ISSN 1063-7745, Crystallography Reports, 2017, Vol. 62, No. 1, pp. 148–156. © Pleiades Publishing, Inc., 2017. Original Russian Text © S.S. Baskakova, V.V. Volkov, T.V. Laptinskaya, M.S. Lyasnikova, A.E. Voloshin, M.V. Koval’chuk, 2017, published in Kristallografiya, 2017, Vol. 62, No. 1, pp. 148–157.
CRYSTAL GROWTH
Nature of Impurities during Protein Crystallization S. S. Baskakovaa,*, V. V. Volkova, T. V. Laptinskayab, M. S. Lyasnikovaa, A. E. Voloshina, and M. V. Koval’chuka,c aShubnikov
Institute of Crystallography of Federal Scientific Research Centre “Crystallography and Photonics”, Russian Academy of Sciences, Moscow, 119333 Russia b Moscow State University, Moscow, 119992 Russia cNational Research Centre “Kurchatov Institute,” Moscow, 123098 Russia *e-mail:
[email protected] Received January 21, 2016
Abstract—Lysozyme crystal growth was studied using reagents of different purity of three trademarks— Seikagaku Corporation (sixfold recrystallized lysozyme), Sigma-Aldrich (threefold recrystallized lysozyme), and Hampton Research (threefold recrystallized lysozyme). Solutions of these reagents were investigated by small-angle X-ray scattering, dynamic light scattering (DLS), ultracentrifugation, and electrophoresis. It was found that crystal-growth and oligomerization processes are more intense in solutions of the reagent of higher purity. The dependences of the fraction of lysozyme oligomers on the supersaturation and purity of the solution are analyzed. DOI: 10.1134/S1063774517010060
INTRODUCTION The incorporation of impurities into the growing crystal is one of the key factors that influence the crystal growth rate and the structural perfection [1–3]. One of the most commonly accepted hypotheses about protein crystals is that protein dimers are the main impurities in protein solutions [4]. These impurities have larger sizes and higher molecular weights compared to monomers. It was shown [5–7] that the presence of dimers decreases the rate of formation and movement of growth steps. This effect is similar to the influence of impurities on water-soluble inorganic crystals. Protein solutions were also studied during the crystallization by inelastic light scattering [8–11]. These studies revealed not only monomeric particles but also larger macromolecular aggregates. In order to validate the hypothesis of the crucial effect of protein dimers on the crystal growth, some research teams analyzed the content of high-molecular-weight impurities in solutions of hen egg-white lysozyme. Since the lysozyme structure was studied fairly well, this protein is suitable for investigations of the mechanisms and kinetics of the protein crystal growth [12, 13]. Lysozyme molecules in solution were studied for the first time by molecular light scattering in 1971 [14]. The translational and rotational diffusion coefficients were determined from the analysis of the spectrum of scattered laser light with a Fabry–Perot interferome-
ter. The experiment was performed using a solution of twofold recrystallized lysozyme in 0.1 M acetate buffer, pH 4.2. In the concentration range of 1–15%, the translational diffusion coefficient was constant and equal to (10.6 ± 0.1) × 10–7 cm2/s. The rotational diffusion coefficient was (16.7 ± 0.8) × 106 s–1 at a concentration of 15%. Consequently, the molecule acquires a hydrodynamic ellipsoidal shape with the major axis of 55 ± 1 Å, two other axes being equal to (33 ± 1) Å. Based on the comparison of these data with the X-ray diffraction data, it was concluded that the molecule is hydrated. The hydration layer thickness is 3.5 Å. In later years, highly stable single-mode lasers have become available, high-speed correlators were developed, and portable computers were designed. This made it possible to analyze scattered light from the correlation function of the signal rather than by measuring the signal at the shifted frequency relative to the laser frequency, as described in [14]. The 1990s witnessed a sharp increase in the number of publications, in which lysozyme solutions were studied by dynamic light scattering (DLS, also known as photon correlation spectroscopy or quasi-elastic light scattering) [15–19]. This method allows the determination of diffusion coefficients of particles in a solvent, from which the correlation lengths or hydrodynamic radii can be extracted. There are currently diverse modifications of light scattering devices, and these devices become available to many laboratories. Therefore, the behavior of lysozyme in different solutions was studied by many
148
NATURE OF IMPURITIES DURING PROTEIN CRYSTALLIZATION
