Packing Space Expansion of Protein Crystallization Screening with

Dec 1, 2010 - ABSTRACT: The applicability of the heteroepitaxic nucleant method using molecular sieves (MS) to sparse matrix screening of...
0 downloads 0 Views 7MB Size
Download PDF
DOI: 10.1021/cg100987g

2011, Vol. 11 110–120

Packing Space Expansion of Protein Crystallization Screening with Synthetic Zeolite as a Heteroepitaxic Nucleant Michihiro Sugahara, Yuko Kageyama-Morikawa, and Naoki Kunishima* Protein Crystallography Research Group, RIKEN SPring-8 Center, Harima Institute, 1-1-1 Kouto, Sayo-cho, Sayo-gun, Hyogo 679-5148, Japan Received July 26, 2010; Revised Manuscript Received October 30, 2010

ABSTRACT: The applicability of the heteroepitaxic nucleant method using molecular sieves (MS) to sparse matrix screening of protein crystallization was examined using xylanase as a test protein. The sparse matrix screening with MS provided three crystal forms I-III under 80% of conditions, which confirmed a much higher yield of crystal formation when compared to the conventional screening without MS. In the form I crystals, an MS-dependent improvement in X-ray diffraction quality was observed. Furthermore, form III of xylanase crystals was obtained only in the presence of MS 5A or 13X, demonstrating an MS-dependent packing-space expansion of protein crystals. 1. Introduction In current protein crystallography, most of the crystallization screening is performed according to the sparse matrix method in which a selected set of crystallization conditions that have a high probability to yield crystals is applied to all protein samples.1 To facilitate the process of the sparse matrix screening, a number of automated protein crystallization facilities have been developed by structural genomics initiatives,2-14 considering the minimization in sample amounts, the reduction of crystallization setup time, and the increment in trial numbers. However, currently there is no useful option when the sparse matrix screening turns out to be unsuccessful, indicating that a structure determination of proteins with poor crystallizability still remains a major problem in protein crystallography. Thus, in order to facilitate structural studies of the recalcitrant proteins, certain effective techniques to enhance the success rate of the sparse matrix crystallization screening are needed.15-17 In response to the demand, a number of protein crystallization methods using nonspecific heterogeneous nucleants have been reported to date: porous silicone,18 mesoporous bioactive gel-glass,19 human hair,20 nucleant cocktail (dried seaweed, horse hair, cellulose, and hydroxyapatite),21 polystyrene,22 and carbon-nanotube-based materials.23 These methods facilitate the crystallization through a nonspecific attraction of protein molecules on the material surfaces. Unfortunately, the nonspecific heterogeneous nucleants do not necessarily improve the diffraction quality of protein crystals because they only elevate the local concentration of proteins thereby producing the same forms of crystals that are also available without nucleants. The diffraction quality of protein crystals generally depends upon the molecular property of protein used. Thus, if the protein sample itself is inherently unstable, a restraint of *Corresponding author address. Tel: þ81-791-58-2937. Fax: þ81-791-582917. E-mail: [email protected]. pubs.acs.org/crystal

Published on Web 12/01/2010

conformational flexibility using various methods (removing flexible parts, introducing mutations, adding ligands, making complexes with counterpart proteins, and so on) may improve the crystal quality. However, even perfectly stable and “firm” protein cannot provide a high quality of crystal if its crystal packing is not suitable to yield the atomic level of structural regularity. In such cases, one of the most effective ways to improve the diffraction quality is searching for new crystal forms. The crystal polymorphism depends upon the variety in packing modes of protein molecules in crystal states. Therefore, certain methodologies to expand the crystal packing space would enhance the success rate of the sparse matrix screening effectively. The packing-space expansion of protein crystals can be achieved by applying heteroepitaxic nucleants that tend to provide new crystal forms through a heterogeneous and directed nucleation. The heteroepitaxic nucleants produce threedimensional (3D) crystals of a target protein from a stacking of two-dimensional (2D) crystals that are formed by specific and regular interactions with the surface of another crystalline material. Certain minerals were the first reported example of the heteroepitaxic nucleants for protein crystallization.24 However, unfortunately, these minerals were effective only for specific proteins probably due to a strict requirement for the lattice matching of crystals between substrate and protein, indicating the limited applicability of minerals as heteroepitaxic nucleants. Recently, we found that molecular sieves (MS), synthetic aluminosilicate crystalline polymers with regular micropores, could be a universal heteroepitaxic nucleant for protein crystallization. The addition of MS to an existing conventional crystallization condition provided higher quality of crystals in a variety of proteins: new crystal forms with better resolutions; larger crystals suitable for diffraction experiments; MS-dependent crystal formation of mutant proteins.25 However, the applicability of the heteroepitaxic nucleant method using MS in the sparse matrix crystallization screening has not yet been r 2010 American Chemical Society

Article

Crystal Growth & Design, Vol. 11, No. 1, 2011

111

Table 1. Crystallographic Statisticsa crystal form nucleant crystallization condition space group unit-cell parameter a (A˚) b (A˚) c (A˚) β (°) resolution range (A˚) no. of unique reflections redundancy completeness (%) Rmergeb (%) mosaicity (°)

I none CS No. 22

I MS 3A CS No. 22

I MS 4A CS No. 22

I MS 5A CS No. 22

P21

P21

P21

P21

P21

56.66 38.83 81.25 94.78 40-1.20 (1.24-1.20) 105854 (9839) 3.6 (3.6) 95.9 (89.8) 8.7 (49.5) 6.8 (3.0) 0.31

56.65 38.90 81.15 94.84 40-0.97 (1.00-0.97) 205244 (19620) 3.6 (3.4) 98.6 (94.8) 4.7 (28.9) 11.3 (4.3) 0.14

56.82 38.86 81.35 94.77 40-1.02 (1.06-1.02) 179691 (17875) 4.0 (3.9) 100 (100) 7.0 (46.1) 7.8 (3.2) 0.18

56.77 38.84 81.34 94.78 40-1.00 (1.04-1.00) 187076 (18168) 3.2 (3.2) 98.1 (96.1) 8.2 (49.0) 7.7 (3.3) 0.16

56.85 38.83 81.36 94.79 40-1.04 (1.08-1.04) 169707 (16874) 3.6 (3.6) 100 (100) 7.5 (46.6) 7.6 (3.0) 0.20

refinement resolution range (A˚) Rcryst/Rfreec (%) no. of molecules in ASU solvent content (%) rms deviation bond lengths (A˚) bond angles (°)

