of new layers on cholesterol monohydrate (001) surfaces using 50:50 IPA:H₂O growth solutions (at 23 °C) with the following cholesterol concentrations: (A) 1.45 mM, (B) 1.5 mM, and (C) 1.35 mM. Images in panel (A) reveal growth of (i) single steps that join a macrostep (white arrow) and (ii) step bunches that join a (iii) macrostep (yellow arrow). (B) Images extracted from Movie S7 reveal bunching of steps around a recess (dashed yellow box). Macrostep generation from the merging of single steps is highlighted by the yellow arrow. (C) Multiple spiral dislocation centers. Callout: enlarged region within the dashed yellow box reveals the asymmetric advancement of layers from a single dislocation center. All scale bars equal 3 µm unless otherwise specified. The listed times (0 and 75) are in seconds.

Conclusions

Here, we examine alcohols as simple surrogates of lipids to create facile environments for in situ characterization of cholesterol crystallization. We show that certain alcohols selected as biomimetic solvents for cholesterol monohydrate crystallization can have an undesirable effect of directing the formation of cholesterol solvates that exclude water, thereby producing nonphysiologically relevant solids. Notably, crystallization in water mixtures containing either methanol or ethanol leads to alcohol solvates, whereas propanol and isopropanol produced the desired triclinic cholesterol monohydrate crystals, identical to the most common form of crystals in vivo. We selected mixtures of water and isopropanol for detailed analysis of cholesterol monohydrate crystallization at both macroscopic and microscopic length scales. Microfluidics experiments reveal that cholesterol monohydrate exhibits a self-inhibition mode of growth as crystals become thicker along the [001] direction. In situ AFM measurements confirmed that monohydrate crystals grow classically with spiral dislocation centers as step sources. The dependence of step velocity on interstep distances indicates that cholesterol molecules incorporate into kinks along the steps after diffusion along the crystal surface and not directly from the solution. Time-resolved AFM observations also reveal the formation of macrosteps from the merging of layers emanating from proximal dislocations. The macrosteps exhibit significantly reduced growth rates compared to single steps, consistent with a self-inhibition mechanism observed in bulk assays. These collective insights provide a foundation for future investigations of cholesterol monohydrate crystallization under different conditions, including alternative solvents with properties (e.g., viscosity and hydrophobicity) that mimic in vivo growth environments for cholesterol crystallization in vascular channels or gallbladders. We posit that the phenomenon of self-inhibition imposed by macrostep formation may be a more general mechanism that applies to other crystals growing by surface diffusion, more likely for growth from water-based solvents. This also has implications for how foreign species affect crystal growth wherein the design objective of (macro)molecular modifiers, which is conventionally to fully suppress the advancement of single layers, may be supplanted by the need to only induce step bunches. The in situ techniques used in this study also offer a platform for future identification and investigation of modifiers that selectively interact with crystal surfaces and impede growth. Advances in the design of such modifiers have the potential to improve human health by counteracting the deleterious effects associated with cholesterol precipitation.

Materials and Methods

Materials.

The following compounds were purchased from Sigma-Aldrich (St. Louis, MO): cholesterol (≥98%), propanol (≥99.5%), isopropanol (≥99.5%), ethanol (≥99.5%), and methanol (≥99.7%). Deionized (DI) water was produced by a Millipore reverse osmosis–ion exchange system (Rios-8 Proguard 2–MilliQ Q-guard).

Extinction Coefficient.

To determine the extinction coefficient ε in each solvent, a known amount of cholesterol was fully dissolved in a known volume of solvent. These stock solutions were diluted to eight concentrations and filtered through 0.22 μm polyethersulfone (PES) filters. Absorbance spectra in the wavelength range 200 to 800 nm were recorded for each concentration using a DU spectrophotometer 800 (Beckman). A wavelength of peak absorbance at 202 nm was identified and ε at that wavelength was determined using the Beer–Lambert law. Each determination was based on three independent concentration series.

Determining the Solubility of Cholesterol.

