Ariel Axelbaum
Axelbaum, 2014
1
The Effect of 2-methyl-2,4-pentanediol on the Crystal Structure of Lysozyme
Thesis Submitted in Partial Fulfillment of the Requirements of the Jay and Jeanie Schottenstein Honors Program
Yeshiva College Yeshiva University May 2014
Ariel Axelbaum Mentor: Prof. Neer Asherie, Departments of Physics and Biology
Axelbaum, 2014
2
Abstract Chiral control of crystallization has ample precedent in the small-molecule world, but relatively little work has been carried out to investigate the role of chirality in protein crystallization. In this study, lysozyme was crystallized in the presence of the chiral additive 2-methyl-2,4-pentanediol (MPD), using the (R) and (S) enantiomers as well as the racemic (RS) mixture. The crystals grown with (R)-MPD had the least disorder and produced the highest resolution protein structures. This result is consistent with the observation that for (R)- and (RS)-MPD, the crystal contacts are made by (R)-MPD demonstrating that there is preferential interaction between lysozyme and this enantiomer. These findings support the hypothesis that chiral interactions are important in protein crystallization. Additionally, to confirm the conformational assignment of MPD in our crystal structures, the conformations of MPD molecules in crystal structures from the Protein Data Bank were examined. Using the number of occurrences of each conformer, the relative energies of the conformers were calculated. It was shown that MPD favors the conformation that allows for an intramolecular hydrogen bond between its hydroxyl groups and maintains an anti-anti configuration for its carbon backbone, which corresponds to the conformer found in our crystals.
1. Introduction The field of protein crystallization is important for many different applications, including many areas of biomedical research. This is because proteins have many vital roles in cells and control most of the functions carried out in cells.1 Most enzymes are proteins and in order to comprehend the reactions that enzymes catalyze, a knowledge of the protein structure is important. Additionally, many diseases, including Alzheimer’s, Parkinson’s, and Huntington’s diseases, have been shown to result from changes in protein structure.2,3 For these reasons there is a large emphasis placed on the determination of protein tertiary structures. Protein structures are traditionally determined through the crystallization of proteins and diffracting x-rays by the crystals. The x-ray diffraction pattern can be used to determine the electron density map of the protein. The beginning of structure determination of protein crystals began in 1958 when John Kendrew determined the atomic structure of myoglobin.4 This achievement was soon followed by the crystallization and determination of the structures of lysozyme and ribonucleases A and S.5,6 Since then, many protein structures have been determined by x-ray diffraction using protein crystals. Currently, there are 30,832 unique protein structures in the Protein Data Bank.7,8 While protein structures can be
Axelbaum, 2014
3
determined by several different techniques, structures determined by x-ray diffraction of protein crystals are useful as they can have high resolution, even greater than 1Å. High resolution is very useful for crystallographers because it corresponds to a high level of detail available in the electron density map. To achieve high resolution structures, the protein crystals must be made of many, well-ordered unit cells. A unit cell is the smallest repeating unit in a crystal and the more ordered the packing of the unit cells, the higher the resolution of the crystal. Salt crystals, which are made of very small, tightly packed ions, make very ordered crystals that can diffract to very high resolutions. Proteins are larger macromolecules that make fewer, weaker bonds than salts and also lose many more degrees of freedom when forming crystals. These properties result in looser packing of the protein crystals that leads to some disorder in the basic repeating unit and ultimately to a limitation of the resolution. Protein crystallization is difficult. Even without considering the requirements for quality diffraction, the crystallization of a protein requires the determination of the specific precipitant or precipitants that will lead to the crystallization of that protein. The precipitant that works for one protein is in no way guaranteed to work for another protein. Proteins are made of long chains of amino acids, and predicting the additives needed for the crystallization of any protein is challenging. In some ways resembling alchemists, scientists mix proteins with a large range of chemicals in the hope of obtaining crystals. From experiments, scientists have discovered the usefulness of alcohols and salts for crystallization and they have become the most common crystallization additives.9 However, even the right additive usually requires a very specific set of conditions that include concentration, pH, and temperature. Protein crystallization continues to stump crystallographers and finding the exact set of conditions needed to form crystals—and, in particular, the conditions that will produce crystals that will diffract to high resolution—can require several months, if not years. Even with many advances in the field of protein crystallization, it is still estimated that fewer than three percent of proteins from cloned genes have had their structure determined.10,11 Growing diffraction quality crystals represents the largest challenge for crystallographers. Furthermore, as the field of protein crystallization has developed, the crystallization of additional proteins has become more difficult for crystallographers. This is because proteins that crystallize with multiple precipitants over a broad range of conditions often have already been crystallized and their structures determined, while proteins that crystallize over a more narrow set of conditions using one precipitant remain elusive.12 While the number of solved protein structures has continued to grow, this is more of an indication of the effort and money that is going into the crystallization of proteins rather than
Axelbaum, 2014
4
an increased success rate.13 Scientists are learning more and more about the crystallization process and, while there is no conclusive formula for the crystallization of proteins, scientists are constantly developing new strategies and new tools to aid in the crystallization process. As part of the search for new crystallization strategies, my group has been studying the effect of chiral precipitants on protein crystallization. Chiral molecules are molecules that have identical chemical formulas but differ in their three-dimensional structure: the atoms are in a different arrangement such that the molecules are mirror images of one another and are non-superimposable (figure 1). In a pair of chiral molecules, each molecule is called an enantiomer. While both enantiomers interact with non-chiral molecules in the same way, when reacting with another chiral molecule the enantiomers react differently. Nearly all amino acids, the building blocks of proteins, are chiral molecules and thus interact with chiral molecules differently. Our experiments were designed to see if we could find a chiral effect in the crystallization of proteins.
Figure 1. A chiral molecule and its mirror image.
In a previous study, my group worked with thaumatin and tartrate as our protein-precipitant pair and discovered that the chirality of the precipitant affected the time needed for the crystallization of thaumatin as well as the crystal habit of the protein crystals.14-16 Yet, however interesting our findings were with thaumatin and tartrate, it was possible that this chiral effect was limited to this one proteinprecipitant pair. In order to show the generality of this chiral effect, it needed to be demonstrated in other protein-precipitant pairs. For our study we decided investigate the interactions between the protein lysozyme and the precipitant MPD. Lysozyme is one of the most studied proteins in the field of protein crystallization.17 First crystallized in 1965 by David Chilton Phillips and his collaborators (PDB ID 1LYZ), lysozyme was the second protein, and the first enzyme, to have its structure determined.5 Due to its early structure
Axelbaum, 2014
5
determination, low cost, stability, and ease with which it can be crystallized, lysozyme has been used by researchers as a model protein to study many aspects of protein crystallization. Lysozyme is found in many different species and between all the various types of lysozyme there are 1,359 structures in the RSCB Protein Data Bank (PDB) [www.pdb.org]. Also, it has the highest resolution structure for a macromolecule with over 100 residues (0.65Å; PDB ID 2VB1). For these reasons we chose it for our study of the generality of the chiral effect. Similarly, one of the most commonly used precipitant additives is 2-methyl-2,4-pentanediol, which is commonly known as MPD. MPD is a chiral molecule with one chiral center at C4, and its two enantiomers are (R)-MPD and (S)-MPD (figure 2). The racemic form of MPD, (RS)-MPD, where both enantiomers are present in equal quantities, is the only form of MPD used until now for crystallization experiments.18 MPD is a ligand in 1158 structures in the Protein Data Bank as either (R)-MPD or (S)MPD. MPD is a diol with two hydroxyl groups that make it soluble in water (up to 118.2g/L at 20 °C)19 and also allow it can interact with the polar groups of the protein. Additionally, MPD has a carbon backbone that allows it to interact with hydrophobic groups. These interactions can drive the protein solution to supersaturation and cause the protein to crystallize. C
Figure 2. A) (R)-MPD B) (S)-MPD. In the pictures, the hydrogen atoms are not shown. Pink spheres represent the oxygen atoms and the other colored spheres represent carbon atoms. C) The chemical structure of MPD. Numbers and letters in blue are for labeling of the carbons (and do not indicate anything about the chemical formula).
Another reason that this protein-precipitant pair was chosen has to do with an inconsistency in the literature. Two groups crystallized lysozyme with MPD and analyzed the binding of MPD to lysozyme at phenylalanine 34, but each group reports a different enantiomer at this site. Weiss et al. report that (R)MPD is found at the site while Michaux et al. report (S)-MPD.20,21 Both assignments are supported by their data. With just these two structures, no preference for one enantiomer over the other could be established. This thesis will attempt to establish whether there is a preference of lysozyme for one
Axelbaum, 2014
6
enantiomer or the other. In order to do this, we crystallized lysozyme with pure solutions of each enantiomer of MPD, in addition to the racemic form. This allowed us to analyze the individual interactions of each enantiomer with the protein. This is useful knowledge in its own right, but it also allowed us to analyze the contribution each enantiomer has in the crystals formed in the racemic solution. Our studies led us into another area of investigation. Our structures, and many of the structures we looked at, made us realize the difficulty in assigning MPD molecules from the electron density obtained by the x-ray diffraction experiments. This is due both to the similarity between the enantiomers and the multiple conformations that MPD can take. As an additional check on our and others’ assignments we looked at the energetic differences between the various conformations that MPD can take.
