The existence of intracrystalline proteins and amino acids in calcium oxalate monohydrate was demonstrated by X-ray synchrotron diffraction studies. Their presence has implications for the destruction of calcium oxalate crystals formed in the urinary tract and the prevention of kidney stones.
Introduction: Although proteins are present in human kidney stones, their role in stone pathogenesis remains unknown. This investigation aimed to characterize the nature of the relationship between the organic and mineral phases in calcium oxalate monohydrate (COM) crystals grown in human urine and in aqueous solutions of proteins and amino acids to clarify the function of proteins in urolithiasis.
Methods: COM crystals were grown in human urine and in aqueous solutions containing either human prothrombin (PT), Tamm-Horsfall glycoprotein (THG), aspartic acid (Asp), aspartic acid dimer (AspAsp), glutamic acid (Glu), glutamic acid dimer (GluGlu), or γ-carboxyglutamic acid (Gla). Controls consisted of COM crystals precipitated from pure inorganic solutions or from human urine that had been ultrafiltered to remove macromolecules. Synchrotron X-ray diffraction with Rietveld whole-pattern peak fitting and profile analysis was used to determine nonuniform crystal strain and crystallite size in polycrystalline samples.
Results: Crystals precipitated from ultrafiltered urine had lower nonuniform strain than those grown in urine or in aqueous PT solution. Nonuniform strain was much lower in crystals grown in distilled water or in the presence of THG. For the amino acids, the highest nonuniform strain was exhibited by crystals grown in Gla solution, followed by Glu. Crystallite size was inversely related to nonuniform strain, with the effect being significantly less for amino acids than for macromolecules.
Conclusions: Selected proteins and amino acids associated with COM crystals are intracrystalline. Although their incorporation into the mineral bulk would be expected to affect the rate of crystal growth, they also have the potential to influence the phagocytosis and intracellular destruction of any crystals nucleated and trapped within the renal collecting system. Crystals impregnated with protein would be more susceptible to digestion by cellular proteases, which would provide access to the crystal core, thereby facilitating further proteolytic degradation and mineral dissolution. We therefore propose that intracrystalline proteins may constitute a natural form of defense against renal stone formation.
Nucleation of crystals and their subsequent entrapment within the renal collecting system are obligate steps in the formation of kidney stones, most of which are composed principally of calcium oxalate (CaOx). The likelihood of crystal nucleation obviously depends on the degree to which urine is supersaturated with CaOx. However, retention of crystals will be favored by any process that causes sufficient enlargement to prevent their harmless expulsion in the urinary stream or by factors that encourage their adhesion to the renal epithelium. The well-documented association of urinary proteins with CaOx kidney stones(1–3) has led to the widespread belief that they might regulate crystallization processes during urolithiasis in a manner similar to that exerted by “control” proteins in healthy biomineralization.(4) However, whether they act as promoters or inadequate retardants of crystal nucleation, growth, or aggregation; whether they are simply nonselective, ineffectual binders; or whether they are products of the stones that then become incorporated into the structure, they will still be associated with the final calculus. Therefore, mere demonstration that protein is present in stone matrices provides few clues about the mechanism by which it came to be there or what effects it may have wrought in the process.
It is now widely accepted that some urinary proteins may help to prevent stone formation by inducing the formation of small crystals or by preventing their aggregation into large clusters likely to be retained within the renal collecting system.(2,3) It is also becoming increasingly recognized that stone formation almost certainly involves direct interaction between newly formed crystals and cells of the renal epithelium, because exposure of cultured renal cells to CaOx results in attachment of the crystals,(5–7) which then are internalized.(8–11) Proteins on crystal surfaces may therefore also act as modulators of the attachment to, and phagocytosis of CaOx crystals by, renal epithelial cells, an assumption supported by at least one report that the adhesion of crystals to cultured cells is inhibited by several macromolecules.(6) However, while the potential roles of superficial proteins may seem obvious, less apparent are those of intracrystalline proteins.