research teams [20]. In all cases, new results were compared with those reported in [14]. However, these results were sometimes inconsistent with those obtained in 1971. First, the difference is that the radius of the smallest particles greatly depends on the salt or a combination of salts added to water before the dissolution. Another difference is that clusters of particles are formed in solution in due course under certain conditions. In [15–19] it was found that clusters grow to form a fractal structure with a fractal dimension of about 1.8. The kinetics of the cluster growth is determined by either the diffusion of particles or the probability of their collisions. In the former case, large crystals can be obtained, whereas small crystals grow in the latter case. At low growth temperatures, a gelatinous precipitate is formed. The computer modeling for 19 globular proteins was performed [21] in order to determine a number of parameters. For the lysozyme monomer, the hydrodynamic radius Rh was evaluated at 2.7 nm; the hydration-shell thickness was 0.9 nm. In [22] a series of solutions of lysozyme purchased from Sigma Aldrich in HEPES buffer, pH 7.8, at different concentrations were studied at 20°С by smallangle X-ray scattering (SAXS), small-angle neutron scattering (SANS), and confocal microscopy. Equilibrium clusters with sizes in the range from 5.1 to 8.5 nm (with a volume fraction of the protein solution 0.024 ≤ φ ≤ 0.15) were found. The association of lysozyme dimers in solution was studied by DLS [23]. In a solution containing only dimers (the refinement was performed by means of chromatography), particles with a hydrodynamic radius Rh = 2.37 nm were found in the presence of NaCl as the precipitant and particles with Rh = 2.53 nm were found in the absence of NaCl. In 2007 Shukla et al. published the results [24] that refute the data reported in [22]. This research team repeated investigations of lysozyme solutions by smallangle X-ray and neutron scattering and also performed a series of other studies. In the cited study, only particles with a size corresponding to the lysozyme monomer were found. Lysozyme solutions with a protein volume fraction φ of 0.012, 0.033, and 0.12 were studied by SAXS [25]. At φ ≤ 0.012 the solution was found to contain about 70% of monomers, whereas the remaining 30% fraction accounts for small equilibrium clusters, predominantly dimers and trimers. The formation of tetramers and pentamers was also not excluded. At φ = 0.033, 80% of the solution contained clusters, whereas at φ = 0.12 the solution was composed mainly of larger clusters. Therefore, the results of the above-considered studies are quite contradictory. It is noteworthy that different research teams studied protein solutions from different manufacturers and of different compositions, which were prepared under different conditions (conCRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
149
centration, temperature, buffer solution, etc.). The influence of the conditions on the crystal growth was not assessed. Therefore, it is impossible to estimate the true role of dimers (or other impurities) in the growth and defect formation processes. The aim of the present work is to study the formation of lysozyme oligomers depending on the supersaturation and the purity of the solution and the influence of the oligomers on the crystal growth. MATERIALS AND METHODS Experiments were performed using solutions of lysozyme of the following three trademarks: dry sixfold recrystallized lysozyme purchased from Seikagaku Corporation (hereinafter, lys6), the threefold recrystallized protein from Sigma-Aldrich (hereinafter, lys3SA), and the threefold recrystallized protein from Hampton Research (hereinafter, lys3HR). Experiments with different reagents were carried out under the same conditions. Lysozyme was placed in a sodium acetate buffer solution. The same buffer solutions were used to prepare solutions of the precipitant NaCl. All solutions were filtered using Whatman membrane filters with a pore size of 0.45 μm in order to additionally purify the solutions from finely dispersed particles and impurities. The crystal growth was studied using the abovementioned protein reagents at a supersaturation of 2.24 in a 0.5 M buffer solution, pH 4.85, and at a supersaturation of 0.9 in a 0.1 M buffer solution, pH 4.6, at 21°C. The concentration of the precipitant (NaCl) was 25 mg/mL. The particle sizes were measured by DLS on an ALV-CGS-5000/60X system (Langen, Germany). A helium-neon laser (wavelength of 632.8 nm, output power of 20 mW) was used as the light source. For each sample, light intensity correlation functions were collected (the ensemble-averaged product of the intensity of a signal accumulated at delay time and the intensity of the newly accumulated signal normalized to the average intensity of the signal). The mathematical processing of the experimental results was performed by either the CONTIN method or the cumulant expansion method. The former method is based on the solution of Laplace’s equation using Tikhonov’s regularization. This method reconstructs the relaxation time distribution, but it gives a large error in the determination of the line width about each maximum. The latter method is applied to solutions, in which it is necessary to extract the main relaxation time and more precisely determine the distribution width. The relaxation time of the correlation function τ is related to the diffusion coefficient D of the particle by the equation 1/τ = Γ = Dq2 [26]. Here q = (4π/λ)nsin(θ/2) is the scattering wave vector, λ = 632.8 nm is the wavelength of helium-neon laser light,
150
BASKAKOVA et al.