22.8-1.20 18.2/20.0 2 42.5

21.6-1.00 18.2/18.5 2 42.2

0.006 1.5

0.011 1.5

PDB code

3AKP

3AKT

crystal form nucleant crystallization condition space group unit-cell parameter a (A˚) b (A˚) c (A˚) β (°) resolution range (A˚) no. of unique reflections redundancy completeness (%) Rmergeb (%) mosaicity (°) refinement resolution range (A˚) Rcryst/Rfreec (%) no. of molecules in ASU solvent content (%) rms deviation bond lengths (A˚) bond angles (°) PDB code

I MS 13X CS No. 22

II none CS No. 33

II MS 5A CS No. 33

III MS 5A CS No. 45

III MS 13X CS No. 45

P21

P21

P212121

P212121

56.63 38.92 40.20 110.47 40-0.97 (1.00-0.97) 94660 (8939) 4.1 (3.9) 97.6 (92.8) 5.5 (35.9) 9.9 (3.9) 0.29

56.76 38.90 40.21 110.48 40-1.10 (1.14-1.10) 66824 (6663) 3.6 (3.6) 99.9 (100) 7.7 (44.4) 7.0 (3.4) 0.36

61.92 37.52 84.30

61.78 37.39 84.42

25-0.97 (1.00-0.97) 116654 (11482) 6.7 (6.4) 99.8 (99.3) 4.1 (47.0) 12.7 (3.5) 0.17

25-1.04 (1.08-1.04) 94491 (9328) 7.5 (7.6) 99.9 (100) 7.2 (46.0) 10.9 (4.1) 0.24

21.9-0.97 18.5/19.2 1 37.7

27.0-1.10 18.2/18.4 1 37.9

21.1-0.97 20.0/20.4 1 47.2

0.011 1.5

0.012 1.5

0.012 1.5

3AKS

3AKR P

3AKQ P

P Values in parentheses are for the outermost shell. b Rmerge = hkl i |Ii(hkl) - |/ hkl i Ii(hkl), ith observation of Pwhere Ii(hkl) is the P c reflection hkl and is the weighted average intensity for all observations i of reflection hkl. Rcryst= hkl ||Fobs| - |Fcalc||/ hkl |Fobs|, where |Fobs| and |Fcalc| are the observed and calculated structure-factor amplitudes, respectively. Rfree was calculated with 5% of the reflections chosen at random and omitted from refinement. a

P

examined. In this report, a sparse matrix screening experiment with and without MS has been performed using xylanase from Trichoderma longibrachiatum as a test protein, showing a clear improvement in the success rate of protein crystallization. 2. Experimental Section 2.1. Crystallization. As a test protein in this work, commercially available xylanase from T. longibrachiatum (EC 3.2.1.8, molecular weight 21 000, isoelectric point 9.0) at 36.0 mg/mL was used without further purification (Hampton research, HR7-106). The protein was screened against 50 sparse matrix crystallization conditions in total, using the Crystal Screen (CS) kit from Hampton Research.

The crystallization screening in the presence/absence of MS was performed manually using the oil-microbatch method.26 A crystallization drop of 1.0 μL was created by mixing a 1:1 of protein solution and precipitant solution in a well of a crystallization plate (Nunc HLA plate, Nalge Nunc International) which was then covered with 20 μL of paraffin oil. After the crystallization set up, the crystallization plate was stored at 293 K. The crystallization with MS was set up using the same procedure as described above except for the addition of MS into the crystallization drop as described elsewhere.25 In this study, MS 3A (Wako Pure Chemical Industries, 034-06095), 4A (037-06085), 5A (130-06075), and 13X (131-07085) were examined as heteroepitaxic nucleants. The morphology of the crystals was observed under a low-power microscope to compare those grown with and without MS. The droplets were observed after

112

Crystal Growth & Design, Vol. 11, No. 1, 2011

1, 2, 4, 7, 14, 21, 28, and 60 days and manually ranked as follows: score 0 for a clear solution; score 1 for a grainless precipitate; score 2 for a granulated precipitate; score 3 for a cluster of microcrystals; score 4 for a mountable cluster of crystals; score 5 for a moutable single crystal (Supporting Information Figure 1). Salt crystals were identified by diffraction experiments. The crystallization experiments were repeated at least six times for each condition in order to confirm the reproducibility of the results. 2.2. Diffraction Experiment and Structure Determination. All crystals were directly mounted in a cryoloop from the crystallization drop and flash-cooled at 100 K in a nitrogen gas stream. Complete diffraction data sets were collected using a Rigaku R-AXIS V image-plate detector with synchrotron radiation at BL26B1 of SPring-8, Japan.27 All data were processed with HKL2000.28

Sugahara et al. Positioning of the xylanase molecules in the asymmetric unit was carried out using the molecular replacement method as implemented in the program MOLREP.29 For the forms II and III from MS 5A, chain A (residues 1-190) of xylanase coordinates deposited in PDB (code 1enx) was used as a search model. For the form I from MS 5A, chain A (residues 2-190) of the form III was used as a search model. The structures of the forms I and II from conventional crystallization were isomorphous to those from MS-mediated crystallization with MS 5A, and thus they were determined by difference Fourier synthesis using the respective model of MS-mediated structure. Manual model revision was performed using QUANTA2000 software (Accelrys Inc.). The program CNS30 was used for the structure refinement and the electron-density map calculation. Each cycle of refinement with bulk-solvent and overall anisotropic B-factor corrections consisted

Table 2. Crystallization Screening of Xylanase with and without MSa

Article

Crystal Growth & Design, Vol. 11, No. 1, 2011

113

Table 2. Continued

a MS-dependent crystal formations are indicated by red shading. *, left and right values denote the buffer’s pH and the final solution’s pH, respectively, from the Hampton research home page (http://hamptonresearch.com/).

of the rigid-body refinement, the simulated annealing incorporating the slow-cool protocol, the positional refinement, and the B-factor refinement (individual or group). Several cycles of model revision and refinement yielded the final models. The stereochemical quality of the final structures was checked using the program PROCHECK.31 Statistics of the data collection and refinement are shown in Table 1. 2.3. Structural Evaluation. The superposition of protein models was performed using the program LSQKAB.32 To compare the interchain arrangement in the corresponding two protein molecules comprising two chains each, the pair of molecules is

superimposed twice in the primary and secondary fittings, according to the multiple superposition method.33 The contact area between two chains of protein molecules was calculated in QUANTA2000 (Accelrys Inc.) using parameters: the probe radius of 1.4 A˚ and the Z spacing factor of 0.01. In the calculation of contact area, solvent molecules and ligands were excluded. When a residue had alternate conformations, only a representative conformer with the lowest B-factor was included in the calculation. Unless otherwise stated, the contact area presented herein denotes a half value of total contact areas on both the relevant chains.