It has been known for many years that CaOx crystals generated from human urine are associated with relatively few selected proteins compared with the large number known to be present in healthy urine.(12) Predominant among these are a urinary form of prothrombin fragment 1 (UPTF1)(13) and osteopontin (OPN),(14) with serum albumin and various derivatives of inter-α-inhibitor being present in lesser quantities.(15) In contrast, the most abundant urinary protein, Tamm-Horsfall glycoprotein (THG) is reportedly absent from demineralized extracts of urinary CaOx crystals.(12) We have recently demonstrated that CaOx crystals deposited from human urine that contains most of its normal protein complement possess a complex, ordered ultrastructure comprised of labyrinthine tunnels and crystalline particles intimately associated with organic material.(4,16) The quantity of organic material diminishes after crystal fracture and proteolytic digestion, showing that it consists principally of protein. In marked contrast, no organic material can be detected in crystals derived from urine that has been ultrafiltered to remove all macromolecules with molecular mass greater than 10 kDa.(4,16)
Those observations led us to postulate that urinary proteins may fulfill a hitherto unsuspected role in the prevention of stone formation. Pure inorganic CaOx crystals internalized by renal cells in culture are known to dissolve over a period of several weeks.(17–21) However, cribriform crystals riddled with macromolecules would be expected to be dismantled and dissolve more rapidly as a result of attack by intracellular and lysosomal proteases released during the phagocytic response, because they would be more vulnerable to enzymatic degradation than solid minerals. Thus, proteins distributed throughout the crystal structure should facilitate excavation into channels deep inside the crystal, enabling burrowing into the crystalline core and vastly increasing the surface area available for further digestive processes and mineral dissolution. Intracrystalline proteins may therefore fulfil an important function as mediators in the routine destruction of retained crystals, and therefore act as a natural form of defense against stone pathogenesis.(4,16)
Current evidence showing the existence of intracrystalline proteins in urinary CaOx has consisted of SDS-PAGE and Western blot analysis,(12,13,15) as well as field emission scanning electron microscopic examination of washed, fractured crystals subjected to subsequent protease treatment.(4,16) Although showing unequivocally that urinary CaOx crystals possess an internal ultrastructure containing protein, those studies provided no direct information about the physical basis of the relationship between the organic and inorganic crystal phases or the nature of the mineral itself. Consequently, the possibility could not be discounted that the CaOx structures examined were not single crystals infiltrated with protein, but rather, crystallites or amorphous mineral stabilized by organic material and assembled into what appeared, at least externally, to be individual structures. A natural equivalent occurs in triradiate calcite spicule of the sponge Calcarea clathrina, which consists of a crystalline nucleus at the junction of the three rays, each of which contains an amorphous core and a crystalline sheath enveloping the entire structure. Despite this complexity, Addadi et al. have shown that the spicule behaves as a single crystal under light microscopy and synchrotron X-ray diffraction (SXRD) and that 80% of scattering intensity is contributed by the amorphous material.(22) Broadening of diffraction peaks enabled the same authors to show that the spicule of the sea urchin consists of a single crystal impregnated throughout with intercalated proteins. Located at the boundaries of the crystal domains, the proteins adsorb on to preferred crystallographic faces, where they induce crystal defects that reduce crystallite size and alter morphology and cleavage fracture characteristics.(23,24)
The aim of this study was to use SXRD analysis of CaOx crystals precipitated from ultrafiltered (UF) urine, centrifuged and filtered (CF) urine, and aqueous solutions containing selected urinary proteins to clarify the nature of the relationship between the organic and mineral phases. CaOx exists as three polymorphs: calcium oxalate monohydrate (COM), dihydrate (COD), and trihydrate (COT). COM was selected for the study because it is regarded as the critical polymorph in urolithiasis,(25,26) is thermodynamically stable, and is the dominant phase found in CaOx renal calculi.(27) Furthermore, because the binding of proteins to mineral crystals will depend significantly on their primary amino acid composition and sequence, similar studies were undertaken using COM crystals grown in the presence of amino acids, which we have shown previously to bind to CaOx.(28) Polycrystalline, rather than single crystal analysis, was carried out because the amount of protein associated with individual crystals in a given preparation is variable,(4) and an average response was required to enable comparisons to be made. Whole-pattern XRD analysis techniques have improved sufficiently to enable this to be accomplished.
MATERIALS AND METHODS
Chemicals and biochemicals
All reagents were of the highest purity commercially available. Sodium oxalate (NaOx), gelatin, Asp, AspAsp, Glu, and GluGlu were obtained from BDH Chemicals Australia (Kilsyth, Victoria, Australia). Calcium chloride, Tris (hydroxymethyl aminomethane), and Gla were supplied by the Sigma Chemical Company (St Louis, MO, USA). Proteinase K was obtained from Boehringer (Mannheim, Germany). COM was purchased from Ajax Chemicals (New South Wales, Australia). THG(29) and PT(30) were prepared as we have described previously. Double distilled water was used in all experiments.