n is the refractive index of the solvent, and θ is the scattering angle. The scattering was observed at an angle of 90°, which corresponds to q = 0.00462 nm–1 (the refractive indices of water and the buffer solutions are 1.33 at room temperature). Then the particle radii were found from the dependence of the diffusion coefficient on the radius of the moving particle described by the equation D = kT/6πηRH. The radius RH is called the hydrodynamic radius and is determined as the radius of a spherical particle, which, under the same conditions, has the same diffusion coefficient as the particle under consideration. In order to determine the particle radii, the viscosity η of the sodium acetate buffer solution was measured using an Ubbelohde capillary viscometer. At 23°C, η was 1.027 and 0.950 mPa s at a concentration of 0.5 and 0.1 M, respectively. At 8°С, η was 1.58 mPa s for the concentration of 0.5 M. The experiments were performed in 10-mm cylindrical glass cells. The solutions used for the determination of the dependence of the hydrodynamic radius on the lysozyme concentration were prepared as follows. The stock solutions of lysozyme of different trademarks at a concentration of 60 mg/mL were prepared in a sodium acetate buffer solutions supplemented with NaCl as the precipitant. The solutions were filtered so that large particles were not detected by DLS within the first 30 min. Then solutions with lower concentrations were prepared using the stock solutions. The latter were divided into approximately two aliquots. One aliquot was diluted with a sodium acetate buffer solution; another, with a NaCl solution. Then these solutions were combined in such a way as to obtain a solution with the required concentration. Both dilution solutions were prefiltered through a 0.22-μm pore filter membrane. The resulting new sample was not filtered. The SAXS measurements were performed on the AMUR-K automatic small-angle X-ray scattering diffractometer equipped with a Kratky collimation system and an OD3M one-coordinate position-sensitive detector at a fixed wavelength λ = 0.1542 nm (CuKα line of a sharp-focus tube, pyrolytic graphite monochromator). The X-ray beam cross-section was 0.2 × 8 mm. The scattering-angle range corresponded to the range of the magnitude of the scattering vector 0.1 < s < 5.0 nm–1 ( s = 4π sin θ, 2θ is the scattering angle). λ Samples were placed in quartz capillaries with a diameter of about 1 mm and a wall thickness of 0.01 mm. The measurement time for one sample was 60 min. The experimental data were normalized to the incident-beam intensity, and then a correction for collimation error was applied. The scattering from the capillary containing the buffer solution was subtracted from the sample scattering intensity. The results of measurements were not converted to the absolute intensity scale, because the structural parameters to be
evaluated are independent of the intensity scale. The measurements were performed according to a certified procedure approved for the AMUR-K diffractometer. This diffractometer is a part of facilities of the Specialized Joint Use Center “Structural Diagnostics of Materials” at the Shubnikov Institute of Crystallography of the Russian Academy of Sciences. Synchrotron radiation (SR) experiments were performed at the Kurchatov Centre for Synchrotron Radiation and Nanotechnology at the DIKSI smallangle X-ray scattering station equipped with a Kratky collimation system and a MAR CCD SX165 digital area detector at a fixed wavelength λ = 0.1625 nm. The X-ray beam cross-section was 0.4 × 1 mm. The scattering-angle range corresponded to the range of the magnitude of the scattering vector 0.35 < s < 6.0 nm–1. The samples were placed in quartz capillaries with a diameter of about 1.5 mm and a wall thickness of 0.01 mm. The measurement time for one sample was 20 min. The experimental data were normalized taking into account a decrease in the incident-beam intensity due to a decay of the storage-ring current. The scattering from the capillary containing the buffer solution (in this case, the solution without a precipitant) was subtracted from the sample scattering intensity. The results of measurements were not converted to the absolute intensity scale because the structural parameters to be evaluated are independent of the intensity scale. The maximum size and the radius of gyration were estimated from the pair-distance distribution function p(r), which was calculated by the indirect Fouriertransform method using the GNOM program according to the equation ∞
p(r ) =
∫ I (s)
s =0
sin ( sr ) ds , sr
where I(s) is the scattering intensity in arbitrary units. The radii of gyration Rg of the molecules were calculated from the SAXS data in the Guinier approximation I exp ( s ) = I ( 0) exp –s 2R g2 /3 , which is valid, with sufficient accuracy, in the range (sRg) < 1.3, and were also determined from the distance distribution function p(r) according to the following equation
(
)
∫ p(r )r dr , = ∫ p(r )dr 2
R g2
by a procedure described in GOST R 8.698–2010 (“State system for ensuring the uniformity of measurements. Dimensional parameters of nanoparticles and thin films. Method for measurement by means of a small angle X-ray scattering diffractometer.”). The function p(r) was used also for the determination of the maximum sizes Dmax of the dissolved molecules
CRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
NATURE OF IMPURITIES DURING PROTEIN CRYSTALLIZATION
EXPERIMENTAL RESULTS Crystal Growth The crystallization conditions were those used for the crystallization of tetragonal lysozyme [27]. Crystal growth experiments with lysozyme of different purity demonstrated that, under the same crystallization conditions, protein crystals appeared in solutions of lysozyme from Seikagaku Corporation within one day after the onset of crystallization, and they grew to a maximum size of ~2 mm. By contrast, crystals did not grow at all or a large amount of small crystallites precipitated in solutions of lysozyme from Sigma-Aldrich and Hampton Research.