114

Crystal Growth & Design, Vol. 11, No. 1, 2011

Figure 1. Photographs of droplets under the crystallization conditions: CS No. 18 without MS (a) and with MS 5A (b); CS No. 47 without MS (c) and with MS 5A (d); CS No. 22 without MS (e) and with MS 5A (f); CS No. 33 without MS (g) and with MS 5A (h); CS No. 45 with MS 5A (i) and with MS 13X (j).

3. Results and Discussion 3.1. Overall Analysis of Screening Results. The sparse matrix crystallization screening of xylanase with/without MS 3A, 4A, 5A, and 13X was carried out using the Hampton CS kit (Table 2). In the absence of MS, two forms, I and II of protein crystals with scores 3-5, grew under 18 conditions. On the other hand, in the presence of MS, 29 conditions with MS 3A, 28 conditions with MS 4A, 34 conditions with MS 5A, and 40 conditions with MS 13X provided three forms I-III of protein crystals. When compared to the control screening without MS, the presence of MS 3A, 4A, 5A, and 13X produced crystals from 11, 10, 16, and 22 more conditions, respectively. Notably, all conditions that provided crystals in the control screen also provided crystals in the screening with MS except for the CS No. 36 with MS 4A, indicating an additive effect of MS to protein crystallization. In the case of xylanase, the most potent nucleating agent was MS 13X. Although xylanase did not crystallize in the absence of MS under 22 conditions (CS No. 2, 5, 8, 10, 11, 18, 19, 20, 25, 27, 30, 31, 34, 35, 37, 42, 43, 45, 46, 48, 49, and 50), the presence of MS allowed crystal formation under the same conditions. Interestingly, these crystallization conditions are predominantly acidic, suggesting that MS tends to facilitate the nucleation of xylanase crystals in acidic pH ranges. Because xylanase has a high pI value of 9.0, it is likely that a scaffold effect of MS in the heteroepitaxic crystal formation mitigates the overall crystallization incompatibility of xylanase in acidic pH ranges due to an electrostatic repulsion between positively charged protein molecules. In the absence of MS, the crystallization conditions CS No. 6, 17, 18, 40, and

Sugahara et al.

47 provided granulated precipitates (Figure 1a) or unusable crystals (Figure 1c). By contrast, single crystals with suitable sizes for X-ray diffraction studies grew on the MS surface (Figure 1b,d) under the same five conditions, which may allow us to skip the laborious optimization of crystallization conditions. These results clearly demonstrate that the presence of MS effectively improves the sparse matrix screening of protein crystallization in terms of yield and morphology of crystals. 3.2. Crystal Form I. The form I crystals have the symmetry of a space group P21 with unit-cell parameters of a = 57, b = 39, c = 81 A˚, β = 95°, and contain two chains of xylanase in the asymmetric unit. The form I crystals are most frequently observed under various crystallization conditions with and without MS (Figure 1e,f). Interestingly, the presence of MS obviously facilitates the formation of form I crystals; two to three times greater numbers of the sparse matrix conditions provided the form I crystals, and all the crystals that were formed only in the presence of MS belonged to the form I, with the exception of CS No. 45 (Table 2). To evaluate the MS effect on the quality of form I crystals, X-ray diffraction data of eight crystals from CS No. 22 with and without MS were collected (Table 1, Supporting Information Table 1). The crystals from the conventional method without MS diffracted X-rays to 1.18-1.30 A˚ resolutions and had mosaicities of 0.31-0.45°. On the other hand, the crystals from the nucleant method with MS 3A, 4A, 5A, and 13X diffracted X-rays to 0.97, 1.02, 1.00, and 1.04 A˚ resolution, respectively, and had sharp diffraction spots with mosaicities of 0.14-0.20°. In both resolution and mosaicity, the MS-mediated crystals showed better diffraction quality as compared with the crystals from the conventional method. All of the form I crystals revealed a layer-like packing structure (Figure 2a) that is typically observed in reported protein crystal structures with the heteroepitaxic growth.34,35 The layer-like structure in the form I crystals was observed only in a plane parallel to the crystallographic c axis. Therefore, assuming that the 3D crystal is composed of a stack of the respective 2D array, heteroepitaxic crystal growth occurs toward the [001] direction in the form I crystals. Each 2D array is composed of a tightly packed monolayer of xylanase molecules. From the definition of epitaxy, this tight packing of monolayer may be favorable to provide a rigid scaffold for the epitaxial crystal growth. Two xylanase chains in the asymmetric unit reside in adjacent two layers separately, thereby making a bilayer structure in the form I lattice. The first layer (z = 0; Figure 2b) contains the A chain and symmetry-related A chains that are related by two parallel crystallographic 21 screw axes. The packing mode in the second layer (z = 1/2; Figure 2c) containing the B chain and symmetry-related B chains is very similar to that in the first layer. From an overall comparison of the two layers, the only difference is the direction of their 21 screw axes that run opposite to each other, resulting in an upside-down alignment of bowl-shaped xylanase molecules in the second layer. To compare the packing mode within the two layers precisely, the interchain arrangement was analyzed by the multiple superposition method.33 First, the second layer was superimposed onto the first layer at corresponding CR atoms of chains A and B, which provided a small root-meansquare-deviation (rmsd) of about 0.2 A˚. Then, the second layer fitted was superimposed again onto the first layer at another corresponding pair of symmetry-related chains, and the rotational angle required for the secondary fitting was

Article

Crystal Growth & Design, Vol. 11, No. 1, 2011

115

Figure 2. Packing diagrams of xylanase crystals from MS 5A: the P21 cell of form I (a, b, c), the P21 cell of form II (d, e), and the P212121 cell of form III (f, g). Xylanase molecules are depicted as CR-trace models with distinguishable coloring. Various plane projections are shown with the unit cell boundary. Drawn in QUANTA2000. The crystallographic 21 screw axes are indicated using crystallographic symbols and identified by labeling. Chains in the asymmetric unit are colored light green and are labeled A or B. The chain labeling SnX denotes the nth symmetry-related chain X.