2. Materials and Methods 2.1 Materials Lysozyme (cat. no. 2933, lot no. 36P9210) was purchased from Worthington Biochemical Corporation (Lakewood, NJ). (R)-MPD and (S)-MPD were synthesized by Reuter Chemische Apparatebau KG (Freiburg, Germany). (RS)-MPD (cat. no. 68340, lot no. 1345630) was purchased from Sigma-Aldrich (St. Louis, MO). Tris base (cat. no. BP512-500), sodium azide (cat. no. S227I-500), and hydrochloric acid (cat. no. A144S-500) were purchased from Fisher Scientific (Pittsburgh, PA). All materials were used without further purification. The purity of the protein was checked using highperformance liquid chromatography (HPLC) and quasielastic light scattering (QLS), as described elsewhere.15 Deionized water was obtained from an Integral 3 deionization system (Millipore, Billerica, MA). Solutions were filtered through a Nalgene disposable 0.22 µm filter unit (Nalge Nunc International, Rochester, NY) prior to use. Concentration measurements were carried out by UV-Vis extinction spectroscopy on a BeckmanCoulter DU800 spectrophotomer. The extinction coefficient of lysozyme at 280 nm was taken to be E0.1% = [2.64 mg ml-1 cm-1].22 Conductivity and pH measurements were performed using an Orion 4-Star conductivity and pH meter with a DuraProbe conductivity cell and a RossSure-Flow pH electrode (Thermo Fisher Scientific, Waltham, MA).
2.2 Crystallization Lysozyme was dissolved in 200 mM Tris (titrated to pH 8.0 with HCl; σ = 9.90 mS/cm), washed three times in the same buffer in an Amicon Ultra-4 centrifugal filter device with a 3 kDa molecular
Axelbaum, 2014
7
weight cutoff (Millipore, Billerica, MA) and then concentrated to approximately 35 mg/ml. Crystals were grown using the hanging drop method in the EasyXtal 15-Well Tool (cat no. 132006; Qiagen, Valencia, CA), which is arranged with columns containing three wells. For each column, drops were made by mixing 10µl of protein solution in 200 mM Tris (pH 8.0) with 10 µl of the reservoir solution. This mixture was vortexed briefly and then a 5 µl drop was dispensed on each of the three crystallization supports in the column. The reservoir solutions (300 µl) were 60% (v/v) (R)-, (S)- or (RS)-MPD in water. A control with only water in the reservoir was also carried out. The crystallization trays were left at 4.0 ± 0.5 °C and inspected periodically by bright field microscopy with an AxioImager A1m microscope (Carl Zeiss, Göttingen, Germany). Crystals of roughly 200 µm in diameter grew in about 5 days with MPD; crystals of similar size took about two weeks to grow in the control. Crystals were harvested with mounted cryoloops (Hampton Research, Aliso Viejo, CA). No cryoprotectant was used, except for crystals grown in the control, which were dipped in Paratone N (Hampton Research, Aliso Viejo, CA) immediately before the diffraction measurements.
2.3 Data Collection and Structure refinement All X-ray diffraction data were recorded at beamline X6A (Brookhaven National Laboratory, National Synchrotron Light Source, Upton, NY, USA) between 13.5 and 15.1 keV. All data were recorded at 100 K using an ADSC Q210 CCD detector (Poway, CA, USA) at X6A. Data were indexed, integrated and scaled in HKL 2000.23 Lysozyme crystal structures were solved by molecular replacement using MOLREP24 and the model from the PDB entry 1IEE. Each model was refined by restrained maximum-likelihood refinement with REFMAC25 with individual anisotropic temperature factors and manual building performed in Coot.26 After the final refinement, stereochemistry of the structures was assessed with PROCHECK.27 All figures were prepared using Coot or PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC).
2.4 MPD analysis 2.4.1 Compilation of Structures Containing MPD Molecules MPD conformations were extracted from the PDB. The PDB was searched using the chemical IDs for either (R)-MPD (MRD) or (S)-MPD (MPD) together with two additional requirements: x-ray resolution between 0 and 1.5 Å and sequence homology of less than 90% with other macromolecules. This search yielded 49 protein structure hits for MRD and 89 protein structure hits for MPD. Some of these hits were labeled with both enantiomers yielding 117 unique protein structures with (R)- or (S)MPD.
Axelbaum, 2014
8
2.4.2 Analysis of PDB Structures First, we note that numerous protein structures had more than one (R)- or (S)-MPD molecule associated with them. These molecules were inspected using Coot with the structure and electron density maps (2Fo-Fc and Fo-Fc) downloaded through the Uppsala Electron Density Server.28 Hits that had no density or no structure factor were discarded. The quality of the electron density map, the local hydrogen bonding and the atomic B-factors of the molecules (described below) were used to check whether the assigned model (R)- or (S)-MPD structure was acceptable as is, i.e., whether the enantiomer and conformer selected were supported by the data. Acceptable structures were kept, while the unacceptable ones, those whose torsion angles could not be determined unambiguously, were either discarded or reassigned to achieve a better agreement between the model and the electron density. The torsion angles of the acceptable and reassigned structures were measured using the built-in function of Coot. In total, 221 molecules were retained: 109 were (R)-MPD and 112 were (S)-MPD. The electron density maps were used to analyze each assignment. The density had to support the assignment of MPD, as well as the assignment of a specific enantiomer and a specific conformation. Because of the similarity between the MPD enantiomers, many structures were ruled out because of insufficient density. This could have been due to a lack of density or to too much density that blurred the outline of the MPD molecule. If the density could not justify an exact assignment, the data could not be used. Good density was density that followed the outline of the MPD molecule without any missing or extra pieces to explain. But even with good density, the exact conformation could be determined only once the positions of the oxygen molecules were determined. To identify the oxygen atoms, we used three different pieces of data. The first was the actual density map. The root-mean-square deviation, or rmsd, value of the density map could be varied and if made high enough the density could be removed from the entire structure. As the rmsd value was increased, the weaker density was removed first and the stronger density remained until higher rmsd values. In general, the regions with the strongest density, and thus apparent at relatively higher rmsd values, should correspond to the position of the oxygen atoms. This was not an absolute rule, but it was used as one of the pieces of data to help guide our decision making. The second piece of data we used was the B-factor of the specific atoms. B-factor is a measure of the mean square displacement of an atom. We were able to look at the B-factors of each atom and compare their values. In general, the oxygen atoms should have the lowest relative B-factor values. Again, this was not an absolute rule, as there were examples where the B-factor of oxygen was higher than that of the nearby carbons, but a relatively high B-factor value for an oxygen atom or a large difference in B-factor values of nearby atoms was a useful piece of data to aid us in our decisions about specific assignments.
Axelbaum, 2014
9
Hydrogen bonding was also used to assess the MPD molecule’s assignment. Hydrogen bonds are moderately strong up to a distance of 3.2Å between the donor and acceptor atoms.29 This distance was measured using the built-in feature in Coot. The hydrogen bonding of the assignment was compared to the available hydrogen bonding of potential reassignments. This was particularly useful for reassignments of C5 and O4 (the oxygen atom connected to C4) where O4 was assigned to a position if it was the only available position for hydrogen bonding. Based on the above method of analysis we would carefully reassign molecules. The data was not always clear-cut and each molecule had to be assessed individually. This was a sensitive point because we needed to guarantee that we did not bias our study towards our expected conclusion through reassignments. In order to combat this bias, my lab partner, Dahniel Sastow, and I split up before our final decisions on the molecules. Individually we would decide to keep, discard, or reassign the molecules and then we compared our decisions as a group. Identical decisions were retained and points of divergence were discussed together until we decided whether or not the difficult structures were able to be kept or reassigned. In this way we were able to better recognize any bias and minimize its effect on our study.
3. Results and Discussion 3.1 Crystal Structures We obtained crystals using (R)-MPD , (S)-MPD, and (RS)-MPD solutions and were also able to crystallize the protein without any MPD additive. The crystals grown with MPD are shown below in figure 3. A
B
C
Figure 3. Crystals grown in A) (R)-MPD B) (S)-MPD and C) (RS)-MPD.
The lysozyme crystals all formed tetragonal crystals with space group P43212, regardless of whether or not any MPD was added. These findings are consistent with similar studies on the crystallization of lysozyme in Tris buffer at pH=8.00.30-34 The crystal structure data is found in table 4 in the appendix, but the most important pieces of information for this study are in table 1 below.
Axelbaum, 2014
Investigator
10
MPD Added
Maximum
Mosaicity (°)
Average B-factor (Å2, protein)
Resolution (Å) This paper
(R)-MPD
1.00
0.26
11.9
This paper
None
1.15
0.32
12.9
This paper
(S)-MPD
1.20
0.46
15.2
This paper
(RS)-MPD
1.25
0.43
13.7
Weiss et al.
(RS)-MPD
1.64
not reported
17.4
Michaux et al.
(RS)-MPD
1.75
not reported
14.6
Table 1. Summary of the crystal structure data from our experiments and the experiments of Weiss et al. and Michaux et al.