To determine the solubilities of cholesterol in different water/alcohol mixtures, cholesterol powder was added to a 10 mL water/alcohol solution in a 20-mL glass vial that was subsequently sealed with Parafilm. Sets of three vials were stored at four temperatures: 23, 30, 37, and 45 °C. A 300-μL aliquot was removed weekly from each of the 12 solutions, diluted with the same water/alcohol mixture, filtered through a 0.22-μm polyethersulfone (PES) filter, and the concentration of dissolved cholesterol was determined spectrophotometrically. This procedure was repeated until the concentrations in each vial reached a plateau, defined by three consecutive readings of similar value. The final steady-state concentrations were averaged over the three samples at each temperature. The resulting mean was used as the concentration C_e of cholesterol in water/alcohol mixtures in equilibrium with crystals at the bottom of the glass vial where the crystal structure was determined by powder X-ray diffraction (PXRD) and single crystal X-ray diffraction (SCXRD). Solubility in AFM studies was determined from two intermediate concentrations which are at the boundary leading to growth or dissolution of steps, respectively.

Preparation of Cholesterol Crystals from Slow Cooling and Slow Evaporation.

To control the size and shape distributions of cholesterol crystals, a small step-change in supersaturation can be achieved through either slow cooling or evaporation. In slow cooling crystallization, excess amount of cholesterol was added into a water/isopropanol mixture (10 mL). The solution was heated to 45 °C for 10 to 15 mins in a water bath, then 5 mL supernatant was filtered through a 0.22-μm PES filter and transferred to an empty vial at 45 °C. The temperature was reduced gradually to room temperature in 5 h. For the slow evaporation method, an excess amount of cholesterol was added to a water/alcohol mixture (10 mL) and stored in the dark for two weeks. The supernatant (5 mL) was filtered through a 0.22-μm PES filter and transferred to an empty vial connected to the atmosphere through a pore. Cholesterol crystals begin to appear after several days and grow to an average size of 500 µm. Crystals synthesized by both methods were stored in the dark for later use.

Determining the Crystal Phase of Cholesterol.

To determine the exact crystal phase of cholesterol (solvates and polymorphs), we performed two separate steps: 1) we precipitated crystals from a supersaturated alcohol–water solution both by slow cooling and slow evaporation; and 2) we incubated anhydrous cholesterol powder in the solvent for more than two weeks. Precipitated crystals and solvent-exchanged crystals were dried and ground properly prior to PXRD analysis using a Rigaku diffractometer with Cu Kα radiation (40 kV, 40 mA) to verify the crystallinity and structure of crystals. Cholesterol monohydrate and cholesterol solvate crystal structures were confirmed by SCXRD at 123 K using a Bruker DUO platform diffractometer equipped with a 4 K CCD APEX II detector and an Incoatec 30 W Cu microsource with compact multilayer optics. Data were collected using a narrow-frame algorithm with scan widths of 0.50 degrees in omega and phi and exposure time of 5 to 30 s/frame at a 4 cm detector distance. Data were integrated using the Bruker SAINT program, with the intensities corrected for the Lorentz factor, polarization, air absorption, and absorption due to variation in the path length through the detector faceplate. Data were scaled and an absorption correction was applied using SADABS. The structure was solved with SHELXT and refined with SHELXL using full-matrix least-squares refinement. The non-H atoms were refined with anisotropic thermal parameters, and all of the H atoms were calculated in idealized positions and refined riding on their parent atoms. The size and morphology of crystals were determined by scanning electron microscopy using an Axia ChemiSEM instrument with a maximum accelerating voltage of 30 kV and a filament (tungsten) Schottky source. Samples for SEM were prepared by adhering crystals to carbon tape. TEM grids were first glow discharged for 40 s using a PELCO easiGLOW (Ted Pella, Inc.) with a current of 20 mA. One drop of suspension was delivered to a continuous carbon copper grid by a pipette. After solvent evaporation, the grid was loaded into a Titan Krios microscope operated at 300 kV. All the 3DED datasets were collected on a Ceta-D camera. The electron dose was set to 0.01 e/Ųs, and the rotation range was set to −30° to +30°.

In Situ Monitoring of the Bulk Cholesterol Crystal Evolution.