calculated. The difference in the interchain arrangement calculated by this method was about 1°, indicating essentially the same packing mode within the two layers. Although the bilayer structure of the form I crystals of xylanase is favorable for the heteroepitaxic crystal growth on MS surfaces, it also allows the spontaneous crystal growth from solution (Table 1, Supporting Information Table 1). Thus, it is intriguing to elucidate the reason why the heteroepitaxic growth improves the diffraction quality of form I crystals. In the form I crystals grown with and without MS, the unit-cell parameters are virtually constant considering the eight diffraction data sets collected. In order to evaluate the degree of isomorphism among these crystals precisely, we calculated a merging R value (Rmerge) from any pair of data sets using the program Scalepack.28 The result clearly shows that these eight form I crystals grown with and without MS are isomorphous (Table 3). This indicates that the heteroepitaxic growth with MS does not affect the global packing mode of the form I crystals. Notably, while a high level of isomorphism is kept among the MS-mediated crystals, a slight fluctuation in Rmerge values is observed in the conventional crystals grown without MS. Therefore, it is most likely

Table 3. Isomorphism in Form I Crystalsa nucleant none 1 none 2 none 3 none 4 MS 3A MS 4A MS 5A MS 13X none 1 none 2 none 3 none 4 MS 3A MS 4A MS 5A MS 13X

9.0 4.3 7.1 4.8 5.0 5.8 6.5

11.1 10.5 11.5 11.7 12.4 13.2

6.8 3.8 5.8 5.5 6.7

6.7 10.3 10.4 11.6

5.4 5.9 6.3

4.8 2.9

5.1

Rmerge was calculated to the resolution limit at 2.0 A˚, from a pair of scaled intensity data sets using the same definition as in Table 1, in which corresponding unique intensities with an index hkl from the two data sets were treated as different observations of hkl. a

that the MS surface provides a scaffold for the growth of form I crystals to reduce certain lattice defects, thereby yielding better diffraction quality and isomorphism of MSmediated crystals. 3.3. Crystal Form II. The form II crystals have the symmetry of a space group P21 with unit-cell parameters of a=57, b=39, c=40 A˚, β = 110°, and contain one chain of xylanase per asymmetric unit. The condition CS No. 33

116

Crystal Growth & Design, Vol. 11, No. 1, 2011

yielded the form II crystals both in the presence and the absence of MS (Figure 1g,h). The crystals from CS No. 33 without MS and with MS 5A yielded 0.97 A˚ and 1.10 A˚ resolution data sets with mosaicities of 0.29° and 0.36°, respectively. Thus, the crystals without MS are better in diffraction quality than those from MS, indicating that MS is not always effective to improve protein crystallization. The form II crystals also show the layer-like crystal packing (Figure 2d,e), while it is not as clear as in the form I case. Notably, the monolayer of xylanase molecules that is seen in the form I crystals is also found in the form II ones, which would be favorable for the heteroepitaxic crystal growth toward the [001] direction. To compare the packing mode within layers precisely, the interchain arrangement was analyzed using the same method as described in the Section 3.2. The CR superposition of chains between forms I and II provided a small rmsd of about 0.5 A˚. As a result, the difference in the interchain arrangement between the form II layer and either form I layer was about 3°, indicating essentially the same packing mode within these layers. This fact may indicate that this common layer structure is an energetically favorable oligomeric state of xylanase in solution. As it is conceivable from the clearly related cell parameters of both the forms (Table 1), the second layer in the form I crystals that contains the B chain is missing in the form II crystals. For that reason, the form II crystals show two-times reduction in the c-axis dimension and the disappearance of the bilayer structure, when compared to the form I crystals. Unfortunately, the scaffold effect of MS seems not to work well in the form II packing. It should be noted that the presence of MS converted the crystal form of xylanase from II to I under the conditions containing 30%(w/v) PEG 4K for CS No. 6 and 2 M ammonium sulfate for CS No. 47. Thus the MS-mediated crystal growth clearly prefers the form I crystal packing when a specific crystallization condition allows both packing modes of forms I and II. The form II crystals grown with and without MS are virtually isomorphous (Rmerge = 4.2% using data at 2 A˚ resolution), indicating a low flexibility of the form II packing that may disturb the adjustment for a slight lattice mismatch between the protein crystal and the MS surface. 3.4. Crystal Form III. Under the condition of CS No. 45 in the presence of MS 5A and 13X, a novel form of xylanase crystals referred to as form III appeared on the MS surfaces after 2-60 days (Figure 1i,j). On the other hand, in the absence of MS, the condition CS No. 45 yielded salt crystals. The form III crystals from MS have the symmetry of a space group P212121 with unit-cell parameters of a = 62, b = 37, c = 84 A˚, and contain one chain of xylanase per asymmetric unit. Diffraction data sets of the form III crystals from MS 5A and 13X were collected at 0.97 A˚ and 1.04 A˚ resolution with mosaicities of 0.17° and 0.24°, respectively (Table 1). Thus, form III exhibits one of the best diffraction qualities of xylanase crystals. The form III crystals also reveal the layer-like crystal packing (Figure 2f,g), suggesting the heteroepitaxic crystal growth toward the [001] direction. In the form III crystal, each 2D array is composed of a tightly packed monolayer of xylanase molecules that are aligned in a different mode as compared with the common layer of other forms I and II. However, in all three forms, the monolayer comprises one asymmetric chain and six symmetry-related chains that are related by two parallel crystallographic 21 screw axes in

Sugahara et al.