COM crystals grown from aqueous solutions:
Crystals of COM were generated by mixing 5 ml each of aqueous solutions of 0.15 M CaCl2 and 0.15 M NaOx into distilled water containing the dissolved protein or amino acid at 37°C. PT (10 mg/liter) was prepared in distilled water. PT was used as a substitute for its derivative, UPTF1, a major component of CaOx urinary crystals(13) and to which it is structurally closely related. PT is easier to purify than UPTF1 and is present in demineralized extracts of CaOx crystals precipitated from human urine.(31) The concentration of THG used (20 mg/liter) was within its normal range in human urine. All amino acids were used at a final concentration of 1 g/liter. CaCl2 and NaOx solutions were added at a rate of 0.4 ml/h from glass syringes using an infusion pump (Sage Instruments, Cambridge, MA, USA). The mixture was gently agitated with an overhead stirrer fitted with a glass stirring rod and incubated for 13 h at 37°C. Precipitated crystals were separated by vacuum filtration, washed with a saturated solution of CaOx, and dried under nitrogen.
COM grown in gelatin:
COM crystals were grown in gelatin at room temperature in accordance with the method outlined by Henisch.(32) Solutions of CaCl2 (0.15 M) and NaOx (0.15 M) were introduced into each side of a glass U tube containing a plug of gelatin gel at the U bend. After a growth period of 12 weeks, COM crystals were removed from the gelatin layer with tweezers, washed with saturated CaOx solution, separated by vacuum filtration, dried under nitrogen, and stored in a dessicator. These slowly grown crystals were assumed to be relatively free of lattice strain and were used as the “strain free” reference for Gaussian and Lorentzian contribution calculations in Eqs. 2 and 3. This assumption was later confirmed by the finding of significantly narrower X-ray diffraction peaks in comparison with COM grown in distilled water.
Collection and treatment of urine samples:
Urine samples were collected over a 24-h period from healthy individuals who had no history of kidney stone disease. The samples were refrigerated during the collection period and during storage before use. Absence of blood from the specimens was confirmed using Multistix test strips (Miles Laboratories Mulgrave, Victoria, Australia). The samples were pooled and centrifuged at 8000g for 15 minutes at 20°C in a Beckman J2–21M/E centrifuge (Beckman Instruments, Palo Alto, CA, USA). The supernatant was filtered through 0.22-μm Millipore filters (GVWP 14250; Millipore Corp., Bedford, MA, USA). A portion of the centrifuged and filtered (CF) urine was ultrafiltered (UF) using an Amicon hollow fiber bundle (Amicon Corp., Danvers, MA, USA), with a nominal molecular mass cut-off of 10 kDa.
COM crystals from human urine:
COM crystals were precipitated from CF and UF urine as described elsewhere.(12) Briefly, the metastable limit of urine with respect to CaOx, that is, the minimum amount of oxalate required to elicit spontaneous detectable crystallization using a Coulter Counter, was first determined by titration with NaOx solution. A standard load of oxalate in excess of that limit was then added to induce precipitation, and the suspension was incubated for 2 h at 37°C in a shaking water bath.
Proteolytic digestion of COM crystals:
A sample of calcium oxalate crystals (5 mg) grown in CF urine was fractured using a diamond cell (High Pressure Diamond Optics, Tuscon, AZ, USA) and incubated at 37°C for 12 h in 2 ml of an aqueous saturated solution of CaOx containing 0.25 mg/ml Proteinase K and 12.5 mM Tris-HCl buffer, pH 7.0. After protease digestion, the crystals were separated by vacuum filtration, washed with a saturated solution of CaOx, and dried under nitrogen. The same procedure was carried out for COM grown in ultrafiltered urine.
Scanning electron microscopy
Crystals were mounted on double-sided carbon tape fixed on aluminum stubs, coated with carbon (25 nm, Speedivac model 12E6), and examined using a JEOL 6300F field emission scanning electron microscope (JEOL, Tokyo, Japan). Imaging was performed with secondary electrons generated by a 3-keV primary electron beam.
Synchrotron X-ray diffraction
Collection of data:
Synchrotron X-ray diffraction (SXRD) patterns were collected on BIGDIFF, a synchrotron diffractometer installed on Beamline 20B, Australian National Beamline Facility (ANBF) at the Photon Factory Synchrotron-Radiation Facility within the National Laboratory for High Energy Physics (KEK), Tsukuba, Japan. The radius of the camera is 573 mm, which gives a scaling factor of 1°/cm along the circumference. The monochromator was set at a wavelength close to the CuKα radiation doublet, and the incident beam dimensions were fixed at a height of 0.8 mm and width of 10 mm. Sample run times were 900 s for both COM and a lanthanum hexaboride (LaB6) line profile standard and 300 s for the silicon wavelength standard. Four imaging plates (400 × 200 mm; Fuji Photo Film Co., Tokyo, Japan) were used to record the diffraction patterns. The diffraction pattern image on the plate was digitized to produce an 8-bit image that gave an angular resolution of 0.01° 2θ. This is less than the ultimate resolution determined by the footprint of the capillary (0.05° for a capillary diameter of 0.5 mm). BIGDIFF is fitted with eight radioactive fiducial markers that were used to determine the 2θ offset for each imaging plate by using the known 2θ values for the markers, as determined by tests with specimens of known Bragg angle (θ) values (i.e., silicon).