Note the higher resolution, B-factor, and mosaicity of the structures grown from pure (R)-MPD than the structures grown in pure (S)-MPD, the racemic (RS)-MPD, or no MPD at all. As mentioned in the introduction, resolution can be seen as a measure of the uniformity of the crystals because higher resolution is a product of higher uniformity in the crystal. A lower value corresponds to higher resolution, indicating a higher level of order in the crystals grown in (R)-MPD. Similarly, mosaicity is a measure of the order of the unit cells in a crystal. A perfectly ordered crystal has a mosaicity value of 0°.9 Additionally, B-factor is a measure of the mean-square deviation of the atoms in the structure, and a lower value corresponds to more uniform atoms throughout the crystal structure. All the data shows that there is a difference in the uniformity of the crystals grown in (R)-MPD from any of the other crystals. The (R)-MPD crystals grow more orderly and pack better than the other crystals. In order to understand this difference, we examined the protein structures to see if there are recognizable differences in the binding of the protein with the MPD.
3.2 Lysozyme-MPD Binding Sites The structure maps from the racemic and enanatiomerically pure crystals gave us the ability to analyze the MPD binding spots in lysozyme from a perspective that was never available before. We could now analyze the way that lysozyme interacts with each enantiomer when it is the only precipitant present as well as when both are present. From our structures we found three MPD-lysozyme interaction sites: tryptophan 63, tryptophan 123, and phenylalanine 34. These sites are shown in figure 4. The tryptophan 123 site is a new site discovered by our group, but the other two had been seen in previous structures.
Axelbaum, 2014
11
Figure 4. Overall structures of lysozyme with the enantiomers of MPD. A ribbon diagram of the Cα backbone is shown for crystals grown with (R)-MPD (blue) and (S)-MPD (green). The MPD molecules associated with each structure are shown in the same color as the protein backbone.
It is clear from our data, that (R)-MPD and (S)-MPD do not bind with lysozyme in the same way. The enantiomers each bind in one site where the other is not found, though they also share one binding site. This can explain why we find different degrees of order in the crystals grown with different enantiomers. However, it must still be explained why (R)-MPD produces more ordered crystals than crystals grown in (S)-MPD or (RS)-MPD. In order to explain this we will look at each binding site of MPD using the data from our crystals alongside the data from the structures published by the Michaux and Weiss groups, and study the interactions between the molecules and the protein. Differences in the enantiomers’ binding with lysozyme will be used to explain the better packing of crystals grown in (R)MPD. A table with the data from the MPD molecules at each interaction site (table 4) can be found in the appendix.
3.2.1 Tryptophan (W) 63 From our structures this site seemed the most straightforward for analysis. Both our (R)-MPD and (RS)-MPD structures showed very good density for an unambiguous assignment of (R)-MPD. The (R)MPD molecule makes hydrogen bonds with W63 and N59. Additionally, our (S)-MPD structure did not show any real density for an MPD molecule in the same site; this site was assigned with water molecules instead. Weiss et al. do not have enough density to support the assignment of an MPD molecule, and they do not assign one for this reason. However, our data seemed to be contradicted by the structure presented
Axelbaum, 2014
12
by the Michaux group. They assigned an (S)-MPD molecule that seemed to fit its density quite well. But once we analyzed their structure more thoroughly, we discovered an error in their assignment. While their assignment seemed to fit the density well, we noticed that the enantiomer that they had assigned occupied nearly the exact same position as our (R)-MPD molecule. The methyl group was in the same position as it was in our molecule and the intramolecular hydrogen bond was unbroken. We recognized that the only way this was possible was if the chiral end of the molecule was the inverted version of our molecule. Upon closer inspection of the needed chiral inversion of their molecule, we noticed that they had made a mistake in fitting their density. They had flipped their molecule so that the (S)-MPD molecule could make the same bond that our (R)-MPD molecule made, but in order to do so they had accidentally placed their C4 atom outside their electron density (figure 5b). The assignment of (S)-MPD was a mistake and in reality they too had an (R)-MPD molecule in their structure. A
B
Figure 5. W63 binding site of MPD from the Michaux structure. A) The (S)-MPD molecule assigned to the site seems to fit the density well and makes two hydrogen bonds (shown by the dashed lines) with the protein. B) The chiral end of the assigned molecule sticks outside the density as indicated by the white arrow. The chiral end should be inverted, thus making the molecule an (R)-MPD molecule.
It is thus clear that (R)-MPD is favored for the W63 binding site according to both our data and the data from the Michaux group.
3.2.2 Tryptophan (W) 123 This interaction site (figure 6) was discovered by our group and not seen in any previous structures. It was found only in our (S)-MPD structure and even then the density could only be seen at low rmsd values. However, the density could not be explained by water molecules and supports the assignment of an additional (S)-MPD molecule. At first glance, it is not clear what role this molecule
Axelbaum, 2014
13
plays in the stabilization of the protein. The (S)-MPD molecule is bound to water molecules that are bound to the symmetry-related protein and to other water molecules, but it has no direct hydrogen bonds with either protein molecules. However, it is important to note that the MPD backbone is favorable located near the hydrophobic aromatic rings of the tryptophan residue and could be acting as a bridge between the residue and the polar water molecules. This site is also intriguing because it is not clear why we do not find density for an MPD molecule in any of our other structures. An (R)-MPD molecule could seemingly fit the site just as well as an (S)-MPD molecule, yet we do not see any density in this site to justify an assignment of an (R)-MPD molecule. Additionally, it would seem difficult to understand why no MPD molecules are seen in this site when (RS)-MPD is used for crystallization-since there are (S)-MPD molecules in the racemic solution. However, this latter point is explained by the decreased concentration of each enantiomer in the solution that formed the (RS)-MPD crystals. The concentration of (S)-MPD molecules in the pure (S)-MPD crystals was 30% MPD (v/v), but in the (RS)-MPD solution there was only 15% (S)-MPD because the racemic solution is made of equal concentrations of each enantiomer. Since the molecule has a low occupancy at this site (0.5), the density is not very strong and a reduced concentration may make this density too weak to be detected.
Figure 6. An (S)-MPD molecule in the W123 interaction site.
Axelbaum, 2014
14
3.2.3 Phenylalanine (F) 34 This interaction site is the point of controversy between the Michaux and Weiss groups. It is a site where the MPD molecules make a crystal contact (it binds to two different protein molecules) through hydrogen bonding with F34 and G22ʹ. (The prime notation in G22ʹ refers to the fact that this residue is found in a symmetry-related protein molecule.) The Weiss group finds an (R)-MPD molecule while the Michaux group finds an (S)-MPD molecule (figure 7). Both groups used (RS)-MPD and in each case their density justifies its assignment. While the Michaux structure does show some additional density above C4 on the chiral end of the molecule, the methyl density (i.e. the density around CM) is stronger and thus justifies their assignment of (S)-MPD. A
B
Figure 7. F34 binding site. A) Structure found by Weiss group. B) Structure found by Michaux group. In both structures the MPD molecule makes intermolecular hydrogen bonds with F34 and G22ʹ.
Without any additional data, a preference for one enantiomer over the other could not be determined for this interaction site. However, with the data from our structures we are able to analyze the individual contributions of each enantiomer to the observed density in the (RS)-MPD structure. Our (R)-MPD shows well-defined MPD density at the F34 site. While the structure (figure 8A) shows a decent amount of density above C4 when the rmsd value for the electron density is low, this is indicative of rotation of the chiral end of the molecule and not of the presence of the second enantiomer. We know that this because we crystallized the protein in the presence of (R)-MPD molecules only.
Axelbaum, 2014
15
A
B1
C
B2
Figure 8. MPD molecules at the F34 binding site. A) (R)-MPD B1) (S)-MPD. There is extra density visible inside the white circle. The density indicates shared occupancy with H2O molecules. B2) (S)-MPD from a different perspective. The white arrow points at C3, where the density is absent. C) (RS)-MPD. The density fits an assignment of (R)-MPD.
However, the (S)-MPD structure (figures 8B1 and 8B2) is different from the (R)-MPD structure. The density is quite poor and does not immediately justify the assignment of an MPD molecule at all. It is only at an electron density rsmd value of 0.5σ that any density appears where C3 is meant to be. At that low of an rmsd value, the density can either imply that the molecule takes many different conformations inside the crystals or that the (S)-MPD molecule is not found in all of the crystal unit cells. In reality, both factors seem to play a role in this structure because the density has different characteristics that work better with each possibility. On one hand, the density around either end of the molecule is fairly strong and thus the molecule is likely present in a large portion of the crystal unit cells. This would imply that the missing density around C3 is from the molecule’s assuming multiple conformations within the crystal.
Axelbaum, 2014
16
Presumably, the methyl group on the (S)-MPD molecule destabilizes the interaction between the protein and the (S)-MPD molecule so that it must rotate to make the intermolecular bond. On the other hand, the (S)-MPD structure has density that is very close to the MPD density but is outside the boundaries of a normal MPD molecule. These cannot be explained by the (S)-MPD molecule, but can be explained by water molecules. Together, the (S)-MPD molecules and the water molecules bond with and stabilize the protein molecules so that it crystallizes. A figure showing the difference between the interactions (R)-MPD and (S)-MPD with lysozyme at this F34 interaction site is shown below. Note the limited density in the (S)-MPD structure which implies a reduced occupancy, which is due to rotation and the presence of water molecules in many of the unit cells. For this reason, the occupancy of the (R)-MPD molecule in this site was assigned as 1.0 while the occupancy of the (S)-MPD was assigned as 0.5 (see table 4 in the appendix).