Figure 3. Zinc-mediated Sc1 crystal packing in form III crystals from MS 5A. The perspective is the same as that in Figure 2g. Drawn in QUANTA2000. (a) Close-up view of zinc binding site with (2Fo - Fc) electron-density map contoured at 2σ level. Protein residues are depicted as liquorice models. Bound zinc, chloride, and water are depicted as yellow, purple, and red spheres, respectively. (b) Protein-zinc binding network. The ab plane projection of form III is shown with the unit cell boundary. Xylanase molecules are depicted as CR-trace models with distinguishable coloring. The zinc clusters are illustrated as yellow bars with labels of bridging histidine residues.

general. For that reason, the form III’s cell parameters are apparently correlated to those of the other forms. To clarify the difference of the packing mode in these layers, the interchain arrangement was analyzed using the same method as described in the Section 3.2. The CR superposition of chains between forms I and III provided a small rmsd of about 0.3 A˚. The result shows that an interchain rearrangement of about 45° within the common layer can produce the form III layer. Therefore, the form III layer is topologically the same as the common layer, while its interchain arrangement is substantially changed probably due to the scaffold effect of MS. Importantly, the form III crystals of xylanase are obtained only in the presence of MS, demonstrating an MS-dependent packing-space expansion of protein crystals. To examine the production of form III crystals in the absence of MS, we performed a grid screen crystallization around the crystallization condition CS No. 45 with a total of 24 conditions (5, 10, 20, and 30%(w/v) of PEG 8K; pH 4.6, 5.8, 6.5, 7.5, 8.5, and 9.0). As a result after 2 months of incubation, the form

Article

Crystal Growth & Design, Vol. 11, No. 1, 2011

117

Figure 4. Superimposed CR-trace models: chains A (red) and B (blue) of form I with MS 5A, chain A (light green) of form II with MS 5A, and chain A (gray) of form III with MS 5A. Drawn in QUANTA2000. (a) Overall view. A major structural difference is seen in the molecular surface loop T100-L105. (b) Conformation of the T100-L105 loop defined by crystal packing. The lower panel shows a close-up view of the loop. In the form II crystal, this loop locates in the vicinity of a crystal packing interface, as shown in the upper panel of packing diagram. The symmetryrelated chain in the form II is depicted as a yellow CR model with a stick model of N143 that causes steric clashes with superimposed models of other forms.

III crystals were not observed, providing a clear counterevidence for the conventional nucleation of this crystal form. Thus, it is intriguing to elucidate the mechanism by which MS induces the form III packing. Reflecting that the form III crystals grew under the crystallization condition containing 0.2 M zinc acetate, bound ions were observed in the crystal structure (Figure 3a). From the comparison of its temperature factor to those of neighboring atoms and from its coordination with histidine residues, the bound ions are most likely to be zinc ions from the crystallization reagent. This interpretation is in agreement with the fact that strong signals are observed at the ion sites in the anomalous Fourier map at CuKR wavelength prepared using the program FFT in CCP4 suite.36 Five zinc ions located in between the His144 of the asymmetric chain and the His22 of a neighboring symmetry-related chain to form a zinc cluster. In the form III crystals, this zinc cluster may act as an interchain linker to form a monolayer of xylanase molecules (Figure 3b). Notably, the crystallization condition CS No. 46, which is essentially the same as the condition CS No. 45 except for the replacement of additive zinc ion by calcium, did not provide the form III crystals in the presence of MS. This indicates that zinc ion but not calcium is selectively required to promote the MSmediated formation of form III crystals. Probably, a breakdown of the spontaneous layer structure that is commonly observed in the other forms I and II is required to expand the packing space of xylanase crystals, which is accomplished by the combination of MS and zinc ion to provide the rearranged zinc-mediated layer in the form III crystals. Because the form III is not available spontaneously from solution, MS can be regarded as the form III specific crystallization catalyst in which zinc ion acts as an essential cofactor. 3.5. Structural Difference of Xylanase Molecules Depending on Crystal Forms. Because the molecular function of a protein is highly correlated with its 3D structure, the structural difference depending on crystal forms should be evaluated carefully. In order to detect conformational changes of xylanase molecules in the different crystal forms I, II, and III, a CR superposition analysis was performed (Figure 4a). In

any pair of xylanase chains from the three forms of crystals grown with and without MS, the CR superposition provided a small rmsd value less than 0.62 A˚. This indicates that crystal packing does not affect the overall conformation of xylanase molecule. However, regarding local structures, a molecular surface loop consisting of residues 100-105 in form II showed a different conformation as compared with those in the other forms I and III. In the form II, this surface loop resides in the vicinity of the crystal packing interface. Notably, a clear steric clash occurs at Asn143 when any corresponding chain in the other forms I and III is superimposed onto the asymmetric chain in the form II (Figure 4b). Therefore, structural flexibility of this surface loop may play a key role in the formation of form II crystals of xylanase, although it does not affect the overall molecular conformation. 3.6. Quantitative Evaluation of Layer-Like Structure. Although the visual inspection of present crystal structures revealed that all the MS-mediated xylanase crystals are likely to be formed by the heteroepitaxic mechanism on the MS surfaces, a quantitative method would be favorable to evaluate the layer-like structure objectively. The layer is defined as a continuous lateral assembly of protein molecules packed in a crystal lattice. One of the most simple and effective methods to evaluate the crystal packing is to measure the contact area at the packing interface of protein molecules in the crystal lattice.37 Thus, we introduce an index for the bias of layer packing in which the contact areas for a protein molecule are classified into two components according to whether they contribute laterally or longitudinally to the layer packing, and the ratio of the lateral contact area to the longitudinal one is calculated (eq 1). rc ¼ X

X ðlateral contact areaÞ= ðlongitudinal contact areaÞ

ð1Þ

The procedure of the rc (ratio of contact area) analysis is as follows. First, all symmetry-related protein molecules surrounding the query molecule are produced by crystallographic

118

Crystal Growth & Design, Vol. 11, No. 1, 2011

Sugahara et al.

symmetry operations. In the case of a monomeric protein, the query molecule is a polypeptide chain selected from the chains in the asymmetric unit of crystal. In the case of an oligomeric protein, the whole biological oligomer containing asymmetric chains is regarded as the query molecule. Then, the lateral or the longitudinal contact area is calculated between the query molecule and any adjacent molecule within the layer or that in the other layers, respectively. An rc value close to 1.0 denotes an unbiased crystal packing, while a large deviation of rc value from 1.0 denotes a strong bias in crystal packing, indicative of the clear layer-like structure. Another important factor to evaluate the layer-like structure is an estimation of protein interaction to the heteroepitaxic substrate surface. Considering of the dimension of spherical protein molecules that is typically more than 30 A˚ in diameter, the surface asperity of zeolite crystals38,39 would be negligible. If we assume the substrate surface as a plane, the density of interacting residues to the substrate plane (dp) can be estimated by eq 2. Z terminus þ δ ðresidue numbers from a layerÞ dz= dp ¼ terminus - δ