Crystals were packed into glass capillaries made of low-absorption lithium borate with a 0.5 mm internal diameter and a wall thickness of 0.01 mm.
The strain contributing to line broadening is described as nonuniform strain, which varies from one crystal grain to another, whereas uniform strain (macrostrain) causes shifts in the Bragg peak positions.(33) Nonuniform strain information was obtained by subtracting the instrumental broadening contribution and separating crystallite size effects from strain effects. For Rietveld analysis, the instrumental contribution was determined using a LaB6 standard (NIST SRM 660),(34) which has little broadening caused by particle size or strain.
Rietveld whole-pattern fitting of X-ray diffractograms:
The X-ray data were subjected to Rietveld refinement(35) using Rietica for Windows 95/98/NT Version 1.72 (ANSTO, Lucas Heights, New South Wales, Australia).(36) Gaussian and Lorentzian contributions to X-ray peak profiles were obtained using a Voigt function.(37,38) These contributions, expressed as whole-pattern refined values, were used in Eqs. 2 and 3 to calculate lattice strain and crystallite size. Parameters refined were phase scale, overall thermal parameter, unit cell parameters (a, b, c, β), atomic coordinates for Ca1, Ca2, background (Cheby 1), zero offset, Gaussian peak broadening coefficient (U), crystallite size, preferred orientation, and Gaussian peak anisotropy (Uanis). The model was set up for COM using the initial parameters of the space group (P 1 21/c 1) and cell parameters (a, b, c, β) obtained from the Inorganic Crystal Structure Database, collection code 30782.(39) A silicon NBS powder standard (NIST SRM 640)(40) was used to determine accurately the wavelength of the X-ray radiation used and to establish the 2θ offset for each image plate according to the fiducial marker images. SXRD data obtained from the LaB6 powder standard was used to construct a plot of full width of the peak at half maximum height (FWHM) versus 2θ to determine the instrument profile. Coefficients U, V, and W of Eq. 1(41,42) were obtained for use in the Rietveld refinements of all X-ray crystal data. The U value was used as an initial value which was open to refinement, but the V and W values were kept constant throughout each refinement, because they are functions of the instrumental conditions.
The nonuniform lattice strain (root mean square) strain, (ϵ2)1/2, was calculated using Eq. 2,(41,42) where U and Ur were determined from the Gaussian contribution to the peak shape profile of the sample and a strain-free reference, respectively, by Rietica.
The error in (ϵ2)1/2 was calculated using the total derivative:
Average crystallite size (D) was computed by Rietica and was derived from the Lorentzian contribution to the peak shape profile according to Eq. 3,(36) where the secθ-dependent term describes particle-size effects and λ is the wavelength of X-rays used (0.149576 nm).
The errors in nonuniform strain and crystallite size are estimated from the statistical variation of the diffraction data and do not include systematic errors.
Crystals were examined under crossed polarizers using an Olympus BH-2 microscope (Olympus, Tokyo, Japan) and ×1000 magnification.
Assessment of the significance of differences in nonuniform strain, crystallite size, and linear relationships was based on the null hypothesis, that is, there is no difference in the two quantities being compared. Both the z value (Eq. 4) and probability (p) of the null hypothesis being true are presented.
where O1 and O2 are the observed values, E1 and E2 are the expected values (null hypothesis, E1 − E2 = 0), σ1 and σ2 are the root mean standard errors of the observed values, and n1 and n2 are the population sizes of the observations.
Because the general aim of this study was to clarify the relationship between the organic and mineral phases in COM, SXRD analysis with Rietveld whole-pattern fitting analysis was carried out in association with light and scanning electron microscopy examination of crystals. This combination of techniques enabled quantification of crystal disorder, expressed in terms of nonuniform crystal strain and crystallite size, to be correlated with direct visualization of the crystal morphology.