logI, arb. units 1 1 0 2 1 3 2
3 0.5
1.0
151
1.5
2.0
2.5
3.0 3.5 s, nm1
Fig. 1. Small-angle X-ray scattering curves (noisy curves) and their comparison with the theoretical scattering intensity (smooth curves) calculated from a monomer + dimer mixture: (1) lys3SA, (2) lys6, (3) lys3HR. Pairs of the curves (experiment + model) are shifted vertically for clarity.
from the maximum distance corresponding to a nonzero value of the distribution. The sedimentation experiments were performed at 20°С using a Spinco analytical ultracentrifuge equipped with an absorption optical scanning system, a monochromator, and a computer. Experiments were carried out using a four-place An-F titanium rotor at 60 000 rpm and double-sector cells. The sedimentation was recorded by measuring the light absorption at a wavelength of 280 nm. All cells were scanned simultaneously. The results were processed using the SEDFIT program. Electrophoresis of the protein was performed under non-reducing and reducing (in the presence of beta-mercaptoethanol) conditions at a separation-gel concentration of 12%. The protein samples were prepared as follows. Tris-glycine buffer, pH 6.8, containing the dye (bromophenol blue) and 1% SDS was added to a protein solution (10 μL). The mixture was refluxed in a water bath for 5 min. The electrophoresis was carried out at room temperature using a Mini PROTEAN Tetra cell. The gels were fixed in a solution containing 10% acetic acid and 25% ethanol for 20 min, stained with 0.15% Coomassie G-250 (Sigma, USA) in a solution containing 10% acetic acid and 30% ethanol for 60 min at room temperature, and washed with a solution containing 15% ethanol and 5% acetic acid. CRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
Studies of Solutions with a Precipitant Small-angle X-ray Scattering Measurements were performed on the AMUR-K laboratory diffractometer using solutions containing the protein at the initial concentration of 25 mg/mL and the precipitant at the concentration of 25 mg/mL in 0.1 M sodium acetate buffer. The radius of gyration determined from the results of measurements was 1.59 ± 0.02 nm for lys3SA, 1.55 ± 0.02 nm for lys3HR, and 1.53 ± 0.02 nm for lys6. These values are similar to the radius of gyration determined for the atomic structure 6LYZ (1.55 ± 0.02 nm). However, the small-angle X-ray scattering curves (Fig. 1) show an increase in the intensity, which is indicative of aggregation of the protein molecules in solutions. Figure 2 presents the pair-distance distribution function p(r). The maximum size of the molecules estimated from the distributions p(r) is 5.5–6 nm. The diameter of the molecule evaluated from the atomic structure 6LYZ extracted from the Protein Data Bank is 4.8 nm taking into account the hydration-shell thickness for the particles in solution (0.25–0.3 nm) increased by a factor of two. The larger experimental value of the maximum particle diameter attests to the aggregation of particles in solution. However, since an increase in the size is insignificant, these are not large particles. Apparently, these aggregates are dimers. The fraction of dimeric molecules in solutions was estimated using the OLIGOMER [28] and CRYSOL programs [29]. In order to evaluate the fraction of molecular associates, the theoretical SAXS intensity curves (Fig. 1) were calculated with the CRYSOL program using the atomic structure 6LYZ from the Protein Data Bank and the structure, which was modeled using the MASSHA program [30] from two monomers of the structure 6LYZ that were arranged as close to each other as possible. The complex was constructed without the application of molecular dynamics methods taking into account only steric restrictions. The three-dimensional structure of 3VFX from the Protein Data Bank was used as an alternative structure of the dimer. The model scattering curves, which were calculated only for the monomers, are
152
BASKAKOVA et al.
p(r), 10‒2 arb. units 1 4 1 nm
(a)
3 2 2
1 3 0
0
1
2
3
4
5
6
7 r, nm
(b)
Fig. 2. Comparison of the pair-distance distribution functions calculated from the SAXS data: (1) lys3SA, (2) lys6, (3) lys3HR.