ðlateral unit cell areaÞ

ð2Þ

The procedure of the dp analysis is as follows. First, all protein molecules from a query layer are produced in a unit cell by crystallographic symmetry operations. Then, the query layer model is scanned along the longitudinal direction z, and CR atoms in a section that has a width ( δ from the bottom or the top of the model are counted and summed. The section width δ means a vertical distance to the substrate plane and is set to 7.0 A˚ in this study. Finally, the dp value is calculated as the CR numbers divided by the lateral area of unit cell (the ab plane area in this study). From its definition, the larger dp value indicates the higher probability of the protein-substrate interactions. In the form I xylanase crystals from MS5A, the first layer contains the asymmetric chain A and six symmetry-related A chains for the lateral interactions (Figure 2b): sym1A(x, y þ 1, z), 111 A˚2; sym2A(x, y - 1, z), 111 A˚2; sym3A(-x, y þ 1/2, -z), 471 A˚2; sym4A(-x, y - 1/2, -z), 471 A˚2; sym5A(-x - 1, y þ 1/2, -z), 361 A˚2; sym6A(-x - 1, y - 1/2, -z), 361 A˚2. Thus, the total lateral contact area for the first layer is 1886 A˚2, whereas the longitudinal interactions with the asymmetric chain A are mediated by three second layer chains including the other asymmetric chain B and two symmetry-related B chains (Figure 2a): sym7B(-x, y 1/2, -z þ 1), 200 A˚2; sym8B(-x, y - 1/2, -z), 171 A˚2; B(x, y, z), 7 A˚2. Thus, the total longitudinal contact area for the first layer is 379 A˚2. As a result, the rc value for the first layer is calculated as 4.98, indicating a strong packing bias that accounts for the clear layer-like structure of form I crystals. In the first layer, four chains sym1A, sym2A, sym3A, and sym4A are related to the asymmetric A chain by one of the crystallographic 21 screw axis referred to as Sc1 that produces 1163 A˚2 for lateral interactions, whereas four chains sym1A, sym2A, sym5A, and sym6A are related to the other 21 screw axis referred to as Sc2 that produces 944 A˚2 in the lateral contact area. This fact indicates that Sc1 and Sc2 interactions are comparable but the Sc1’s one seems to be a little stronger than that of Sc2 by about 200 A˚2 difference in contact area. The dp value for the first layer was calculated using the chains A and sym3A to be 1.09  10-2/A˚2. The

second layer showed similar results; the lateral/longitudinal contact areas and the values rc and dp were 1891 A˚2, 379 A˚2, 4.99, and 1.22  10-2/A˚2, respectively. We also compared the form I crystals with and without MS and got similar results; in the conventional crystal, the lateral/longitudinal contact areas and the values rc and dp were 1909 A˚2, 381 A˚2, 5.01, and 1.09  10-2 /A˚2 for the first layer, and 1918 A˚2, 381 A˚2, 5.04, and 1.23  10-2 /A˚2 for the second layer, respectively. A little larger dp values in the second layers may indicate that the heteroepitaxic nucleation occurs in the second layer. In the form II crystals from MS5A, the single layer contains the asymmetric chain A and six symmetry-related A chains for the lateral interactions (Figure 2e): sym1A(x, y þ 1, z), 118 A˚2; sym2A(x, y - 1, z), 118 A˚2; sym3A(-x, y þ 1/2, -z), 485 A˚2; sym4A(-x, y - 1/2, -z), 485 A˚2; sym5A(-x - 1, y þ 1/2, -z), 387 A˚2; sym6A(-x - 1, y - 1/2, -z), 387 A˚2. Thus, the total lateral contact area for the form II layer is 1980 A˚2, which is similar to those in the form I layers. The Sc1/Sc2 interactions represent contact areas of 1206 A˚2 and 1010 A˚2, respectively, which is also in good agreement with the form I case. By contrast, the longitudinal interactions with the asymmetric chain A are different from the form I case and are mediated by four symmetry-related A chains in adjacent layers (Figure 2d): sym7A(x, y, z þ 1), 262 A˚2; sym8A(-x, y þ 1/2, -z þ 1), 218 A˚2; sym9A(-x, y 1/2, -z þ 1), 218 A˚2; sym10A(x, y, z - 1), 262 A˚2. Thus, the total longitudinal contact area for the form II layer is 960 A˚2. As a result, the rc value for the form II is calculated as 2.06, indicating a moderate packing bias that is consistent with the unclear layer-like structure of form II crystals. Total contact area of the form II layer is 2939 A˚2, which is 1.3 times greater than those of the form I layers. The greater contact area in the form II layer indicates the more intimate interlayer interactions. This may be relevant to the fact that the heteroepitaxic crystal formation does not work well in the form II, in addition to the potential lattice mismatch between MS and the form II layer. The dp value for the form II layer was calculated using the chains A and sym3A to be 1.04  10-2/A˚2. This form II’s dp value is a little smaller than those of form I layers, which is also consistent with the preference of form II crystals for the spontaneous nucleation rather than for the heteroepitaxic one. We compared the form II crystals with and without MS and got essentially the same results; in the conventional crystal, the lateral/longitudinal contact areas and the values rc and dp were 1989 A˚2, 965 A˚2, 2.06, and 1.04  10-2/A˚2, respectively. In the form III crystals from MS5A, the single layer contains the asymmetric chain A and six symmetry-related A chains for the lateral interactions (Figure 2g): sym1A(x, y þ 1, z), 219 A˚2; sym2A(x, y - 1, z), 219 A˚2; sym3A(-x þ 1, y þ 1/2, -z þ 1/2), 4 A˚2; sym4A(-x þ 1, y - 1/2, -z þ 1/2), 4 A˚2; sym5A(-x, y þ 1/2, -z þ 1/2), 362 A˚2; sym6A(-x, y 1/2, -z þ 1/2), 362 A˚2. Thus, the total lateral contact area for the form III layer is 1169 A˚2, which is only about 60% of the corresponding area in the other form’s layers. The Sc1/Sc2 interactions represent contact areas of 445 A˚2 and 1161 A˚2, respectively, indicating that the zinc binding destroys the Sc1 protein-protein interactions. The longitudinal interactions with the asymmetric chain A are also different from the other form’s case and are mediated by only two symmetry-related A chains in adjacent layers (Figure 2f): sym7A(-x þ 1/2, -y, z þ 1/2), 257 A˚2; sym8A(-x þ 1/2, -y, z - 1/2), 257 A˚2. Thus, the total longitudinal contact area for the form III layer is 514 A˚2. As a result, the rc value for the form III is

Article

Crystal Growth & Design, Vol. 11, No. 1, 2011

119

Figure 5. Putative MS recognition sites. Common residues at layer interfaces are labeled. Drawn in QUANTA2000. (a) The ac plane projection of form I crystal. (b) The ac plane projection of form III crystal.