Figure 9. Interaction of A) (R)-MPD and B) (S)-MPD with lysozyme at the F34 interaction site. The protein molecule is shown in green and the symmetry-related protein is shown in blue. The density in gray is contoured at 1.5σ.
Using this data, we can better analyze the (RS)-MPD site (figure 8c), which would otherwise be very difficult to understand. The density favors the assignment of (R)-MPD, but when the rmsd value is lowered there is extra density that appears above C4 on the chiral end of the molecule that could be explained either by the rotation of the molecule about the C3-C4 bond or by the presence of (S)-MPD molecules. However, since our data shows that (S)-MPD cannot explain the density of the backbone on its own, we have chosen the assignment of an (R)-MPD molecule to better explain the data. This assignment is also favored because the methyl density is stronger than the extra density over C4. The Michaux group, however, has methyl density that more strongly implies an (S)-MPD molecule. In order to best assign the (RS)-MPD structure of the Michaux group, it is important to again
Axelbaum, 2014
17
point out the contribution of each enantiomer. In our (RS)-MPD structure, we have extra density on C4 that indicates occupancy of the site by (S)-MPD molecules. But the overall density for the molecule in this interaction site has a significant density for the MPD backbone, which the pure (S)-MPD structure does not. This density must come from occupancy of site by (R)-MPD molecules, whose structure shows pronounced backbone density. Remember that the density of any structure is an average of the density of that site in every unit cell in the crystal. In an (RS)-MPD structure, if some proteins are bound to (R)MPD and some to (S)-MPD, the density of the MPD molecule seen in that site will resemble a hybrid of the conformations taken by the two molecules. Our crystallization of lysozyme with pure (S)-MPD shows that the (S)-MPD molecule cannot account for the density around C3 and thus the presence of density on the carbon backbone in the (RS)-MPD structure must be due to the presence of (R)-MPD molecule. When the average combines all of the different structures, we end up with a molecule that has both a defined backbone and density above C2 and C4. For this density, an assignment of an (S)-MPD molecule is not incorrect, as our structure also shows that (S)-MPD molecules are likely present in this interaction site, but it is misleading because we have shown that (S)-MPD cannot explain the density on its own. From our structures it is clear that an assignment of (R)-MPD better explains the density since it can account for both the backbone density and the methyl density above C2. The extra density above C4 is then explained by the rotation of the chiral end of the molecule about the C3-C4 bond. This rotation is seen in the structure from the crystals made with pure (R)-MPD. While an (S)-MPD molecule can also be assigned to explain this density, it must not replace the assignment of the (R)-MPD which can completely explain the density.
3.2.4 Summary of MPD Interaction Sites We used structures from crystals made with pure (R)-MPD or (S)-MPD to supplement the structures from crystals grown in (RS)-MPD to resolve the conflicts in the assignments of one MPD enantiomer over the other. Additionally, this method led to the discovery of a third, unknown MPD binding site. The interaction sites of MPD molecules with lysozyme are shown in figure 4. It is clear that only (R)-MPD is present in the W63 interaction site and only (S)-MPD is present in the W123 interaction site. Finally, (R)-MPD is the favored assignment for the F34 interaction site, though we have shown that the assignment of (S)-MPD, in addition to the (R)-MPD molecule, is also appropriate for structures grown from (RS)-MPD. Using this data, the differences in the packing of the crystals grown with the different solutions can be explained. The W63 binding site only binds with (R)-MPD molecules and the F34 site, where the MPD molecules make a crystal contact between protein molecules, favors the interaction with (R)-MPD
Axelbaum, 2014
18
molecules. These interactions apparently lead to better, tighter packing of the protein molecules that leads to higher resolution and lower mosaicity and B-factor values. (S)-MPD molecules bind with the protein at W123, but this site has a low occupancy value (0.5) and while this site may contribute to the packing of the protein, it does not benefit the crystal as much as the interaction of the (R)-MPD molecules. Additionally, (S)-MPD does not bind as well at the F34 crystal contact site and this gives the crystal less uniformity. In the (RS)-MPD crystal structure, the competition between the (R)- and (S)-MPD molecules leads to worse overall packing than if crystallized in the presence of pure (R)-MPD. Since (S)-MPD molecules occupy the F34 site in some of the protein interactions, this interferes with the bonding within the crystal and results in lower resolution. It is interesting to note that the crystals grown without any MPD additive have higher resolution than the crystals grown in (S)- or (RS)-MPD. This means that the (S)-MPD molecules actually stabilize the protein molecules less than the other molecules in the solution. It is only the pure (R)-MPD molecules that help give the protein molecules added stability.
3.3 MPD Conformational Analysis So far we have assigned the same conformer, where the hydroxyl groups are parallel to one another, to our MPD molecules in our structures. Additionally, we have seen the same conformation in the structures shown by the Michaux and Weiss group. However, we have mentioned that the MPD molecule can, and does, rotate thus giving the molecule a different conformation. Additionally, the MPD molecule in the (S)-MPD structure was assumed to occupy many different conformations so that it could maintain the crystal contact for the proteins while still maintaining the intramolecular hydrogen bond. In order to understand rotations of MPD molecules, we must analyze the conformations that MPD can assume.
3.3.1 MPD Conformations The C2-C3 and the C3-C4 single bonds allow for rotations that affect the overall conformation of the molecule. The dihedral angles, or torsion angles, made up by atoms C1-C2-C3-C4 and C2-C3-C4-C5 are named ψ1 and ψ2 respectively (figure 10).
Axelbaum, 2014
19
Figure 10. The torsion angles of (R)- and (S)-MPD.
If the consecutive substituents on the MPD backbone are to assume a stable staggered conformation, there are a limited number of conformations for the molecule. For simplicity, we will only consider these locally stable conformations. With each rotation between staggered conformations being approximately 120° this allows for three rotations per bond giving nine possible conformations. These are shown in figure 11. We named the conformers as follows: we chose 1a to refer to the conformation where the torsion angles are each 180° (the backbone of the MPD molecule has an M or W shape to it). This backbone conformation has been shown as favorable.35 Additionally, in this conformation, the C-O bonds are both parallel to one another which would allow for a strong intramolecular hydrogen bond. The other eight conformations are labelled in relation to this 1a conformation.
Axelbaum, 2014
20
Figure 11. The nine locally stable conformations of MPD. The conformation assumed to be the most stable is labeled 1a. Rotations clockwise about the C3-C2 angle, when looking down the bond from C3 to C2, change the number of the conformation. Rotations clockwise about the C4-C3 angle, when looking down the bond from C4 to C3, change the letter of the conformation.
3.3.2 Relative Free Energies To measure the strength of the intramolecular hydrogen bond, we studied the conformations of MPD molecules found in structures found in the Protein Data Bank (PDB). Our analysis started with 117 PDB structures containing MPD molecules (see materials and methods). During our analysis of the torsion angles, we began to recognize the difficulty in assigning specific MPD conformations.
Axelbaum, 2014
21
One of the difficulties that we faced with our current work, which we did not face before, is the similarity of the density of the enantiomers of MPD. Tartrate, the precipitant used by our group to crystallize thaumatin,14 was a chiral molecule with two chiral centers and the enantiomers were sufficiently different such that there was little ambiguity when it came to assignment. MPD is different from tatrate in that respect. The MPD molecule is fairly small and while the enantiomers can never take the exact same conformation in 3-dimensional space, the molecules can get very close to resembling one another. This was unacceptable for our study of the conformations because a misassignment of enantiomers would result in a mix-up of the values of ψ1 and ψ2. Note that assignment of chirality was not an issue for the structures of our crystals which were made with pure (R)-MPD or (S)-MPD because we knew that only one enantiomer was present. But for our analysis of our structure made with (RS)-MPD and for our analysis of MPD structures from the Protein Data Bank, where all structures were made with the racemic (RS)-MPD, the similarity of the MPD enantiomers presented quite a difficulty. Our analysis led us to recognize two signs that allowed us to distinguish between enantiomers so that we could confidently assign ψ1 and ψ2. These are signs that are necessary for the determination of chirality. The first piece of density we needed was the methyl density. MPD has a methyl group on its non-chiral end and this density must be seen to determine the chirality of the molecule. Without being able to determine the location of the methyl density, the density can be explained by either enantiomer. The determination of the methyl density would seem straightforward, but we discovered that quite often there was also density above C4 (figure 12). This is density that is less pronounced than the methyl density, but appears on the chiral end of the MPD molecule. This density complicated our analysis because it could have resulted from the presence of both chiral molecules or from rotation of the chiral end of the MPD molecule. This is because the density that we analyzed is not a snapshot of the protein structure, but rather represents an average of the individual protein orientations throughout the entire crystal. For instance, if exactly half of the unit cells had (R)-MPD bound to the protein and exactly half of the unit cells had (S)-MPD the density would have shown that with weak methyl density on both halves of the molecule. Additionally, if some of the molecules had the chiral end in one conformation and others had the chiral end in a different rotation, the density would have reflected this and we would have seen density that would correspond to both conformations.