visually almost not distinguishable from the monomer + dimer model. Nevertheless, the inclusion of dimers in the model led to the improvement of the goodness-of-fit defined by the chi-squared statistic χ2. Therefore, Fig. 1 presents only the results of modeling with monomer + dimer mixtures. The OLIGOMER program utilizes theoretical scattering intensities to construct a matrix of a system of linear equations with unknown coefficients for the components (monomer, dimer, tetramer, and octamer). The right-hand side of the equations included experimental scattering data with a correction for collimation error. The calculations using the structure 3VFX demonstrated that the results of measurements for the lys3HR sample are consistent with the description of the scattering from a solution containing the monomer (95.6 ± 1.7%) and the dimer (4.4 ± 0.6%) with χ2 = 1.05. If the scattering from the dimer is ignored, χ2 = 1.1. This is indicative of an insignificant fraction of dimers in the solution. A somewhat different pattern was observed for a lys6 solution. Thus, the fraction of dimeric molecules is approximately two times larger (7.5 ± 0.4%, χ2 = 1.15), which is reflected in the poorer goodness-of-fit in the case when dimers are ignored (χ2 = 1.44). For the lys3SA sample, the calculated fraction of dimeric molecules is 1.6 ± 0.7% (χ2 = 1.3). Due to the negligibly small fraction of dimers, the fitting without taking the latter into account gave almost the same value of χ2. The shape of lysozyme molecules was reconstructed from the small-angle scattering data with the DAMMIN program [31]. In the case of protein solu-
(c)
(d)
Fig. 3. Comparison of the results of the dummy atom modeling of the macromolecular shape based on the SAXS data with the structure of the molecule extracted from the Protein Data Bank: (а) 6LYZ (PDB), (b) lys3SA, (c) lys3HR, (d) lys6.
tions containing the precipitant (Fig. 3), the particle shape for the lys3HR and lys6 samples corresponded to an associate consisting of two lysozyme molecules. The lower fragments of the structures shown in Figs. 3c and 3d increase the sizes of the molecules and are associated with that the solution contains dimers as an impurity, resulting in the virtual increase in the linear size of the models found on the assumption that the solution contains identical particles. The smaller size of the added particle is attributed to the fact that the modeling was performed based on the total scattering data, which account for the presence of both the dimers and monomers in the solution. The model presented in Fig. 3b almost exactly corresponds to the shape of the molecule 6LYZ and does not contain an added particle due to an insignificant fraction of dimers in a lys3SA solution. Analytical Ultracentrifugation The presence of oligomers in lysozyme solutions was analyzed by analytical ultracentrifugation. The particle weights were determined using solutions containing the smallest and largest fractions of dimers (lys3SA and lys6) according to estimates by SAXS. The protein and precipitant concentrations in 0.1 M sodium acetate buffer were 8 and 5 mg/mL, respectively. It was found that approximately 97.4% of the particles in the lys6 solution have the molecular weight of 17.3 kDa and about 2.6% of the particles have the
CRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
NATURE OF IMPURITIES DURING PROTEIN CRYSTALLIZATION
RH, nm 3.0
153
G21 100
1 2 3 4 5 6
2.8 2.6
101
2.4
4 6
10
3
1, 2
2.2 20
30
40
102
50 60 С, g/L
103
molecular weight of 53.9 kDa. In the lys3SA solution, approximately 98.3% of the particles have the molecular weight of 15.8 kDa, which is close to the weight of the lysozyme monomer (the molecular weight of the protein monomer is 14.3 kDa), and approximately 1.7% of the particles have the molecular weight of 55.8 kDa, which is approximately 3–4 times larger than the weight of the lysozyme monomer. Dynamic Light Scattering Since the SAXS experiments and analytical ultracentrifugation were performed in solutions with different concentrations and showed that solutions contained different fractions of high-molecular-weight complexes, the solutions were studied by DLS at different supersaturation levels. A series of lys3SA and lys6 solutions was studied at different concentrations. The initial solution was composed of 0.1 M sodium acetate buffer supplemented with the protein at a concentration of 60 mg/mL and the precipitant at a concentration of 25 mg/mL. To perform measurements at lower concentrations, the initial solutions were diluted. The DLS method is not suitable for determining the fraction of oligomers in solution because the mathematical processing is unable to distinguish similar values of the hydrodynamic radii. Experiments using size-calibrated latex spheres demonstrated that the CONTIN method reliably resolves two sizes in the ensemble if these sizes differ by at least ten times [32]. In the present study, the average value was determined from the particle ensemble. It was slightly overestimated due to the cubic dependence of the contribution to the scattering intensity on the radius. As a rule, the samples under study contained, apart from the main particles (the maximum at hydrodynamic radii from 1.9 to 3 nm), a small fraction (~0.001%) of particles with sizes of 200–400 nm and Vol. 62
No. 1
2 1
Fig. 4. Plots of the hydrodynamic radius RH of molecular associates of lysozyme lys3SA (asterisks) and lysozyme lys6 (solid squares) versus the protein concentration C in solution.