calculated as 2.28, indicating a substantial packing bias that is in between rc values of the other forms I and II. Total contact area of the form III layer is 1683 A˚2, which is the smallest in the three forms, indicating that the zinc binding effectively compensates the decreased contact area to provide the excellent diffraction quality of form III crystals. The dp value for the form III layer was calculated using the chains A and sym3A to be 1.38  10-2 /A˚2, which was the largest in the three forms. From its definition, the heteroepitaxic nucleation should occur through short-range contacts between the protein layer and the substrate surface. Since the dp value represents the density of protein residues proximal to the substrate plane (within 7.0 A˚), the higher dp value suggests more contacts that may be favorable for the heteroepitaxic crystal nucleation. Therefore, the high dp value of the form III layer may be a reason for form III’s exclusive preference for the MS-mediated heteroepitaxic nucleation. 3.7. Protein-MS Interaction. From a comparison of crystal packing between the three forms I-III, residue Glu107 is commonly observed at the exposed surface on the monolayer of xylanase molecules and looks favorable for the interaction with MS surfaces (Figure 5). Therefore, Glu 107 is one of the candidates for the contact point with MS surfaces, in which the glutamate side-chain head may fit to the larger size of the MS 13X lattice. The dp analysis identified 21 residues of xylanase for the putative protein-MS interactions: K104-I117, R142, N154, N157, A158, Q161-G163. MS used in this research is an aluminosilicate crystalline material with identical and ordered pores: 3A, 4A, 5A, and 13X with pore sizes of 3, 4, 5, and 13 A˚, respectively.38,39 Each MS material noncovalently binds specific metal ions with regular coordination architecture: Kþ for MS 3A, Naþ for MS 4A, Ca2þ for MS 5A, and Naþ for MS 13X. In our previous report on five test proteins,25 the most successful

nucleating agent was MS 5A. However, in the case of xylanase, the success rate of crystallization from MS 13X is higher than those from other MS, indicating that the suitable type of MS for the heteroepitaxic crystal formation varies from protein to protein. The MS-mediated crystallization may depend on the hydrophobicity of MS materials that can be estimated from the Si/Al ratio. The ratio is 1.00 for the type A and 1.23 for the type X, indicating that the MS 13X is more hydrophobic than the MS 5A. If the van der Waals interactions dominate the heteroepitaxic nucleation of xylanase crystals, the more hydrophobic nature of MS 13X may be the reason for its higher success rate in the xylanase crystallization. 4. Conclusion Although the sparse matrix crystallization screening is widely used in protein crystallography, coarse and unadjustable samplings of the screening often fail in obtaining high quality of crystals especially in the case of proteins with poor crystallizability. In this work, the heteroepitaxic nucleant method using MS was applied to the sparse matrix crystallization screening of xylanase, which provided formation of larger single crystals with better diffraction quality as compared with the conventional screening. In many cases, MS induced crystal formation under the conditions that did not provide any protein crystal using the conventional crystallization method without MS. In current protein crystallography, the available sample amount for crystallization trials is limited especially when the expression level of the target protein is low. In such cases, a minimum set of sparse matrix crystallization screening in the presence of MS would be the most effective way to achieve the diffraction quality of protein crystals. Importantly, the structure of the MS-dependent new

120

Crystal Growth & Design, Vol. 11, No. 1, 2011

crystal form III revealed a synergistic effect of MS and zinc ion to expand the packing space of protein crystals. This form III is not available spontaneously from solution, which is a clear counter-evidence for the conventional nucleation. We also observed this type of counter-evidence for four other proteins in the previous report.25 Furthermore, all protein crystals from MS, including the crystals in the previous report, reveal the layer-like packing structure, being indicative of the heteroepitaxic crystal growth. Therefore, we believe that the MSmediated protein crystallization is based on a kind of epitaxic scaffold effect of MS surfaces. However, currently we do not know why MS can induce the heteroepitaxic nucleation on a wide variety of proteins, which remains an open question. It is conceivable that a strict lattice matching may not be required for the heteroepitaxic protein crystallization with MS when compared to those with minerals24 such as mica. Acknowledgment. M.S. was the main contributor to this study, designed and performed all the experiments, and wrote the paper; Y.K.M. codesigned and assisted the crystallization screening experiment; N.K. supervised this work, performed the quantitative structural evaluation, and cowrote the paper. We thank the beamline staff for assistance during data collection at BL26B1 of SPring-8 (Proposal No. 20080101, No. 20090106). This work was supported in part by an “Incentive Research Grant” from RIKEN to M.S., and by a “Grants-in Aid for Scientific Research” from the Ministry of Education, Culture, Sports, Science and Technology of Japan (No. 22770116) to M.S. Supporting Information Available: Data collection statistics (SI Table 1) and figure of categories for the evaluation of droplet images (SI Figure 1) are available free of charge via the Internet at http:// pubs.acs.org.

References (1) Jancarik, J.; Kim, S. H. J. Appl. Crystallogr. 1991, 24, 409–411. (2) Stevens, R. C. Curr. Opin. Struct. Biol. 2000, 10, 558–563. (3) Sugahara, M.; Miyano, M. Tanpakushitsu Kakusan Koso 2002, 47, 1026–1032. (4) Sulzenbacher, G.; Gruez, A.; Roig-Zamboni, V.; Spinelli, S.; Valencia, C.; Pagot, F.; Vincentelli, R.; Bignon, C.; Salomoni, A.; Grisel, S.; Maurin, D.; Huyghe, C.; Johansson, K.; Grassick, A.; Roussel, A.; Bourne, Y.; Perrier, S.; Miallau, L.; Cantau, P.; Blanc, E.; Genevois, M.; Grossi, A.; Zenatti, A.; Campanacci, V.; Cambillau, C. Acta Crystallogr. 2002, D58, 2109–2115. (5) Watanabe, N.; Murai, H.; Tanaka, I. Acta Crystallogr. 2002, D58, 1527–1530. (6) Hui, R.; Edwards, A. J. Struct. Biol. 2003, 142, 154–161. (7) Adachi, H.; Takano, K.; Matsumura, H.; Niino, A.; Ishizu, T.; Inoue, T.; Mori, Y.; Sasaki, T. Jpn. J. Appl. Phys. 2004, 43, L76– L78. (8) Shah, A. K.; Liu, Z. J.; Stewart, P. D.; Schubot, F. D.; Rose, J. P.; Newton, M. G.; Wang, B. C. Acta Crystallogr. 2005, D61, 123–129. (9) Miyatake, H.; Kim, S. H.; Motegi, I.; Matsuzaki, H.; Kitahara, H.; Higuchi, A.; Miki, K. Acta Crystallogr. 2005, D61, 658–663. (10) Walter, T. S.; Diprose, J. M.; Mayo, C. J.; Siebold, C.; Pickford, M. G.; Carter, L.; Sutton, G. C.; Berrow, N. S.; Brown, J.; Berry, I. M.; Stewart-Jones, G. B.; Grimes, J. M.; Stammers, D. K.; Esnouf, R. M.; Jones, E. Y.; Owens, R. J.; Stuart, D. I.; Harlos, K. Acta Crystallogr. 2005, D61, 651–657.

