Research Article

Crystal Growth Inhibitors for the Prevention of l-Cystine Kidney Stones Through Molecular Design

See allHide authors and affiliations

Science  15 Oct 2010:
Vol. 330, Issue 6002, pp. 337-341
DOI: 10.1126/science.1191968


Crystallization of l-cystine is a critical step in the pathogenesis of cystine kidney stones. Treatments for this disease are somewhat effective but often lead to adverse side effects. Real-time in situ atomic force microscopy (AFM) reveals that l-cystine dimethylester (L-CDME) and l-cystine methylester (L-CME) dramatically reduce the growth velocity of the six symmetry-equivalent {100} steps because of specific binding at the crystal surface, which frustrates the attachment of l-cystine molecules. L-CDME and L-CME produce l-cystine crystals with different habits that reveal distinct binding modes at the crystal surfaces. The AFM observations are mirrored by reduced crystal yield and crystal size in the presence of L-CDME and L-CME, collectively suggesting a new pathway to the prevention of l-cystine stones by rational design of crystal growth inhibitors.

Kidney stones comprising l-cystine (Fig. 1) affect at least 20,000 individuals in the United States. Although this number is substantially smaller than the number of individuals afflicted by calcium oxalate monohydrate (COM) stones (approximately 10% of the U.S. population), l-cystine stones are larger, recur more frequently, and are more likely to cause chronic kidney disease (1). The formation of l-cystine stones is a consequence of excessive levels of l-cystine in the urine because of defective reabsorption of filtered cystine (2). This condition is the result of an autosomal recessive disorder caused by mutations in one of the two genes, either SLC3A1 on chromosome 2 (type A cystinuria) or SLC7A9 on chromosome 19 (type B cystinuria), which code for components of the major proximal renal tubule cystine and dibasic amino acid transporter (3). This condition is exacerbated by the low solubility of l-cystine (4), which favors facile formation of crystals that aggregate into stones, often with centimeter dimensions (Fig. 2A).

Fig. 1

Molecular structures of l-cystine and the inhibitors L-CDME and L-CME.

Fig. 2

(A) Human stones with millimeter-scale dimensions [courtesy of M. Lewis, International Cystinuria Foundation]. (B) A hexagonal l-cystine crystal prepared in vitro. The faint lines on the top surface of the crystal, parallel to the edges, are the {100} steps. (C) Two adjacent helices of l-cystine molecules, viewed on the (100) plane, each winding about a 61 screw axis that coincides with the c axis. Six l-cystine molecules, denoted C1 to C6, span the 5.6-nm c axis. Key intermolecular interactions include amine-carboxylate hydrogen bonds along the helix (I, dN…O = 2.87 Å) and S…S interactions (II, dS…S = 3.47 Å) between helices at intervals of c/2, depicted here for C1 and C4 along the [010] direction (identical S…S interactions occur at symmetry-related sites along the other five equivalent directions). (D) Intermolecular amine-carboxylate hydrogen bonds in the (001) plane (III, dN…O = 2.79 Å and IV, dN…O = 2.81 Å). Atom color code is gray, carbon; red, oxygen; blue, nitrogen; yellow, sulfur; and white, hydrogen. (E) Schematic illustration of a hexagonal l-cystine crystal, with Miller indices. The six planes flanking (001) belong to the {100} family.