We employed a microfluidic platform described in previous studies (37, 38) for in situ analysis of crystal growth. We used a supersaturated solution to directly seed crystals in the microfluidic channel and then passed supersaturated solutions at different cholesterol concentrations to grow the crystals. This system was able to monitor the growth of cholesterol monohydrate crystals under continuous supply of fresh growth solution using an inverted optical microscope (Leica DMi8 instrument). Growth solutions prepared with different cholesterol concentrations (1.35 to 1.50 mM) in isopropanol-water (50% by volume) were used in microfluidic studies at 23 °C. As cholesterol crystallization is highly sensitive to temperature, we made a concentric heating syringe with continuously flowing hot water on the outer volume attached to water circulator at 30 °C. This prevented instantaneous crystallization and kept the solution supersaturated when it reached inlet of the microfluidic channel. The growth solution was then delivered to the device at a rate of 6 mL h⁻¹ for over 100 min. We observed two different populations of crystals (thin and thick) based on the contrast and measured growth rates for each population separately. The growth rates were measured for 10 to 15 crystals for each population at different cholesterol concentrations.

In Situ Monitoring of the Cholesterol Crystal Surface Evolution.

We used a Multimode 8 AFM from Bruker for all experiments in this study. AFM mages were collected in tapping mode using Scanasyst-Fluid probes (Silicon nitride, 0.7 N/m spring constant) with a tapping frequency of 150 kHz. Image sizes ranged from 10 to 40 μm and scan rates were selected between 5 and 12 Hz. The captured images contained 256 scan lines at angles depending on the orientation of the monitored crystal. The temperature outside the fluid cell was controlled with the same heating syringe as microfluidics experiments to prevent crystallization. Cholesterol crystals were removed from the synthesis solution with tweezers and gently placed on an AFM disk (Ted Palla Inc.) coated with clear quickset Gorilla epoxy. Samples on the disks were placed on the AFM scanner and images were collected over 30 min. The evolution of unfinished layers on cholesterol crystal surfaces was characterized by the velocity ν of advancing steps, which were monitored by tracking the displacement of 5 to 10 individual steps. Between 10 to 15 measurements were taken for each step and the average growth rates were reported.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Movie S1.

Dynamics of seeded cholesterol monohydrate crystal growth in a microfluidic device (50:50 isopropanol:water). The video reveals growth of a thin crystal at a cholesterol concentration of 1.35 mM for 20 min total imaging time. The imaging area is 50 μm × 65 μm.

Movie S2.

Dynamics of seeded cholesterol monohydrate crystal growth in a microfluidic device (50:50 isopropanol:water). The video reveals growth of a thick crystal at a cholesterol concentration of 1.35 mM for 100 min total imaging time. The imaging area is 50 μm × 65 μm.

Movie S3.

Dynamics of seeded cholesterol monohydrate crystal growth in a microfluidic device (50:50 isopropanol:water). The video reveals growth of a multi-step on the crystal surface at a cholesterol concentration of 1.35 mM for 30 min total imaging time. The imaging area is 100 μm × 85 μm.

Movie S4.

Dynamics of cholesterol monohydrate (001) crystal surface dissolution during in situ AFM experiment. The video reveals single steps receding and etch pits forming in an undersaturated solution (50:50 isopropanol:water) with cholesterol concentration of 1.3 mM (compared to C_e = 1.32 mM). The imaging area is 8 μm × 8 μm.

Movie S5.

Dynamics of cholesterol monohydrate (001) crystal surface growth during in situ AFM experiments. The video reveals layer nucleation and growth in a supersaturated solution (50:50 isopropanol:water) with a cholesterol concentration of 1.4 mM. The imaging area is 10 μm × 17 μm.

Movie S6.

Dynamics of cholesterol monohydrate (001) crystal surface growth during in situ AFM experiments. The video reveals layers growing in the presence of macrosteps for a supersaturated solution (50:50 isopropanol:water) prepared with a cholesterol concentration of 1.45 mM. The imaging area is 10 μm × 5 μm.

Movie S7.

Dynamics of cholesterol monohydrate (001) crystal surface growth during in situ AFM experiments. The video reveals the emergence of macrosteps during growth in a supersaturated solution (50:50 isopropanol:water) with a cholesterol concentration of 1.5 mM. The imaging area is 15 μm × 13 μm.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information. Previously published data were used for this work [Unit cell parameters of cholesterol crystal structures obtained from the CCDC and/or literature (1–5) compared to our samples prepared in house and analyzed by SCXRD (SI Appendix, Table S1)].