Calcium oxalate morphology
COM was the only polymorph of CaOx generated from aqueous solutions and the major polymorph precipitated from UF and CF urine. All crystals exhibited the classic COM hexagonal plate morphology, but some thinning of the plates and rounding of edges was observed for those grown in UF urine, CF urine, and in the PT solution. X-ray diffraction lines of the sole minor phase, COD (<5%), did not interfere significantly with COM diffraction lines of the urinary crystals, and both single and two phase Rietveld refinement of the crystallographic parameters produced the same results. Peak shifts in the SXRD patterns were negligible, and the crystallographic parameters of all samples fell in a very narrow range (0.6294 nm < a < 0.6303 nm, 1.4585 nm < b < 1.4600 nm, 1.0120 nm < c < 1.0123 nm, 109.41° < β < 109.49°). These are in excellent agreement with published data for COM (a = 0.6290 nm, b = 1.4583 nm, c = 1.0116 nm, and β = 109.50°).(39)
Microscopic examination under crossed polarizers of a random selection of crystals (6–50 μm) grown in each of the various aqueous and urine media showed the classic four distinct extinctions per 360° rotation of birefringent single crystals in every instance. The extinctions were uniform and demonstrated that, optically, all behaved as single crystals rather than individual structures comprised of aggregates of randomly oriented microcrystals.
Electron microscopy of fractured crystals grown in distilled water (Fig. 1A) and in UF urine (Fig. 1B) revealed solid interiors with little or no signs of pitting. Identical observations were made with fractured crystals precipitated from amino acid and THG solutions (data not shown). In contrast, Figs. 1C and 1D show a varying cross-sectional density for COM grown in the presence of PT and in CF urine, indicating a heterogeneous internal structure. Protease treatment led to significant material losses, with most erosion occurring at crystal centers (Fig. 2A). Higher magnification (Fig. 2B) shows that the boundaries of the residual material consisted mainly of columns arrayed along the (100) face.
The FWHM trends of SXRD peaks for each type of crystal, as defined by Eq. 1 and determined by Rietveld refinement, showed a marked variation in the rate of peak broadening as a function of 2θ (Fig. 3). COM grown in gelatin showed the smallest broadening rate, which confirmed it to be the least disordered. A greater rate of broadening, and hence greater disorder, was observed for COM grown in distilled water (control) and in THG solution. Solutions of GluGlu, Asp, AspAsp, and Glu produced COM with progressively greater broadening rates, whereas the most marked increases in broadening rates were found for UF and CF urine, Gla, and PT. Protease treatment of COM grown in CF urine reduced its broadening rate markedly.
Nonuniform strain and crystallite sizes calculated from the fitted peak width trends using Eqs. 2 and 3 range from 0.040% to 0.128% and from 0.31 to 2.59 μm, respectively, (Fig. 4A). Rietveld refinement uses a single variable to describe peak width variation as a function of diffraction angle (U, Eq. 1) for the whole X-ray diffraction pattern. Hence, it can return only single values for crystallite size and nonuniform strain, which for a polycrystalline sample is an average over both the assemblage of crystallites in the crystal and the three crystallographic axes. The morphology of the crystallites thus cannot be deduced as they are treated as spheres, nor can anisotropy in the nonuniform strain be determined. The estimated relative SD error for both nonuniform strain and crystallite size was approximately 10%. COM grown in distilled water and in THG solution was found to have the least nonuniform strain and some of the largest crystallite sizes, relative to crystals grown in gelatin. The presence of GluGlu, Asp, or AspAsp in solution gave rise to a slight increase in nonuniform strain but had little or no effect on crystallite size, while Glu led to a more significant increase in nonuniform strain (z = 5.6, p ∼ 0) and decrease in crystallite size (z = 1.8, p < 0.04). COM grown in UF urine was more strained (z = 2.74, p < 0.003) and had a significantly smaller crystallite size (z = 5.7, p ∼ 0), which was approximately 20% of that of the control crystals grown in distilled water. In contrast, COM grown in Gla solution was even more strained (z = 2.5, p < 0.006) but had a much greater crystallite size (z = 6.9, p ∼ 0) than that grown in UF urine, which was approximately 40% of that recorded in the control. The highest nonuniform strains and smallest crystallite sizes were found for COM grown in the PT solution (0.128 ± 0.008% and 0.31 ± 0.03 μm, respectively) and in CF urine (0.128 ± 0.009% and 0.31 ± 0.03 μm, respectively). Protease treatment of the crystals grown in CF urine reduced nonuniform strain and increased crystallite size to the levels observed for crystals grown in UF urine (z = 3.8, p ∼0; z = 1.5, p < 0.07).