Current treatments for l-cystine stone prevention, including dilution through high fluid intake (5) and increasing urine pH through ingestion of alkalinizing potassium or sodium salts (5, 6), can suppress but may not completely prevent stone formation. Severely afflicted individuals often rely on additional treatment with l-cystine–binding thiol drugs such as d-penicillamine [HSC(CH3)2CH(NH2)COOH] and α-mercaptopropionylglycine [HSCH(CH3)C(O)NHCH2COOH], which react with l-cystine to generate more soluble asymmetric disulfides (2). These drugs, however, have an unpleasant odor and can cause adverse side effects, such as nausea, fever, fatigue, skin allergies, and hypersensitivity (5). For patients with very high l-cystine concentrations, l-cystine–binding thiol drugs may not reduce l-cystine levels sufficiently at dosages (up to 2000 mg per day) regarded as below the threshold for toxicity, and patient adherance to these prescribed medications and high fluid intake can be problematic (5). We report an alternative approach to the prevention of l-cystine kidney stones that is based on crystal growth inhibition achieved through the binding of tailored growth inhibitors—l-cystine dimethylester (L-CDME) and l-cystine methylester (L-CME)—to specific crystal surfaces, as revealed by in situ real-time atomic force microscopy (AFM) (7, 8) and parallel bulk crystallization studies (9).

l-cystine stones are aggregates of individual crystals with hexagonal habits (Fig. 2). l-cystine can be crystallized in vitro at physiological pH (6 ≤ pH ≤ 8) by means of slow evaporation (10), acidification of basic l-cystine solutions to neutral pH (11), or gradual cooling of solutions supersaturated with l-cystine (12). Under these conditions, l-cystine crystallizes as hexagonal plates (Fig. 2B) with large (001) basal surfaces that can achieve widths of 400 μm and are bounded by six equivalent {100} faces. The typical thickness of these crystals ranges from 10 to 30 μm. The crystal structure (hexagonal P6122 space group, a = b = 0.5422 nm, c = 5.6275 nm) reveals l-cystine molecules organized as a helix about the 61 screw axis so that six cystine molecules span the ~5.6-nm unit cell length of the c axis (13). The l-cystine molecules exhibit intermolecular NH3+O(C=O) hydrogen bonding along the 61 screw axis (Fig. 2C, I), intermolecular S…S interactions between the helices at intervals of c/2 along each of the six equivalent {100} directions (Fig. 2C, II), and NH3+O(C=O) hydrogen bonding (Fig. 2D, III and IV) between adjacent helices in the (001) plane. The hexagonal plate habit reflects the multiple strong intermolecular interactions in the (001) plane. The basal surfaces of l-cystine grown at neutral pH are decorated with {100} steps that are observable through either optical (Fig. 2B) or scanning electron microscopy (SEM) (fig. S2).

Crystal growth near equilibrium is commonly described by the terrace-ledge-kink model (14), in which steps created by dislocations advance across crystal terraces by the addition of solute molecules to kink sites along the ledge (a ledge is the intersection of a step and terrace). Steps originating from screw dislocations typically exhibit a spiral growth pattern, with the first turn occurring once every step has reached its critical length (15). Real-time in situ AFM of the l-cystine (001) face during growth in aqueous solutions containing l-cystine revealed steps emanating from screw dislocations, generating hexagonal hillocks in a spiral growth pattern. Occasionally, multiple dislocations were observed (Fig. 3, A and B, and movie S1), merging to generate a range of step heights from 1 nm to 60 nm, with the larger steps observed distant from the dislocation cores, where step bunching would be expected (fig. S3). In contrast, hillocks generated by single isolated dislocations were bounded by six well-defined major {100} steps, each with a ~6-nm height corresponding to the unit cell length along c, separating (001) terraces. Each hillock terrace was decorated with six minor {100} steps at 60° intervals, each with a ~1-nm height corresponding to a single l-cystine molecule, creating the appearance of a pinwheel. These minor steps most likely reflect a splitting of the dislocation into six equivalent dislocations described by a Burgers vector having a magnitude of c/6. Consecutive images during crystal growth revealed a clockwise rotation of the pinwheel at the dislocation core (a left-handed screw) accompanied by continuous generation of new hillocks. Attachment of l-cystine molecules to both the minor and major steps on the surrounding terraces results in outward advancement of the steps with respect to the dislocation core (Fig. 3C and movie S2). The spiral growth pattern also is observed for d-cystine, the unnatural enantiomer, but with counterclockwise (a right-handed screw) rotation of the pinwheel (Fig. 3D and movie S3). A preference for screw dislocations of opposite handedness for enantiomeric crystals has been predicted (16).