Axelbaum, 2014
22
Figure 12. (S)-MPD molecule (yellow) with extra density above C4, shown inside the white circle. This density can be explained by rotation of the chiral end of the (S)-MPD molecule, around the C3-C4 bond, or by the presence of an (R)-MPD molecule (green). The oxygen atoms are pink for both molecules.
The second piece of density we needed was around chiral end of our MPD molecules. This also sounds straightforward, but more than just needing density around the chiral end of the molecule, we needed to be able to determine the exact shape of the chiral end (around C4). If the density around the chiral end wasn’t clear enough or wasn’t complete, we couldn’t definitively assign the chirality of the molecule. An inversion of any group bonded to C4, the chiral end of the molecule, results in the opposite enantiomer, so we had to exclude the possibility of this “chiral flip.” This was found in many of the structures analyzed for our Protein Data Bank analysis of MPD molecules, including the MPD molecule near the W63 interaction site assigned in the Michaux group’s lysozyme structure discussed in section 3.2.1.
Recognizing these difficulties, our group worked through the many MPD molecules contained in the 117 PDB structures and measured the torsion angles of the final assignments. We then binned the data from the torsion angles into nine bins, each corresponding to one conformation that the MPD molecules can take. These torsion angles and bins can be seen in figure 13 below. Notice the concentration of points in the bin for the 180°, 180° corresponding to the 1a conformation, which was presumed to be the preferred conformation.
Axelbaum, 2014
23
Figure 13. The torsion angles of MPD molecules found in the Protein Data Bank. Red squares correspond to (R)MPD molecules and blue squares to (S)-MPD molecules.
Using the number of data points in each bin we were able to calculate the relative Gibbs free energies:
Gi RT ln
Ni N 1a
Here N1a is the number of MPD molecules found in the 1a conformation and Ni is the number of MPD molecules found in any bin, i. The free energies are in the table 2.
Axelbaum, 2014 Conformer
24 G (kJ/mol) 300K
1a
0.00
2a
3.70
3a
4.71
3c
6.99
1b
7.33
1c
7.33
2c
7.33
2b
7.71
3b
8.72
Table 2. Relative free energies calculated from the conformations found in the Protein Data Bank.
From the table it is clear that there is a noticeable energy gap between the 1a conformation and the next lowest two lowest conformations, 2a and 3a. The energy difference is larger than kT, which at 300K is ~2.4kJ/mol. This energy gap nearly doubles when considering the difference between the 1a conformation and the rest of the conformations. This is an interesting result as 2a, and 3a are obtained by the rotation about the C2-C3 bond, which is a rotation of the non-chiral end of the molecule. Even though the intramolecular bond is broken, these rotations are more thermodynamically stable than rotations of the chiral end of the molecule. We also note that 3c, the next most stable conformer, corresponds to a conformation where the intramolecular hydrogen bond is still maintained. It is for steric reasons having to do with the carbon backbone that this conformation is relatively unfavored. However, it is more favorable than the remaining conformers where no intramolecular hydrogen bond is made. These findings are supported by quantum chemical calculations and molecular dynamics simulations by our collaborators. (Mark Stauber, Jean Jakoncic, Jacob Berger, Jerome Karp, Ariel Axelbaum, Dahniel Sastow, Sergey V. Buldyrev, Bruce J. Hrnjez and Neer Asherie; unpublished results.) From our study of the PDB it is clear that 1a is the favored conformer. This result is important because it shows that 1a is the favored conformer even in real solutions where many hydrogen bonds can be made to other molecules. This means that when crystallographers are considering the assumed
Axelbaum, 2014
25
conformation in which they should assign their MPD molecules, they should initially assume the 1a conformation, unless their data shows that the MPD molecule has taken a different conformation. This finding is also useful for our discussion about (S)-MPD in lysozyme. We mentioned in section 3.2.3 that (S)-MPD in the F34 interaction site has very little density for the backbone, but the ends of the molecule still have strong density. It seems that the (S)-MPD molecule rotates in space to accommodate its less desirable methyl group in order to maintain this intramolecular hydrogen bond. The molecule rotates, without changing its conformation, and this allows it to bond with the protein and make the crystal contact, but results in very little apparent density for C3. This result is also relevant to the extra density above C4 that we saw in our structure that was grown from pure (R)-MPD. We explained that this density was explained by rotation of the chiral end of the molecule, but our data indicates that rotations of the chiral end of the molecule around the C3-C4 bond would not be expected. If we did not know that our protein had been crystallized with only one enantiomer, it would be better to assume the presence of the second enantiomer instead of assuming rotation of the chiral end of the molecule. Another interesting point is the potential chiral effect that we might expect to see in the analysis of the PDB data. It is possible that one enantiomer would show a stronger preference for the 1a conformation when interacting with proteins in comparison to the other enantiomer. Any difference we see, however, is not statistically significant. This shows that while we have shown that chirality matters in the interaction of MPD with lysozyme, thermodynamically both enantiomers are equally stable.
4. Conclusion Our crystallization of lysozyme with pure (R)- and (S)-MPD, along with the racemic (RS)-MPD, allowed us to generalize our previous findings of a chiral effect in the crystallization of thaumatin and tartrate. (R)-MPD forms crystals with higher resolution and lower mosaicity and B-factor than crystals grown in (S)-MPD or (RS)-MPD. These are the result of a higher level of order in the packing of the proteins in the crystals grown in (R)-MPD. We showed that this packing is the result of the different interactions with each enantiomer in specific binding sites. Additionally, we were able to use our structures from crystals grown in enantiomerically pure MPD to analyze the individual contribution of each enantiomer to the electron density seen in the structure grown with the racemic MPD. This allowed us to resolve the inconsistency of the data from the previously published structures. These findings point
Axelbaum, 2014
26
to the importance of a recognition of the chirality of precipitants as an aid in crystallization and assignment in structures. Finally, we were able to use the conformations of MPD molecules in structures in the Protein Data Bank to measure the relative Gibbs free energy between the nine locally stable conformations of MPD. We established that 1a, the conformation where the intramolecular hydrogen bond is maintained while the carbon backbone takes an anti-anti conformation, is the favored conformation for MPD molecules. These findings again help in the interpretation of electron density and the assignment of MPD molecules in structures. The 1a conformation should be assumed for MPD unless the density confirms that another confirmation is favored.
5. Acknowledgments I’d like to thank Professor Neer Asherie for his continual guidance and mentorship. I would also like to thank the other members of the group, Dahniel Sastow, Mark Stauber, Jacob Berger, and Jerome Karp, as well as our collaborators, Jean Jakoncic, Sergey Buldyrev, and Bruce Hrnjez. I thank the National Science Foundation (DMR 1206416) for providing summer stipends and funding for a trip to the ACA conference in Hawaii where I was able to present my work as a poster and receive helpful feedback. Lastly, I thank Yeshiva University and the Jay and Jeanie Schottenstein Honors Program for their contributions to this project.
6. References 1. Fersht, A. (1999). Structure and mechanism in protein science: a guide to enzyme catalysis and protein folding. Macmillan. 2. Selkoe, D. J. (2004). Cell biology of protein misfolding: the examples of Alzheimer's and Parkinson's diseases. Nature cell biology, 6(11), 1054-1061. 3. Li, S. H., & Li, X. J. (2004). Huntingtin–protein interactions and the pathogenesis of Huntington's disease. TRENDS in Genetics, 20(3), 146-154. 4. Kendrew, J. C., Bodo, G., Dintzis, H. M., Parrish, R. G., Wyckoff, H., & Phillips, D. C. (1958). A three-dimensional model of the myoglobin molecule obtained by x-ray analysis. Nature, 181(4610), 662-666. 5. Blake, C. C. F., Koenig, D. F., Mair, G. A., North, A. C. T., Phillips, D. C., & Sarma, V. R. (1965). Structure of hen egg-white lysozyme: a three-dimensional fourier synthesis at 2 Å resolution. Nature, 206(4986), 757-761. 6. Raines, R. T. (1998). Ribonuclease a. Chemical reviews, 98(3), 1045-1066. Wyckoff, H. W., Hardman, K. D., Allewell, N. M., Inagami, T., Tsernoglou, D., Johnson, L. N., & Richards, F. M. (1967). The structure of ribonuclease-S at 6 A resolution. Journal of Biological Chemistry, 242(16), 3749-3753. 7. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., & Bourne, P. E. (2000). The protein data bank. Nucleic acids research, 28(1), 235-242.