CRYSTALLOGRAPHY REPORTS
5
2017
102
101
100
101
102
103 t, ms
Fig. 5. Scattering intensity correlation functions (G2) for a solution of lysozyme lys6 at a concentration of 60 mg/mL in 0.5 M sodium acetate buffer containing the precipitant at different temperatures: (1) the initial solution at 23°C, (2) at 17°C, (3) at 9°C, (4) at 8°C, (5) the solution was kept at 8°C for one hour, (6) at 23°C after 20 h; t (ms) is the delay between two multiplied signals.
1–1.5 μm (these particles are referred to as clusters in [15–20]). A virtually linear increase in the measured particle radius from 2.2 to 2.9 nm was observed with an increase in the protein concentration (Fig. 4). The measurement accuracy is approximately ±0.1 nm. As can be seen in Fig. 4, the particle radius in the lys3SA solutions was always larger than the particle radius in the lys6 solutions, except for the points corresponding to the concentrations of 15 and 60 mg/mL. However, since this difference is within experimental error, it is impossible to draw an unambiguous conclusion. Changes in the structure of the solution associated with changes in the temperature were studied using a 0.5 M sodium acetate buffer solution containing the protein at the initial concentration of 60 mg/mL and the precipitant at a concentration of 25 mg/mL. The correlation function plots are presented in a doublylogarithmic scale (Fig. 5) in order to separate the contributions of large relaxation times. The initial solution of lysozyme lys6 contained particles with a radius of 2.6 nm and a small fraction of large particles. The number of latter slightly decreased with a decrease in the temperature to 17°С due apparently to the precipitation of the largest particles (oligomers). The maximum in the plot was shifted to RH = 2.4 nm. Curve 3 (Fig. 5) was obtained at 9°С after 30 min (RH = 2.4 nm); curve 4, at 8°С one hour after the measurement of curve 3 (RH = 2.9 nm). During this period of time, the solution was unstable. The crystallites that precipitated were clearly visible under
154
BASKAKOVA et al.
logI 3.0
p, 102 arb. units 2
5 1 2.5
4
1 2
3
2.0 2 1.5
1 1
2
3
4 s2, nm2
0
Fig. 6. Scattering curves plotted as the logarithm of intensity versus the square modulus of the scattering vector s. (1) lys3HR, (2) lys6. The straight lines represent approximation lines for the determination of the radius of gyration Rg.
1
2
3
4
5 s, nm
Fig. 7. Pair-distance distribution functions for two protein samples calculated from the SAXS data: (1) lys3HR, (2) lys6.
a laser beam. After the storage of the solution for more than one hour, the precipitation of crystallites stopped (curve 5). The supernatant contained particles with a radius of 3.1 nm and aggregates with sizes in the range of several hundred nanometers. Then the thermostat was switched off, after which the temperature spontaneously increased to 23°С. After 20 h the particle radius in the equilibrium supernatant (curve 6) was restored to the value of 2.6 nm. Crystallites were observed at the bottom and on the walls of the cell. Studies of Solutions without a Precipitant Small-angle X-ray Scattering Solutions of lys6 and lys3HR without a precipitant were studied using synchrotron radiation at the DIKSI station (Kurchatov Centre for Synchrotron Radiation and Nanotechnology). The solutions were composed of 0.1 M sodium acetate buffer containing the protein at a concentration of 5 mg/mL. The radius of gyration determined from the SAXS data was 1.46 nm for lys3HR and 1.49 nm for lys6. These values are almost equal to the radius of the monomer. As can be seen in Fig. 6, there is a slight decrease in the intensity in the small-angle region due to interparticle interference. Figure 7 presents a pair-distance function. The maximum size of the molecules evaluated from the pair-distance distribution function is 4.1–4.4 nm. This value is smaller compared to the solutions containing the precipitant. In the case of solutions without a precipitant (Fig. 8), there is virtually no difference between the structure extracted from the Protein Data Bank and the models constructed based on the scattering data. These structures have similar sizes and shapes.
Electrophoresis The SAXS data for lysozyme solutions without a precipitant were validated by electrophoresis (Fig. 9). Solutions of lysozyme of two trademarks (lys3HR and lys6) were studied in 0.1 M sodium acetate buffer containing the protein at a concentration of 2 mg/mL without a precipitant. Only particles with a weight of about 14.5 kDa, which is almost equal to the weight of the lysozyme monomer, were found in the solutions. No particles with other weights were observed.
1
2
3 Fig. 8. Comparison of the results of the modeling based on the SAXS data obtained on the DIKSI station with the structure extracted from the Protein Data Bank (PDB): (1) the structure from the PDB, (2) the structure of lys3HR, (3) the structure of lys6.
CRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
NATURE OF IMPURITIES DURING PROTEIN CRYSTALLIZATION
116 kDa 66 45 35
25
18.5
1
2
3
4
14.5
Fig. 9. Investigation of lysozyme solutions by electrophoresis under (1) and (2) reducing conditions and (3) and (4) non-reducing conditions: (1) and (3) lys3HR, (2) and (4) lys6.
RESULTS AND DISCUSSION The fact that either no crystallization was observed or the bulk crystallization occurred in solutions of the threefold recrystallized protein samples purchased from Sigma-Aldrich and Hampton Research is apparently attributed to the presence of impurities, which interfere with the crystal growth and the content of which is higher in these samples compared to the sixfold recrystallized protein from Seikagaku Corporation. The SAXS data, as well as the results of analytical ultracentrifugation, indicate that the maximum concentration of oligomers (dimers and, probably trimers and/or tetramers) is present in a lys6 solution containing the precipitant. This is unambiguous evidence that it is not the оligomers of the protein molecules that interfere with the crystal growth. Studies of the particles with medium sizes depending on the protein concentration (in fact, on the supersaturation level) by DLS demonstrated that the degree of aggregation of the molecules increases with an increase in the supersaturation level, which is a natural consequence of the precrystallization state of the solution. This is also confirmed by the SAXS measurements and the results of electrophoresis for solutions containing the protein at a low concentration without a precipitant (in an unsaturated state). In this case, the sizes and weights of the detected particles are almost equal to those of the lysozyme monomer and, consequently, the formation of oligomers does not occur. The most remarkable are DLS experiments with supercooled solutions. A rapid decrease in the temperature from 23 to 9°C led to a decrease in the average particle radius in the supernatant from 2.6 to 2.4 nm accompanied by the crystal formation. Apparently, larger particles (with a size of about 3–4 nm) observed CRYSTALLOGRAPHY REPORTS
Vol. 62
No. 1
2017
155
in the supercooled solution reached the critical nucleus size and served as crystallization sites. After the storage of the solution at a constant temperature of 8°С, when the precipitation of crystallites stopped, the particle size increased from 2.9 to 3.1 nm within one hour. Then, as the mixture was allowed to slowly warm to room temperature, this value was restored back to 2.6 nm. Consequently, the oligomer formation is a reversible process and reaches the thermodynamic equilibrium at a constant temperature, the fraction of oligomers in the solution being increased with an increase in the degree of supercooling. The SAXS measurements of lysozyme solutions at a concentration of 40 mg/mL in the presence and in the absence of a precipitant also demonstrated [33] that oligomers are not formed in a unsaturated solution (without of a precipitant). Dimers and octamers were found in a solution containing the precipitant. The presence of octamers is apparently associated with a higher protein concentration in the solution under study because the presence of octamers was not observed in either SAXS or sedimentation experiments. It should be noted that it is the solutions of the lys6 of higher purity that ensure the best crystal growth; however, these solutions also contain the largest fraction of oligomers. This suggests that the formation of protein crystals and associates are processes, which are related to each other and are influenced by impurities in a similar way. Apparently, this effect is that impurity particles are added to protein molecules, thus changing the surface charge distribution and, consequently, the conditions of the complex formation with other molecules. Since the solutions of the threefold and sixfold recrystallized reagents contained substantially different fractions of associates, the amounts of the protein and impurity molecules in solutions are comparable in the order of magnitude. Since the weight fraction of impurities in the reagents under study is not higher than 0.1%, the molecular weight of impurities should be several ten Da. For example, amino acids produced by denaturation of the protein molecules can be such impurities. Therefore, oligomers by themselves do not influence the protein crystal growth, at least at concentrations examined in the present study. Due to their small fractions in solution, oligomers cannot serve as the main building units in the crystal growth. However, they can be rather readily incorporated into the crystal lattice with almost no elastic deformation. This is evidenced by a large value of the equilibrium distribution coefficient of dimers in lysozyme estimated at 9 in [7]. CONCLUSIONS Large protein crystals were grown using sixfold recrystallized lysozyme from Seikagaku Corporation, whereas either large amounts of small crystals were
156
BASKAKOVA et al.