Fig. 3

(A and B) Real-time in situ AFM images of a l-cystine crystal, acquired 12 min apart. A pair of hexagonal hillocks generated by two closely spaced dislocations serve as landmarks. (C and D) AFM images of a single dislocation center of (C) l-cystine and (D) d-cystine crystal during growth. (E and F) AFM image of a hexagonal growth hillock on the (001) face of l-cystine (E) before and (F) after addition of L-CDME (5 mg/liter; 0.02 mM), revealing roughening of the {100} steps due to step pinning. Images were acquired in aqueous solutions containing 2 mM l-cystine.

Quantitative determination of crystal growth rates at the near-molecular level were obtained by using AFM to measure the {100} step velocity (Fig. 4, A and B, and fig. S4). In order to achieve growth rates within a reasonable measurement timeframe, the l-cystine concentration was adjusted to 480 mg/liter (2 mM), which is five times larger than l-cystine solubility at room temperature (17). The {100} step velocity (Vo), determined through measurement of the step advance in successive images, was 11 nm/s, which is equivalent to 50 molecules/nm2 s. The step velocity was equivalent along all six directions, as expected for the hexagonal symmetry.

Fig. 4

(A and B) The position of the 5.6-nm-high {100} steps, as measured from the center of the spiral dislocations during growth in aqueous solutions containing 2 mM l-cystine with various concentrations of L-CDME or L-CME. The step velocities in the absence of inhibitor, determined from the slopes of the lines, are Vo = 11.4 ± 0.3 and 11.3 ± 0.1 nm/s, respectively. The SD is based on the average of the velocities determined for three independent steps. (C) Comparison of the effectiveness of L-CDME and L-CME on the inhibition of the {100} step velocity, expressed as V/Vo. (D) The mass yield of l-cystine crystals obtained after crystallization for 72 hours in the presence of various concentrations of L-CDME or L-CME. The error bars represent 1 SD based on three measurements for each inhibitor concentration.

Crystallization outcomes such as habit, chirality, and polymorphism can be influenced by tailored growth inhibitors that reduce crystallization rates through binding at specific step sites (18, 19). These inhibitors, which have been described as “imposters,” (15) consist of a binder moiety that emulates a critical structural element of the solute that attaches to a specific crystal site and a perturber moiety that obstructs the approach of additional solute molecules to neighboring sites, pinning step motion. Growth inhibitors may be monomers that closely resemble the solute, as demonstrated for amino acid and adipic acid crystals (2022). The effect of molecular inhibitors on the crystal growth of β-hematin—a synthetic malaria pigment—and their potential role as antimalarial drugs has been reported (23, 24). Polyvalent macromolecules capable of binding to multiple crystal sites, such as peptides and proteins, also can influence ice crystallization (25) and the formation of biominerals, including calcium carbonate (26, 27) and calcium oxalates (2830).

AFM revealed that addition of L-CDME, a structural mimic of l-cystine in which the carboxylate groups are replaced by methylester groups, resulted in roughening of the otherwise highly linear {100} step edges and rounding of the hillock corners, which is consistent with step pinning through adsorption of L-CDME at the {100} steps (Fig. 3, E and F). This effect was reversible because the steps once again become linear after addition of aqueous solutions containing only l-cystine (fig. S5). The step velocity decreased monotonically with increasing L-CDME concentration, becoming negligible above 30 mg/liter (Fig. 4, A and C). Similarly to the step roughening, the inhibitor effect was reversible, with the rates returning to the original value once the growth medium was replaced with aqueous solutions containing only l-cystine. The reduction of the step velocity was equivalent along all six directions in the (001) plane, as expected for the hexagonal symmetry of the crystal. These observations are consistent with attachment of L-CDME to the {100} step planes through a combination of intermolecular (cystine)S…S(L-CDME), (cystine)C(=O)O+H3N(L-CDME), and (cystine)N-H…O=C(L-CDME) interactions in a manner that mimics the attachment of l-cystine solute molecules at the {100} steps, with the ester methyl groups of bound L-CDME molecules blocking the attachment of l-cystine solute molecules at neighboring crystal sites. The steric bulk of the ester methyl group is not sufficient to prevent binding of L-CDME to the {100} steps.