The data show that two reciprocal relationships exist between nonuniform strain and crystallite size, with proteins other than THG generating a smaller crystallite size than the amino acids for same degree of nonuniform strain (Fig. 4A). The slopes of the transformed relationships (Fig. 4B) show that the two populations, proteins (slope = 32.3 ± 0.6) and amino acids (slope = 6.6 ± 0.3), are significantly different (z = 95, p ∼ 0). Figure 4 thus clearly shows that the degree of crystal disorder, that is, increased nonuniform strain and decreased crystallite size, is influenced by the size and chemical composition of the guest molecules.
Although it has been known for at least 10 years that CaOx crystals precipitated from human urine are irreversibly associated with proteins,(12) there has been continuing debate as to whether they are bound only to the crystal surface or are actually located within the mineral structure. The distinction is critical, because the presence of proteins inside crystals could influence their phagocytosis by epithelial cells of the renal collecting system and facilitate their subsequent intracellular disposal,(4,16) a possibility which, if true, has implications for the design of drugs for preventing stone formation. The broad objective of this study was to obtain quantitative data to corroborate our previous qualitative findings, which indicated that selected proteins associated with CaOx crystals precipitated from human urine are intracrystalline. All crystals have imperfections caused by dislocations and/or foreign molecules that create defects and disorder and hence are an assemblage of smaller defect-free perfectly crystalline subregions (crystallites) whose alignment will exhibit some degree of disorder. An increase in the number of defects thus causes a reduction in crystallite size and increases the randomness of their orientation, the latter being manifest as an increase in nonuniform lattice strain. Consequently, a decrease in crystallite size and associated increase in nonuniform lattice strain, relative to a control, would indicate the presence of extraneous molecules inserted into the crystal bulk and thereby constitute proof that they are intracrystalline.(43)
Our specific aim was, therefore, to measure crystal strain and crystallite domain size in COM crystals precipitated from human urine and from aqueous solutions of selected proteins and amino acids to determine whether the organic additives are occluded within the mineral bulk. PT was selected for study because it has been strongly implicated in the pathogenesis of renal stones.(3) It is a potent inhibitor of both CaOx crystal growth and aggregation in an inorganic medium(44,45) and in undiluted UF urine.(46) The protein's inhibitory potency, as well as its strong affinity for CaOx crystals deposited from urine,(12) can be attributed to its Gla domain.(45) Located in the N-terminal region of the molecule, this region contains 10 Gla residues, an amino acid renowned for its extraordinary capacity for binding calcium ions(45) and which we have shown previously to adsorb strongly onto COM.(28) For those reasons, Gla was also studied. Asp and Glu were included because of their moderate adsorption affinity for COM relative to Gla,(28) as were their dimers, AspAsp and GluGlu, whose use enabled determination of the effect of doubling single amino acid chain lengths. Although lacking Gla, THG was studied because it is the most abundant urinary protein and has been reported to function in urine both as an inhibitor and promoter of CaOx crystallization(47) and also as a regulator of COM adhesion to urothelial cells.(48) A further reason for studying PT and THG is that PT is present in demineralized extracts of alkali-washed crystals precipitated from healthy male urine,(31) whereas THG is not.(12) Crystals were also grown in urine that had been centrifuged and filtered, as well as in the same urine after ultrafiltration to remove all macromolecules with molecular masses greater than 10 kDa. This enabled the generation of crystals from urine samples with an identical complement of low molecular weight components, but markedly different macromolecular compositions. COM grown in distilled water was used as the control to allow effects resulting specifically from the influence of proteins and amino acids to be identified.
COM was the only CaOx polymorph formed in distilled water and in the presence of aqueous amino acid and protein solutions, whereas two polymorphs, COM and COD, were precipitated in UF urine and in CF urine. This is in agreement with the view that the urinary matrix stabilizes the less thermodynamically stable COD phase.(49) The thinning of the hexagonal-shaped plates and rounding of edges observed for COM crystals grown in the aqueous PT solution, UF urine, and CF urine indicate a reduction in the growth rates of the (100) faces. This reduction, caused by adsorbed species, produces the familiar “coffin”-shaped COM crystals commonly found in urine. All refined Rietveld crystallographic values were close to those in published data,(39) indicating that the unit cell dimensions were not altered significantly by the presence of amino acids and proteins. Thus, the organic molecules, where present, were either surface-adsorbed or dispersed about the crystallites or both, and not intercalated in the COM crystallite lattices.