Axelbaum, 2014
27
8. This number was determined by searching for structures that are labeled as protein and removing homologous structures with a 90% sequence similarity. URL (March 2014): http://www.rcsb.org/pdb/search/advSearch.do 9. McPherson, A. (2004). Introduction to protein crystallization. Methods, 34(3), 254-265. Dumetz, A. C., Chockla, A. M., Kaler, E. W. & Lenhoff, A. M. (2009). Cryst. Growth Des. 9, 682-691. 10. Gabanyi, M. J., Adams, P. D., Arnold, K., Bordoli, L., Carter, L. G., Flippen-Andersen, J., Gifford, L., Haas, J., Kouranov, A., McLaughlin, W. A., Micallef, D. I., Minor, W., Shah, R., Schwede, T., Tao, Y. P., Westbrook, J. D., Zimmerman, M. & Berman, H. M. (2011). J. Struct. Funct. Genomics 12, 45-54. 11. Retreived from URL (May 2013): http://targetdb.sbkb.org/statistics/TargetStatistics.html#general 12. Chayen, N. E. (2002). Trends Biotechnol. 20, 98-98. Chayen, N. E. & Saridakis, E. (2008). Nat. Methods 5, 147-153. 13. Pusey, M. L., Liu, Z. J., Tempel, W., Praissman, J., Lin, D. W., Wang, B. C., Gavira, J. A. & Ng, J. D. (2005). Prog. Biophys. Mol. Biol. 88, 359-386. 14. Asherie, N., Ginsberg, C., Blass, S., Greenbaum, A. & Knafo, S. (2008). Cryst. Growth Des. 8, 1815-1817. 15. Asherie, N., Ginsberg, C., Greenbaum, A., Blass, S. & Knafo, S. (2008). Cryst. Growth Des. 8, 4200-4207. 16. Asherie, N., Jakoncic, J., Ginsberg, C., Greenbaum, A., Stojanoff, V., Hrnjez, B. J., Blass, S. & Berger, J. (2009). Cryst. Growth Des. 9, 4189-4198. 17. Chayen, N. E. & Saridakis, E. (2001). J. Cryst. Growth 232, 262-264. Magay, E. & Yoon, T. S. (2011). J. Appl. Crystallogr. 44, 252-253. 18. Anand, K., Pal, D. & Hilgenfeld, R. (2002). Acta Crystallogr D 58, 1722-1728. 19. Sigma-Aldrich. Hexylene Glycol [Material Safety Data Sheet]. Retrieved from URL (May 2014): http://www.sigmaaldrich.com/MSDS/MSDS/DisplayMSDSPage.do?country=US&language=en& productNumber=112100&brand=ALDRICH&PageToGoToURL=http%3A%2F%2Fwww.sigma aldrich.com%2Fcatalog%2Fproduct%2Faldrich%2F112100%3Flang%3Den 20. Weiss, M. S., Palm, G. J. & Hilgenfeld, R. (2000). Acta Crystallogr D 56, 952-958. 21. Michaux, C., Pouyez, J., Wouters, J., & Privé, G. G. (2008). Protecting role of cosolvents in protein denaturation by SDS: a structural study. BMC structural biology, 8(1), 29. 22. Aune, K. C. & Tanford, C. (1969). Biochemistry 8, 4579-4585. 23. Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307-326. 24. Vagin, A. & Teplyakov, A. (1997). J. Appl. Crystallogr. 30, 1022-1025. 25. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Acta Crystallogr D 53, 240-255. 26. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Crystallogr D 66, 486-501. 27. Laskowski, R. A., Macarthur, M. W., Moss, D. S. & Thornton, J. M. (1993). J. Appl. Crystallogr. 26, 283-291. 28. Kleywegt, G. J., Harris, M. R., Zou, J. Y., Taylor, T. C., Wahlby, A. & Jones, T. A. (2004). Acta Crystallogr D 60, 2240-2249. 29. Jeffrey, G. A., & Jeffrey, G. A. (1997). An introduction to hydrogen bonding (Vol. 12). New York: Oxford University Press. 30. Bujacz, G., Wrzesniewska, B. & Bujacz, A. (2010). Acta Crystallogr D 66, 789-796. 31. Helliwell, J. R. & Tanley, S. W. M. (2013). Acta Crystallogr D 69, 121-125. 32. Michaux, C., Pouyez, J., Wouters, J., & Privé, G. G. (2008). Protecting role of cosolvents in protein denaturation by SDS: a structural study. BMC structural biology, 8(1), 29. 33. Weiss, M. S., Palm, G. J. & Hilgenfeld, R. (2000). Acta Crystallogr D 56, 952-958. 34. Judge, R. A., Jacobs, R. S., Frazier, T., Snell, E. H. & Pusey, M. L. (1999). Biophys. J. 77, 15851593. 35. Salam, A. & Deleuze, M. S. (2002). J. Chem. Phys. 116, 1296-1302.
Axelbaum, 2014
28
7. Appendix Accession Codes Our structures: Lysozyme without MPD, 4B49; lysozyme with (R)-MPD, 4B4E; lysozyme with (S)-MPD, 4B4I; lysozyme with (RS)-MPD, 4B4J. Michaux et al., 3B72; Weiss et al., 1DPW. (Both are lysozyme with (RS)-MPD.)
Table 3. Crystallographic data and refinement statistics. PDB entry
4B49
4B4E
4B4I
4B4J
MPD added
none
(R)
(S)
(RS)
Data Collection Space group
P43212
P43212
P43212
P43212
Unit cell dimensions (Å)
76.84
77.53
77.44
77.67
a = b, c (Å); (α = β = γ = 90°)
38.69
37.89
37.94
37.70
Resolution (Å) a
20.00-1.15
15.00-1.00
30.00-1.20
30.00-1.25
(1.17-1.15)
(1.02-1.00)
(1.22-1.20)
(1.27-1.25)
Total reflections
562743
853125
512028
456741
Unique reflectionsa
41718 (2040)
62397 (3076)
36597 (1806)
32449 (1584)
Rmerge (%)a
0.08 (0.76)
0.07 (0.82)
0.08 (0.79)
0.06 (0.56)
I/σIa
41.5 (3.2)
43.3 (2.5)
45.1 (5.6)
45.9 (5.5)
Completeness (%)a
99.8 (99.1)
99.3 (99.0)
99.7 (100.0)
99.8 (99.6)
Redundancya
13.5 (10.9)
13.7 (11.2)
14.0 (12.7)
14.1 (13.2)
Mosaicity (°)
0.32
0.26
0.46
0.43
B factor (Å2) (Overall)
15.5
13.4
17.2
15.1
VM (Å3 /Da) / Solvent content (%)
2.03 / 39.6
2.03 / 39.6
2.03 / 39.4
2.03 / 39.3
Axelbaum, 2014
29
Refinement Resolution (Å) a
18.67-1.15
13.30-1.00
21.49-1.20
27.05-1.25
(1.18-1.15)
(1.03-1.00)
(1.23-1.20)
(1.28-1.25)
Reflections [Rcryst +Rfree (5%)]
39425 + 2089
59136 + 3155
34682 + 1825
30732 + 1643
Rcryst / Rfreea
12.7 / 15.1 (22.2 /
12.4 / 14.4 (25.0 /
12.9 / 16.7 (18.8 /
13.05 / 15.57 (18.1 /
25.6)
25.7)
22.1)
22.7)
Protein (no. of residues) b
1001 (129)
1000 (129)
1000 (129)
1001 (129)
MPD c
—
16 (2; R)
16 (2; S)
16 (2; R)
Ions / Ligands (no. of molecules)
11 (4)
3 (3)
2 (2)
2 (2)
Water
267
196
188
162
Protein
12.9
11.9
15.2
13.7
MPD
—
12.3
21.4
13.1
Ions / Ligands
13.1
14.2
17.7
15.4
Water
26.2
21.6
28.3
24.6
Bond length (Å)
0.009
0.008
0.007
0.007
Bond angles (°)
1.349
1.306
1.214
1.272
Most favored (%)
90.3
90.3
88.5
87.6
Additionally favored (%)
9.7
9.7
11.5
12.4
Disallowed (%)
0.0
0.0
0.0
0.0
Number of atoms
B factors (Å2)
RMS deviation from ideal
Ramachandran plot
a
Number in parentheses refers to the highest resolution shell. Atom OXT (residue L129) was omitted during the refinement for structures 4B4E and 4B4J. c In parentheses: (no. of molecules in the structure; enantiomer). b
Axelbaum, 2014
30
Table 4. MPD molecule data from lysozyme-MPD interaction sites.