The AFM measurements demonstrate that L-CDME slows growth along the naturally fast growth directions within the (001) plane. This microscopic behavior was mirrored in bulk crystallization by a gradual change in the crystal habit from (001) plates to small hexagonal needles oriented along the [001] direction as the L-CDME concentration was increased (Fig. 5A and fig. S6). At L-CDME concentrations as low as 5 mg/liter (0.02 mM; equivalent to 1% of the l-cystine concentration), the area of the (001) face was reduced by a factor of 1000, whereas the length of the needles was comparable with the thickness of the hexagonal plates grown in the absence of L-CDME (~30 μm), resulting in crystals that were 1000 times smaller. Moreover, the total crystallization yields decreased with increasing L-CDME concentration, approaching complete inhibition above 2 mg/liter (Fig. 4D), which is consistent with the AFM observations. The dramatic change in crystal habit in the presence of L-CDME and L-CME may influence the aggregation of crystals and their attachment to renal cells, which are also critical steps in stone formation (31). L-CDME also promoted the formation of small amounts of minute crystals of tetragonal l-cystine (Fig. 5B). The formation of the tetragonal polymorph, which ordinarily crystallizes only under more basic conditions (pH > 8) (32), can be attributed to the suppressed growth of the hexagonal form by L-CDME.

Fig. 5

(A) Minute l-cystine crystals grown in the presence of L-CDME (5 mg/liter; 0.02 mM) exhibit a hexagonal needle-like habit with prominent {100} faces and high c/a aspect ratios, approaching 30 for many crystals. (B) Small quantities of the tetragonal P41 polymorph are formed in the presence of L-CDME (5 mg/liter). (C) L-CME (10 mg/liter; 0.04 mM) produces tapered hexagonal needles with six {101} faces. Some crystals exhibit the tapered habit at both ends of the crystal, as expected for the crystal symmetry (fig. S7). The observation of only one half of the tapered crystal suggests that growth often begins on a surface. (D) The tapered needles formed in the presence of 10 mg/liter L-CME occasionally grow from a nidus that may be an amorphous l-cystine particle or a microscopic unidentified foreign object. Crystal growth was performed in aqueous solutions containing 3 mM l-cystine (700 mg/liter).

The unsymmetrical L-CME, a cystine mimic with only one ester methyl group, also inhibited l-cystine crystallization. AFM measurements revealed that the {100} step velocity declined with increasing L-CME concentration, but to a lesser extent than L-CDME (Fig. 4, B and C). L-CME also reduced the size of l-cystine crystals substantially, and it reduced the crystal yield, but to a lesser extent than did L-CDME (Fig. 4D). Step roughening was observed in the presence of L-CME, but at higher concentrations as compared with that of L-CDME (fig. S5). These observations are consistent with weaker inhibition of L-CME as compared with that of L-CDME. The l-cystine crystals grown in the presence of L-CME exhibited a tapered hexagonal habit, with six faces intersecting the basal (001) plane at an angle of 85°, which is attributable to the emergence of six equivalent {101} surfaces (Fig. 5, C and D, and fig. S7). The different crystal habits generated by L-CDME and L-CME provide interesting insight into the binding modes of these inhibitors at the l-cystine crystal surface.