Electron microscopy of fractured crystals grown in distilled water, the amino acid solutions, and UF urine showed solid interiors, indicating that these media have little or no impact on the homogeneity of the crystal structure. However, transverse sections of fractured crystals grown in PT solution and in CF urine showed marked textural variation. Furthermore, when the crystals were treated with protease, only those grown in CF urine and PT solution showed significant erosion, particularly near their centers. As protease specifically digests proteins, it follows that the crystal cavities must have resulted, at least in part, from the removal of proteins. However, the cavities (0.3–1 μm) are undoubtedly far larger than the dimensions of individual proteins,(22) which suggests that selected, possibly aggregated proteins might induce nucleation of mineral around them. It is possible that such aggregation might be caused by the addition of the high oxalate concentration used to induce artificial nucleation of crystals in CF urine, and consequently that crystals formed in vivo would lack the intracrystalline spaces evident in those nucleate in vitro. However, similar intracrystalline spaces were present in COM crystals that were deposited from an aqueous PT solution by the slow addition of calcium and oxalate ions, and we have also observed a similar internal structure in CaOx crystals formed naturally in fresh urine on cooling (results not shown). Moreover, physiological induction of crystalluria within the renal collecting system probably occurs in response to spasmodic pulses of high local oxalate and calcium concentrations resulting from dehydration or ingestion of meals rich in those ions. Alternatively, proteins with a strong affinity for CaOx could cluster into large assemblages of individual molecules competing for limited binding sites on the surfaces of nascent crystals. Concentration of organic material toward the crystal cores can be explained easily in terms of the relative quantities of protein and solute ions in solution. As solute deposition continues, protein becomes depleted, thereby allowing the deposition of CaOx relatively free of organic material and the formation of a solid mineral shell enveloping the protein-mineral composite that was once the crystal embryo. The orientation of the residual columnar material in the crystals grown in CF urine and then treated with protease (Fig. 2B) indicates that proteins prefer to dock on to the COM (100) faces. The reason for this is not known, but it is likely that preferential binding to those faces may be related to the matching of protein binding amino acid groups with the position of calcium ions on the crystal faces.
COM grown in Asp, AspAsp, GluGlu, and THG solutions exhibited nonuniform strains and crystallite sizes comparable with those of the control COM grown in distilled water. It is therefore unlikely that those molecules were occluded into the mineral. Crystals grown in solutions of Gla, Glu, and PT, as well as those generated from UF and CF urine, however, had notably higher nonuniform stains and lower crystallite sizes than the control. Because increased nonuniform strain and decreased crystallite size result from incorporated molecules or ions that increase the number of stacking faults and zones of disorder within crystals,(43) it can be concluded that Gla, Glu, PT, and endogenous urinary proteins and molecules were incorporated into COM during crystal growth. The increase in crystallite size and decrease in nonuniform strain observed after protease treatment of COM crystals grown in CF urine indicate that the smaller crystallite material, more intimately associated with proteins than with bulk COM, was liberated during the protease treatment.
A dominant factor that influences occlusion of amino acids into COM is their ability to bind calcium. Gla, which is well known for its strong affinity for calcium, is likely to attach irreversibly to the COM surface and be enveloped by a growing crystal front, whereas amino acids with lower binding affinity are less likely to become interred within the mineral bulk. This seems to be the case for Asp and Glu. Why Glu should create more crystal disorder than Asp, AspAsp, and GluGlu is not clear, but may be associated with the stereochemistry of their binding to calcium ions on the crystal face.(28) Stereochemistry may also be a determining factor in reducing the binding capacity of Asp and Glu dimers for COM, producing lower nonuniform strains and higher crystallite sizes than occurred with their corresponding monomers. The type and sequence of amino acids comprising the primary structure of proteins are of great importance in determining the nature of protein-crystal interactions.(25,50) The presence of PT within COM crystals can almost certainly be attributed to the strong calcium-binding properties of its Gla residues. However, affinity for calcium ions is insufficient to guarantee that a protein will be incarcerated within growing COM. THG has no Gla, but still has a considerable affinity for calcium,(47) and therefore might be expected to be incorporated into CaOx crystals. However, the protein is not occluded into the mineral,(12,51) probably because of its size. THG, which has a monomeric molecular mass of approximately 94 kDa, condenses in urine and at high concentrations in aqueous solutions into polymers with molecular masses in the millions. In contrast, OPN, another urinary protein thought to play a major role in stone formation and which also lacks Gla residues, is, nonetheless, an intracrystalline component of urinary CaOx crystals.(52) It has been generally assumed that the inhibitory effects of OPN on CaOx crystallization, as well as its association with CaOx crystals, can be attributed to its high content of Asp residues, which mediate its binding to the CaOx crystal surface. However, our demonstration that Asp is not an intracrystalline component of CaOx crystals suggests that OPN′s properties, as well as its presence in stones and CaOx crystals, are dictated by factors other than its component Asp moieties. Collectively, these facts show that the inclusion of an individual protein into the COM mineral bulk is not solely determined by its calcium-binding ability or by the presence of Gla within its primary chain.