Sugahara et al. (11) Hazes, B.; Price, L. Acta Crystallogr. 2005, D61, 1165–1171. (12) Berry, I. M.; Dym, O.; Esnouf, R. M.; Harlos, K.; Meged, R.; Perrakis, A.; Sussman, J. L.; Walter, T. S.; Wilson, J.; Messerschmidt, A. Acta Crystallogr. 2006, D62, 1137–1149. (13) Hiraki, M.; Kato, R.; Nagai, M.; Satoh, T.; Hirano, S.; Ihara, K.; Kudo, N.; Nagae, M.; Kobayashi, M.; Inoue, M.; Uejima, T.; Oda, S.; Chavas, L. M.; Akutsu, M.; Yamada, Y.; Kawasaki, M.; Matsugaki, N.; Igarashi, N.; Suzuki, M.; Wakatsuki, S. Acta Crystallogr. 2006, D62, 1058–1065. (14) Iino, H.; Naitow, H.; Nakamura, Y.; Nakagawa, N.; Agari, Y.; Kanagawa, M.; Ebihara, A.; Shinkai, A.; Sugahara, M.; Miyano, M.; Kamiya, N.; Yokoyama, S.; Hirotsu, K.; Kuramitsu, S. Acta Crystallogr. 2008, F64, 487–491. (15) Sugahara, M.; Asada, Y.; Shimizu, K.; Yamamoto, H.; Lokanath, N. K.; Mizutani, H.; Bagautdinov, B.; Matsuura, Y.; Taketa, M.; Kageyama, Y.; Ono, N.; Morikawa, Y.; Tanaka, Y.; Shimada, H.; Nakamoto, T.; Sugahara, M.; Yamamoto, M.; Kunishima, N. J. Struct. Funct. Genomics 2008, 9, 21–28. (16) Chayen, N. E.; Saridakis, E. Nat. Methods 2008, 5, 147–153. (17) Saridakis, E.; Chayen, N. E. Trends Biotechnol. 2009, 27, 99–106. (18) Chayen, N. E.; Saridakis, E.; El-Bahar, R.; Nemirovsky, Y. J. Mol. Biol. 2001, 312, 591–595. (19) Chayen, N. E.; Saridakis, E.; Sear, R. P. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 597–601. (20) Georgieva, D. G.; Kuil, M. E.; Oosterkamp, T. H.; Zandbergen, H. W.; Abrahams, J. P. Acta Crystallogr. 2007, D63, 564–570. (21) Thakur, A. S.; Robin, G.; Guncar, G.; Saunders, N. F. W.; Newman, J.; Martin, J. L.; Kobe, B. PLoS ONE 2007, 10, e1091. (22) Kallio, J. M.; Hakulinen, N.; Kallio, J. P.; Niemi, M. H.; K€arkk€ainen, S.; Rouvinen, J. PLoS ONE 2009, 4, e4198. (23) Asanithi, P.; Saridakis, E.; Govada, L.; Jurewicz, I.; Brunner, E. W.; Ponnusamy, R.; Cleaver, J. A. S.; Dalton, A. B.; Chayen, N. E.; Sear, R. P. Appl. Mater. Interfaces 2009, 1, 1203–1210. (24) McPherson, A.; Shlichta, P. Science 1988, 239, 385–387. (25) Sugahara, M.; Asada, Y.; Morikawa, Y.; Kageyama, Y.; Kunishima, N. Acta Crystallogr. 2008, D64, 686–695. (26) Chayen, N. E.; Shaw Stewart, P. D.; Maeder, D. L.; Blow., D. M. J. Appl. Crystallogr. 1990, 23, 297–302. (27) Ueno, G.; Kanda, H.; Hirose, R.; Ida, K.; Kumasaka, T.; Yamamoto, M. J. Struct. Funct. Genomics 2006, 7, 15–22. (28) Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276, 307–326. (29) Vagin, A.; Teplyakov, A. J. Appl. Crystallogr. 1997, 30, 1022– 1025. (30) Br€ unger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.; Nilges, M.; Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren, G. L. Acta Crystallogr. 1998, D54, 905–921. (31) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J. M. J. Appl. Crystallogr. 1993, 26, 283–291. (32) Kabsch, W. Acta Crystallogr. 1976, A32, 922–923. (33) Kunishima, N.; Asada, Y.; Sugahara, Ma.; Ishijima, J.; Nodake, Y.; Sugahara, Mitsu.; Miyano, M.; Kuramitsu, S.; Yokoyama, S.; Sugahara, M. J. Mol. Biol. 2005, 352, 212–228. (34) Edwards, A. M.; Darst, S. A.; Hemming, S. A.; Li, Y.; Kornberg, R. D. Nat. Struct. Biol. 1994, 1, 195–197. (35) Hemming, S. A.; Bochkarev, A.; Darst, S. A.; Kornberg, R. D.; Ala, P.; Yang, D. S.; Edwards, A. M. J. Mol. Biol. 1995, 246, 308– 316. (36) Collaborative Computational Project, Number 4. Acta Crystallogr. 1994, D50, 760–763. (37) Mizutani, H.; Saraboji, K.; Malathy Sony, S. M.; Ponnuswamy, M. N.; Kumarevel, T.; Krishna Swamy, B. S.; Simanshu, D. K.; Murthy, M. R. N.; Kunishima, N. Acta Crystallogr. 2008, D64, 1020–1033. (38) Reed, T. B.; Breck, D. W. J. Am. Chem. Soc. 1956, 78, 5972– 5977. (39) Olson, D. H. J. Phys. Chem. 1970, 74, 2758–2764.