The P6122 space group symmetry generates inequivalent projections of the l-cystine molecules on each flank of a hexagonal hillock. Each projection winds around the hillock by translations of ±c/6 on adjacent faces. In the 5.6-nm span of the hillock, diad axes create two distinct pairs of symmetry-related projections in which the l-cystine molecules are oriented in opposite directions along c—for example, C1/C3 and C4/C6 on the (010) face (Fig. 6). The two remaining sites, located on the diad special positions, create two additional distinct projections along directions containing the aforementioned intermolecular S…S contacts. Molecular models reveal that binding of L-CDME at (001)∩{100} ledge sites [illustrated in Fig. 6 by the intersection of (001) and (010) planes] is precluded by the steric obstruction introduced by one of the ester methyl groups (Fig. 6). Consequently, attachment of L-CDME must occur at l-cystine sites above the ledge sites. It is reasonable to suggest that L-CDME would bind preferentially at the diad sites because of the additional S…S interactions—for example, the C2 and C5 sites on the (010) face. Binding of L-CDME would prevent steps from advancing further above the inhibitor site, leading to step bunching that would terminate in {100} planes and producing hexagonal needles with small cross sections. L-CDME binding to other sites cannot be excluded; the projections of l-cystine molecules at each site of a diad-related pair (blue-blue or green-green on each hillock face in Fig. 6) differ only with respect to their “up” or “down” orientation (denoted by the white arrows). The C2 symmetry of L-CDME would result in equivalent binding to either site of a diad pair.

Fig. 6

(Top left) Schematic representation of an l-cystine hillock, color-coded to denote the distinct projections of l-cystine along each flank of the 5.6-nm-high hillock. The six flanks are depicted in the unraveled version of the hillock on the right. Equivalent projections on adjoining flanks are related by the 61 screw axis. Diad axes, denoted by black ovals, generate two identical projections that differ only with respect to their up or down orientation (denoted by white arrows). (Bottom right) L-CDME binding at a (001)∩{010} ledge site is frustrated by one of the ester methyl group, emphasized here by the white arc. (Bottom left) L-CDME binding to the (010) step: The ester methyl groups of L-CDME, depicted by purple spheres, block attachment of l-cystine to adjacent sites. Binding to the C1 and C4 sites on the {010} step allows S…S and hydrogen-bonding interactions between l-cystine molecules projecting from the step surface and L-CDME. In contrast, in the up orientation the carboxylic acid terminus of L-CME can bind to the ledge sites in a manner like l-cystine, through hydrogen-bonding interaction I, at the (001) terrace, as well as S…S interactions and hydrogen bonding at the (010) step. Crystallographically equivalent (001)∩(010) ledge sites, depicted here for the C6″-C1 and C6-C1′ combinations, are spaced at intervals of c, generating a vicinal surface equivalent to a (011) plane. This plane intersects the (001) plane at an angle of 85°, which is consistent with the tapered faces observed in the presence of L-CME.

In contrast, L-CME can bind at (001)∩{100} ledge sites in its up orientation because the carboxylic acid terminus is indistinguishable from the ends of l-cystine (the ester methyl group would prevent binding at the ledge in the down orientation). The terrace of the ledge site effectively breaks the twofold up-down symmetry of the diad-related pairs. Consequently, each flank of the hexagonal hillock contains six crystallographically, and therefore chemically, inequivalent (001)∩{100} ledge sites along the 5.6-nm span of the c axis (fig. S8). The tapered crystals grown in the presence of L-CME can be explained only by highly selective binding of L-CME at one of these sites. Crystal step motion would then be pinned at these sites, repeating at an interval of c, creating a vicinal face with a tangent plane intersecting the (001) plane at an angle of 85°, which is identical to the angle of the tapered faces on the macroscopic crystals and assignable to {101} planes. The different mode of binding in the presence of L-CDME and L-CME, particularly the discrimination of L-CME for one of the six possible ledge sites, is a remarkable illustration of molecular recognition between tailored inhibitors and crystal surface sites. Moreover, the step velocity measurements, step roughening, crystal habit effects, and crystal yields represent a rare example of a correspondence between inhibitor effects at the microscopic level and macroscopic crystallization behavior.