obtained or crystals did not precipitate at all in experiments performed under the same conditions using threefold recrystallized lysozyme samples from Sigma-Aldrich and Hampton Research. Lysozyme solutions were studied by dynamic light scattering, small-angle X-ray scattering, ultracentrifugation, and electrophoresis in order to find whether these solutions contain dimers. The latter, according to the theory reported in [4], have an effect on the crystal formation and the crystal growth rate. It was demonstrated that protein solutions of both types contain a few percentage of high-molecularweight particles. After the addition of the precipitant, the solutions of the sixfold recrystallized protein contain a larger fraction of high-molecular-weight particles compared to solutions of the threefold recrystallized protein. Lysozyme solutions without a precipitant do not contain oligomers. Apparently, the formation of protein crystals and associates are processes, which related to each other and are influenced by impurities in a similar way. Denaturation products of the protein molecules can be such impurities. ACKNOWLEDGMENTS The study was supported by the Russian Foundation for Basic Research, grant no. 13-02-12163-ofi-m. This study was performed using the equipment of the Joint Use Center at the Shubnikov Institute of Crystallography of Federal Scientific Research Centre “Crystallography and Photonics” of the Russian Academy of Sciences with the financial support of the Ministry of Education and Science of the Russian Federation. REFERENCES 1. L. A. Monaco and F. Rosenberger, J. Cryst. Growth 129, 465 (1993). 2. A. J. Malkin, Yu. G. Kuznetsov, and A. McPherson, Surf. Sci. 393, 95 (1997). 3. A. McPherson, Crystallization of Biological Macromolecules (Cold Spring Harbor Laboratory Press). 4. B. R. Thomas, P. G. Vekilov, and F. Rosenberger, Acta Crystallogr. 52, 776 (1996). 5. T. Nakada, G. Sazaki, S. Miyashita, et al., J. Cryst. Growth 196, 503 (1999). 6. I. Yoshizaki, A. Kadowaki, Y. Iimura, et al., J. Synchrotron Radiat. 11, 30 (2004). 7. D. C. Carter, K. Limit, J. X. Ho, et al., J. Cryst. Growth 196, 623 (1999).
8. P. G. Vekilov, M. Ataka, and T. Katsuura, J. Cryst. Growth 130, 317 (1993). 9. Z. Kam, H. B. Shore, and G. Feher, J. Mol. Biol. 123, 539 (1978). 10. Y. Georgalis and W. Saenger, Adv. Colloid Interface Sci. 46, 165 (1993). 11. A. Malkin and A. McPherson, J. Cryst. Growth 128, 1232 (1993). 12. C. C. F. Blake, L. N. Johnson, G. A. Mair, et al., Proc. R. Soc. London B 167, 378 (1967). 13. C. C. F. Blake, D. F. Koenig, G. A. Mair, et al., Nature 206, 757 (1965). 14. S. B. Dubin, N. A. Clark, and G. B. Benedek, J. Chem. Phys. 54, 5158 (1971). 15. W. Ebershtein, Y. Georgalis, and W. Saenger, Eur. Biphys. J. 22, 359 (1993). 16. M. Mushol and F. Rosenberg, J. Chem. Phys. 103, 10424 (1995). 17. G. Chirico, S. Beretta, and G. Baldini, J. Chem. Phys. 110, 2297 (1999). 18. S. Tanaka, K. Ito, R. Hayakawa, and M. Ataka, J. Chem. Phys. 111, 10330 (1999). 19. Y. Georgalis, P. Umbach, and W. Saenger, J. Am. Chem. Soc. 121, 1627 (1999). 20. T. Wakamatsu, Am. J. Anal. Chem. 5, 581 (2014). 21. J. Garcıa de la Torre, Biophys. Chem. 93, 159 (2001). 22. A. Stradner, H. Sedgwick, F. Cardinaux, et al., Lett. Nature 432, 492 (2004). 23. K. Onuma and K. Inaka, J. Cryst. Growth 310, 1174 (2008). 24. A. Shukla, E. Mylonas, E. Di Cola, et al., Proc. Natl. Acad. Sci. USA 105, 5075 (2008). 25. P. Kowalczyk, A. Ciach, P. A. Gauden, and A. P. Terzyk, J. Colloid Interface Sci. 363, 579 (2011). 26. W. Brown, Dynamic Light Scattering: the Method and Some Applications (Clarendon, Oxford, 1993). 27. A. V. Svanidze, S. G. Lushnikov, and L. A. Shuvalov, Crystallogr. Rep. 50 (5), 789 (2005). 28. P. V. Konarev, V. V. Volkov, A. V. Sokolova, et al., J. Appl. Crystallogr. 36, 1277 (2003). 29. D. Svergun, C. Barberato, and M. H. J. Koch, J. Appl. Crystallogr. 28, 768 (1995). 30. P. V. Konarev, M. V. Petoukhov, and D. I. Svergun, J. Appl. Crystallogr. 34, 527 (2001). 31. D. I. Svergun, Biophys. J. 76, 2879 (1999). 32. M. Shibayama, T. Karino, and S. Okabe, Polymer 47, 6446 (2006). 33. M. A. Marchenkova, V. V. Volkov, A. E. Blagov, et al., Crystallogr. Rep. 61 (1), 5 (2016).
CRYSTALLOGRAPHY REPORTS
Translated by T. Safonova
Vol. 62
No. 1
2017