Interaction Site
Tryptophan 63
PDB ID: Molecule Label
Atom
B-factor
Average B- Occupancy
(Å )
factor (Å2)
C1
76.86
34.2275
C2
44.06
1
O2
21.81
1
CM
15.5
1
C3
31.84
1
C4
34.87
1
O4
23.53
1
C5
25.35
1
4B4E: A 1131 MRD
C1
10.81
(R)-MPD
C2
11.06
0.6
O2
9.94
0.6
CM
14.82
0.6
C3
13.35
0.6
C4
13.85
0.6
O4
13.75
0.6
C5
17.01
0.6
4B4J: A 1132 MRD
C1
12.47
(RS)-MPD
C2
12.17
0.5
O2
10.85
0.5
CM
15.87
0.5
C3
14.39
0.5
C4
15.67
0.5
O4
15.2
0.5
C5
18.86
0.5
MPD Additive
Label
3B72: A 1 MPD (RS)-MPD
2
13.0738
14.435
1
0.6
0.5
Axelbaum, 2014 Tryptophan 123
Phenylalanine 34
31 4B4I: A 1132 MPD
C1
20.53
(S)-MPD
C2
20.55
0.5
O2
23.35
0.5
CM
22.94
0.5
C3
21.15
0.5
C4
22.01
0.5
O4
25.32
0.5
C5
22.42
0.5
C1
13.44
C2
33.55
1
O2
25.46
1
CM
53.77
1
C3
22.42
1
C4
30.19
1
O4
30.14
1
C5
28.32
1
C1
21.48
C2
21.71
1
O2
21.81
1
CM
21.16
1
C3
21.16
1
C4
20.69
1
O4
20.32
1
C5
20.45
1
4B4E: A 1130 MRD
C1
12.35
(R)-MPD
C2
10.66
1
O2
10.76
1
CM
13.31
1
C3
11.14
1
C4
11.39
1
O4
10.53
1
C5
12.17
1
3B72: A 5 MPD (RS)-MPD
1DPW: A 400 MRD (RS)-MPD
22.2838
29.6613
21.0975
11.5388
0.5
1
1
1
Axelbaum, 2014
32
4B4I: A 1133 MPD (S)-MPD
C1
19.63
20.6286
C2
19.63
0.5
O2
20.45
0.5
CM
20.16
0.5
C3
20.02
0.5
C4
26.63
0.5
O4
17.88
0.5
C5 4B4J: A 1133 MRD (RS)-MPD
0.5
0.5
C1
11.38
11.6813
0.55
C2
10.92
0.55
O2
12.07
0.55
CM
13.73
0.55
C3
12.13
0.55
C4
11.37
0.55
O4
10.12
0.55
C5
11.73
0.55
1
The Effect of 2-methyl-2,4-pentanediol on the Crystal Structure of Lysozyme
Thesis Submitted in Partial Fulfillment of the Requirements of the Jay and Jeanie Schottenstein Honors Program
Yeshiva College Yeshiva University May 2014
Ariel Axelbaum Mentor: Prof. Neer Asherie, Departments of Physics and Biology
Axelbaum, 2014
2
Abstract Chiral control of crystallization has ample precedent in the small-molecule world, but relatively little work has been carried out to investigate the role of chirality in protein crystallization. In this study, lysozyme was crystallized in the presence of the chiral additive 2-methyl-2,4-pentanediol (MPD), using the (R) and (S) enantiomers as well as the racemic (RS) mixture. The crystals grown with (R)-MPD had the least disorder and produced the highest resolution protein structures. This result is consistent with the observation that for (R)- and (RS)-MPD, the crystal contacts are made by (R)-MPD demonstrating that there is preferential interaction between lysozyme and this enantiomer. These findings support the hypothesis that chiral interactions are important in protein crystallization. Additionally, to confirm the conformational assignment of MPD in our crystal structures, the conformations of MPD molecules in crystal structures from the Protein Data Bank were examined. Using the number of occurrences of each conformer, the relative energies of the conformers were calculated. It was shown that MPD favors the conformation that allows for an intramolecular hydrogen bond between its hydroxyl groups and maintains an anti-anti configuration for its carbon backbone, which corresponds to the conformer found in our crystals.
1. Introduction The field of protein crystallization is important for many different applications, including many areas of biomedical research. This is because proteins have many vital roles in cells and control most of the functions carried out in cells.1 Most enzymes are proteins and in order to comprehend the reactions that enzymes catalyze, a knowledge of the protein structure is important. Additionally, many diseases, including Alzheimer’s, Parkinson’s, and Huntington’s diseases, have been shown to result from changes in protein structure.2,3 For these reasons there is a large emphasis placed on the determination of protein tertiary structures. Protein structures are traditionally determined through the crystallization of proteins and diffracting x-rays by the crystals. The x-ray diffraction pattern can be used to determine the electron density map of the protein. The beginning of structure determination of protein crystals began in 1958 when John Kendrew determined the atomic structure of myoglobin.4 This achievement was soon followed by the crystallization and determination of the structures of lysozyme and ribonucleases A and S.5,6 Since then, many protein structures have been determined by x-ray diffraction using protein crystals. Currently, there are 30,832 unique protein structures in the Protein Data Bank.7,8 While protein structures can be
Axelbaum, 2014
3
determined by several different techniques, structures determined by x-ray diffraction of protein crystals are useful as they can have high resolution, even greater than 1Å. High resolution is very useful for crystallographers because it corresponds to a high level of detail available in the electron density map. To achieve high resolution structures, the protein crystals must be made of many, well-ordered unit cells. A unit cell is the smallest repeating unit in a crystal and the more ordered the packing of the unit cells, the higher the resolution of the crystal. Salt crystals, which are made of very small, tightly packed ions, make very ordered crystals that can diffract to very high resolutions. Proteins are larger macromolecules that make fewer, weaker bonds than salts and also lose many more degrees of freedom when forming crystals. These properties result in looser packing of the protein crystals that leads to some disorder in the basic repeating unit and ultimately to a limitation of the resolution. Protein crystallization is difficult. Even without considering the requirements for quality diffraction, the crystallization of a protein requires the determination of the specific precipitant or precipitants that will lead to the crystallization of that protein. The precipitant that works for one protein is in no way guaranteed to work for another protein. Proteins are made of long chains of amino acids, and predicting the additives needed for the crystallization of any protein is challenging. In some ways resembling alchemists, scientists mix proteins with a large range of chemicals in the hope of obtaining crystals. From experiments, scientists have discovered the usefulness of alcohols and salts for crystallization and they have become the most common crystallization additives.9 However, even the right additive usually requires a very specific set of conditions that include concentration, pH, and temperature. Protein crystallization continues to stump crystallographers and finding the exact set of conditions needed to form crystals—and, in particular, the conditions that will produce crystals that will diffract to high resolution—can require several months, if not years. Even with many advances in the field of protein crystallization, it is still estimated that fewer than three percent of proteins from cloned genes have had their structure determined.10,11 Growing diffraction quality crystals represents the largest challenge for crystallographers. Furthermore, as the field of protein crystallization has developed, the crystallization of additional proteins has become more difficult for crystallographers. This is because proteins that crystallize with multiple precipitants over a broad range of conditions often have already been crystallized and their structures determined, while proteins that crystallize over a more narrow set of conditions using one precipitant remain elusive.12 While the number of solved protein structures has continued to grow, this is more of an indication of the effort and money that is going into the crystallization of proteins rather than
Axelbaum, 2014
4
an increased success rate.13 Scientists are learning more and more about the crystallization process and, while there is no conclusive formula for the crystallization of proteins, scientists are constantly developing new strategies and new tools to aid in the crystallization process. As part of the search for new crystallization strategies, my group has been studying the effect of chiral precipitants on protein crystallization. Chiral molecules are molecules that have identical chemical formulas but differ in their three-dimensional structure: the atoms are in a different arrangement such that the molecules are mirror images of one another and are non-superimposable (figure 1). In a pair of chiral molecules, each molecule is called an enantiomer. While both enantiomers interact with non-chiral molecules in the same way, when reacting with another chiral molecule the enantiomers react differently. Nearly all amino acids, the building blocks of proteins, are chiral molecules and thus interact with chiral molecules differently. Our experiments were designed to see if we could find a chiral effect in the crystallization of proteins.
Figure 1. A chiral molecule and its mirror image.
In a previous study, my group worked with thaumatin and tartrate as our protein-precipitant pair and discovered that the chirality of the precipitant affected the time needed for the crystallization of thaumatin as well as the crystal habit of the protein crystals.14-16 Yet, however interesting our findings were with thaumatin and tartrate, it was possible that this chiral effect was limited to this one proteinprecipitant pair. In order to show the generality of this chiral effect, it needed to be demonstrated in other protein-precipitant pairs. For our study we decided investigate the interactions between the protein lysozyme and the precipitant MPD. Lysozyme is one of the most studied proteins in the field of protein crystallization.17 First crystallized in 1965 by David Chilton Phillips and his collaborators (PDB ID 1LYZ), lysozyme was the second protein, and the first enzyme, to have its structure determined.5 Due to its early structure
Axelbaum, 2014
5
determination, low cost, stability, and ease with which it can be crystallized, lysozyme has been used by researchers as a model protein to study many aspects of protein crystallization. Lysozyme is found in many different species and between all the various types of lysozyme there are 1,359 structures in the RSCB Protein Data Bank (PDB) [www.pdb.org]. Also, it has the highest resolution structure for a macromolecule with over 100 residues (0.65Å; PDB ID 2VB1). For these reasons we chose it for our study of the generality of the chiral effect. Similarly, one of the most commonly used precipitant additives is 2-methyl-2,4-pentanediol, which is commonly known as MPD. MPD is a chiral molecule with one chiral center at C4, and its two enantiomers are (R)-MPD and (S)-MPD (figure 2). The racemic form of MPD, (RS)-MPD, where both enantiomers are present in equal quantities, is the only form of MPD used until now for crystallization experiments.18 MPD is a ligand in 1158 structures in the Protein Data Bank as either (R)-MPD or (S)MPD. MPD is a diol with two hydroxyl groups that make it soluble in water (up to 118.2g/L at 20 °C)19 and also allow it can interact with the polar groups of the protein. Additionally, MPD has a carbon backbone that allows it to interact with hydrophobic groups. These interactions can drive the protein solution to supersaturation and cause the protein to crystallize. C
Figure 2. A) (R)-MPD B) (S)-MPD. In the pictures, the hydrogen atoms are not shown. Pink spheres represent the oxygen atoms and the other colored spheres represent carbon atoms. C) The chemical structure of MPD. Numbers and letters in blue are for labeling of the carbons (and do not indicate anything about the chemical formula).