The reciprocal relationship between nonuniform strain and crystallite size that we found for COM powders has been noted in single crystals studies of other systems, for example, NiO,(53) where nonuniform strain and the reciprocal of crystallite size were both shown to vary linearly with uniform strain, and Addadi et al.'s qualitative observation for calcite crystals in sponges and sea urchins.(22) In those cases, it is clear that a reduction in crystallite size increased misalignment and hence nonuniform strain. However, to our knowledge, our observation that the relationship differs greatly between amino acids and proteins is the first demonstration that, for a given increment in nonuniform strain, the corresponding decrease in crystallite size is markedly greater for macromolecules than for small molecules. We attribute the difference to the large disparity between the sizes and shapes of proteins and amino acids. Because of their small size, amino acids would be expected to inhibit growth only at, or very near to, the sites at which they bind to the crystal surface and therefore to cause only small local imperfections. On the other hand, a macromolecule such as a protein has the potential to interfere with the deposition of solute ions over a much larger region of the crystal surface. Delivery and attachment of solute ions to a crystal will be prevented at any point on the surface covered by the specific binding domain of any protein. However, such domains account for only a small proportion of a protein's structure, whose remaining molecular bulk also has the potential to hinder physically the transport of solute ions to the crystal surface. Crystals formed in the presence of macromolecules would therefore be more likely to incur larger disruptions to their structures than they would from the inclusion of smaller impurities such as amino acids. Consequently, they would exhibit more pronounced non-uniform strain, because greater crystallite packing disorder would occur with decreased crystallite size. Because of their large size, proteins are unlikely to intercalate into the COM lattice, but once adsorbed onto preferred crystal faces, they could be occluded into the structure as a consequence of engulfment by the advancing growth front.
The relationship between macromolecules and mineral that we have observed in urinary COM crystals can thus be considered analogous to that observed for calcite in some sponges and sea urchins.(24) The structures consist of morphologically identifiable single crystals containing a macromolecular core with which are associated small crystallites or amorphous material, on which is built a layer of larger and more ordered crystallites. The presence in urine of proteins with the capacity to become incorporated inside CaOx crystals is now viewed as having important implications for the development or prevention of urolithiasis.(4,16,54) Although much work has documented the possible role of various urinary macromolecules, particularly proteins,(55) which may inhibit the nucleation of CaOx crystals or interdict their subsequent growth or aggregation, the possible significance of intracrystalline proteins is only recently beginning to be appreciated. It is now widely acknowledged that attachment of newly formed crystals to the renal epithelium is a probable step in urolithiasis, but paradoxically, it could also constitute a natural, routine mechanism of avoiding stone formation. Dissolution of CaOx crystals phagocytosed by cultured renal cells(17–21) has led to the notion that intracellular destruction of crystals could act as a natural form of defense against urolithiasis,(21) a proposal that we have extended to include a specific role for intracrystalline macromolecules.(4,16) Proteins within crystals could offer a delayed, second line of defense if earlier protective mechanisms fail, because they could accelerate the dismantling and ultimate dissolution of their mineral hosts, effected by a combination of proteolytic digestion and the acidic interior of the phagolysosome. Their distribution throughout the structure would provide a series of channels through which cellular proteases could penetrate to the crystal core, thus increasing the area of exposed mineral surface to the acidic environment and increasing the rate of dissolution. This notion is supported by our observation of the presence of intracrystalline interconnecting furrows and subcrystalline particles in urinary CaOx crystals subjected to proteolytic digestion(4,16) and demonstration that protease inhibitors prevent the surface etching that occurs when urinary CaOx crystals are incubated in fresh, whole human urine.(54)
This work was supported by grants from the Australian Institute of Nuclear Science and Engineering, the Australian Synchrotron Research Grant Program, the National Health and Medical Research Council of Australia (Grant No. 980336), the Urological Foundation of Australia, the Flinders Medical Research Institute, and Flinders 2000. The assistance of Drs James Hester and Garry Foran with collection of the SXRD data and the advice of Dr Kevin Ho on proteolytic digestion are gratefully acknowledged.