L-CDME is extraordinarily effective compared with inhibitors examined for other kidney stone–forming materials. Molecular additives have been found to be substantially less effective as crystal growth inhibitors of COM as compared with that of their polymeric forms; for example, aspartic acid is more than 1000 times less active for COM inhibition than poly(aspartic acid), in which the carboxylate side chains are thought to bind to calcium sites on the crystal surfaces. The effective inhibition observed for polymers can be attributed to the entropic benefit associated with binding of multiple carboxylate groups on a single chain. Surprisingly, the concentrations at which L-CDME becomes effective for l-cystine inhibition (~2 mg/liter) is comparable with those observed for poly(amino acid) inhibitors of COM crystallization (28). The importance of molecular recognition between L-CDME and the l-cystine step planes is underscored by our observations that other additives with proximal carboxylate and amine groups, such as those in l-cystine, had a negligible influence on l-cystine growth. For example, L-cysteine, the thiol relative of l-cystine, reduced the size of l-cystine crystals somewhat (at 10 mg/liter), but its effect on crystallization yield was negligible (fig. S9). Urinary proteins such as osteopontin, human serum albumin, and Tamm-Horsfall protein afforded only modest reductions in crystallization yield at concentrations comparable with physiological values (2 mg/liter), suggesting a negligible role for these substances in the regulation of cystine stone formation.

Collectively, the AFM and bulk crystallization behavior for l-cystine suggest that L-CDME is a viable therapeutic agent for the prevention of l-cystine kidney stones. This approach to stone prevention uses a potentially benign crystal growth inhibitor at low concentrations rather than drugs that rely on a chemical reaction with l-cystine (l-cystine–binding thiol drugs), increases in urine alkalinity (which are often accompanied by undesirable side effects), or dramatic increases in urine volume (which can be unreliable owing to patient nonadherance). The reduction in mass yield in the presence of inhibitors is a kinetic effect that maintains a metastable supersaturated l-cystine concentration, but from a pathological perspective this is a sufficient condition for preventing stone formation. l-cystine stone formers typically have urinary l-cystine concentrations ranging from 250 to 1000 mg/liter (equivalent to 1 to 4 mM), which is comparable with the concentrations we used for the AFM and bulk crystallization studies. Therefore, L-CDME concentrations near 2 mg/liter (<0.01 mM), at which inhibition of l-cystine growth was highly effective, may be adequate for therapeutic effect. Cell culture data, acquired for the purpose of evaluating cystine exodus from lysosomes, show loss of cell viability at approximately 1 mM L-CDME, and studies in rats, performed to measure oxidative stress in the brain cortex, demonstrated adverse effects at dosages of approximately 500 mg/kg (mass of rat) per day (3335). Although the pharmacokinetics of L-CDME are not well known, on the basis of typical daily urine volumes a L-CDME dose of 10 to 50 mg per day—far below toxic levels but greater than the amount needed for crystal growth inhibition in vitro—may prove sufficient to achieve adequate L-CDME concentrations in urine for crystal growth inhibition in vivo.

Supporting Online Material

Materials and Methods

Figs. S1 to S10

Table S1


Movies S1 to S3

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was supported primarily by NIH (NIDDK R01-DK068551) and the NYU Molecular Design Institute. The authors also acknowledge support from the Office of Rare Disease Research (1U54DK083908-01) and the Advanced Photon Source, ChemMatCARS Sector 15, which is principally supported by NSF/U.S. Department of Energy (DOE) (NSF/DOE; CHE-0535644). The authors thank C. Hu, M. Li, Y.-S. Chen, and G. Kowach for technical assistance and B. Kahr for helpful discussions.
View Abstract

Stay Connected to Science

Navigate This Article