Another reason that this protein-precipitant pair was chosen has to do with an inconsistency in the literature. Two groups crystallized lysozyme with MPD and analyzed the binding of MPD to lysozyme at phenylalanine 34, but each group reports a different enantiomer at this site. Weiss et al. report that (R)MPD is found at the site while Michaux et al. report (S)-MPD.20,21 Both assignments are supported by their data. With just these two structures, no preference for one enantiomer over the other could be established. This thesis will attempt to establish whether there is a preference of lysozyme for one
Axelbaum, 2014
6
enantiomer or the other. In order to do this, we crystallized lysozyme with pure solutions of each enantiomer of MPD, in addition to the racemic form. This allowed us to analyze the individual interactions of each enantiomer with the protein. This is useful knowledge in its own right, but it also allowed us to analyze the contribution each enantiomer has in the crystals formed in the racemic solution. Our studies led us into another area of investigation. Our structures, and many of the structures we looked at, made us realize the difficulty in assigning MPD molecules from the electron density obtained by the x-ray diffraction experiments. This is due both to the similarity between the enantiomers and the multiple conformations that MPD can take. As an additional check on our and others’ assignments we looked at the energetic differences between the various conformations that MPD can take.
2. Materials and Methods 2.1 Materials Lysozyme (cat. no. 2933, lot no. 36P9210) was purchased from Worthington Biochemical Corporation (Lakewood, NJ). (R)-MPD and (S)-MPD were synthesized by Reuter Chemische Apparatebau KG (Freiburg, Germany). (RS)-MPD (cat. no. 68340, lot no. 1345630) was purchased from Sigma-Aldrich (St. Louis, MO). Tris base (cat. no. BP512-500), sodium azide (cat. no. S227I-500), and hydrochloric acid (cat. no. A144S-500) were purchased from Fisher Scientific (Pittsburgh, PA). All materials were used without further purification. The purity of the protein was checked using highperformance liquid chromatography (HPLC) and quasielastic light scattering (QLS), as described elsewhere.15 Deionized water was obtained from an Integral 3 deionization system (Millipore, Billerica, MA). Solutions were filtered through a Nalgene disposable 0.22 µm filter unit (Nalge Nunc International, Rochester, NY) prior to use. Concentration measurements were carried out by UV-Vis extinction spectroscopy on a BeckmanCoulter DU800 spectrophotomer. The extinction coefficient of lysozyme at 280 nm was taken to be E0.1% = [2.64 mg ml-1 cm-1].22 Conductivity and pH measurements were performed using an Orion 4-Star conductivity and pH meter with a DuraProbe conductivity cell and a RossSure-Flow pH electrode (Thermo Fisher Scientific, Waltham, MA).
2.2 Crystallization Lysozyme was dissolved in 200 mM Tris (titrated to pH 8.0 with HCl; σ = 9.90 mS/cm), washed three times in the same buffer in an Amicon Ultra-4 centrifugal filter device with a 3 kDa molecular
Axelbaum, 2014
7
weight cutoff (Millipore, Billerica, MA) and then concentrated to approximately 35 mg/ml. Crystals were grown using the hanging drop method in the EasyXtal 15-Well Tool (cat no. 132006; Qiagen, Valencia, CA), which is arranged with columns containing three wells. For each column, drops were made by mixing 10µl of protein solution in 200 mM Tris (pH 8.0) with 10 µl of the reservoir solution. This mixture was vortexed briefly and then a 5 µl drop was dispensed on each of the three crystallization supports in the column. The reservoir solutions (300 µl) were 60% (v/v) (R)-, (S)- or (RS)-MPD in water. A control with only water in the reservoir was also carried out. The crystallization trays were left at 4.0 ± 0.5 °C and inspected periodically by bright field microscopy with an AxioImager A1m microscope (Carl Zeiss, Göttingen, Germany). Crystals of roughly 200 µm in diameter grew in about 5 days with MPD; crystals of similar size took about two weeks to grow in the control. Crystals were harvested with mounted cryoloops (Hampton Research, Aliso Viejo, CA). No cryoprotectant was used, except for crystals grown in the control, which were dipped in Paratone N (Hampton Research, Aliso Viejo, CA) immediately before the diffraction measurements.
2.3 Data Collection and Structure refinement All X-ray diffraction data were recorded at beamline X6A (Brookhaven National Laboratory, National Synchrotron Light Source, Upton, NY, USA) between 13.5 and 15.1 keV. All data were recorded at 100 K using an ADSC Q210 CCD detector (Poway, CA, USA) at X6A. Data were indexed, integrated and scaled in HKL 2000.23 Lysozyme crystal structures were solved by molecular replacement using MOLREP24 and the model from the PDB entry 1IEE. Each model was refined by restrained maximum-likelihood refinement with REFMAC25 with individual anisotropic temperature factors and manual building performed in Coot.26 After the final refinement, stereochemistry of the structures was assessed with PROCHECK.27 All figures were prepared using Coot or PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC).
2.4 MPD analysis 2.4.1 Compilation of Structures Containing MPD Molecules MPD conformations were extracted from the PDB. The PDB was searched using the chemical IDs for either (R)-MPD (MRD) or (S)-MPD (MPD) together with two additional requirements: x-ray resolution between 0 and 1.5 Å and sequence homology of less than 90% with other macromolecules. This search yielded 49 protein structure hits for MRD and 89 protein structure hits for MPD. Some of these hits were labeled with both enantiomers yielding 117 unique protein structures with (R)- or (S)MPD.
Axelbaum, 2014
8
2.4.2 Analysis of PDB Structures First, we note that numerous protein structures had more than one (R)- or (S)-MPD molecule associated with them. These molecules were inspected using Coot with the structure and electron density maps (2Fo-Fc and Fo-Fc) downloaded through the Uppsala Electron Density Server.28 Hits that had no density or no structure factor were discarded. The quality of the electron density map, the local hydrogen bonding and the atomic B-factors of the molecules (described below) were used to check whether the assigned model (R)- or (S)-MPD structure was acceptable as is, i.e., whether the enantiomer and conformer selected were supported by the data. Acceptable structures were kept, while the unacceptable ones, those whose torsion angles could not be determined unambiguously, were either discarded or reassigned to achieve a better agreement between the model and the electron density. The torsion angles of the acceptable and reassigned structures were measured using the built-in function of Coot. In total, 221 molecules were retained: 109 were (R)-MPD and 112 were (S)-MPD. The electron density maps were used to analyze each assignment. The density had to support the assignment of MPD, as well as the assignment of a specific enantiomer and a specific conformation. Because of the similarity between the MPD enantiomers, many structures were ruled out because of insufficient density. This could have been due to a lack of density or to too much density that blurred the outline of the MPD molecule. If the density could not justify an exact assignment, the data could not be used. Good density was density that followed the outline of the MPD molecule without any missing or extra pieces to explain. But even with good density, the exact conformation could be determined only once the positions of the oxygen molecules were determined. To identify the oxygen atoms, we used three different pieces of data. The first was the actual density map. The root-mean-square deviation, or rmsd, value of the density map could be varied and if made high enough the density could be removed from the entire structure. As the rmsd value was increased, the weaker density was removed first and the stronger density remained until higher rmsd values. In general, the regions with the strongest density, and thus apparent at relatively higher rmsd values, should correspond to the position of the oxygen atoms. This was not an absolute rule, but it was used as one of the pieces of data to help guide our decision making. The second piece of data we used was the B-factor of the specific atoms. B-factor is a measure of the mean square displacement of an atom. We were able to look at the B-factors of each atom and compare their values. In general, the oxygen atoms should have the lowest relative B-factor values. Again, this was not an absolute rule, as there were examples where the B-factor of oxygen was higher than that of the nearby carbons, but a relatively high B-factor value for an oxygen atom or a large difference in B-factor values of nearby atoms was a useful piece of data to aid us in our decisions about specific assignments.
Axelbaum, 2014
9
Hydrogen bonding was also used to assess the MPD molecule’s assignment. Hydrogen bonds are moderately strong up to a distance of 3.2Å between the donor and acceptor atoms.29 This distance was measured using the built-in feature in Coot. The hydrogen bonding of the assignment was compared to the available hydrogen bonding of potential reassignments. This was particularly useful for reassignments of C5 and O4 (the oxygen atom connected to C4) where O4 was assigned to a position if it was the only available position for hydrogen bonding. Based on the above method of analysis we would carefully reassign molecules. The data was not always clear-cut and each molecule had to be assessed individually. This was a sensitive point because we needed to guarantee that we did not bias our study towards our expected conclusion through reassignments. In order to combat this bias, my lab partner, Dahniel Sastow, and I split up before our final decisions on the molecules. Individually we would decide to keep, discard, or reassign the molecules and then we compared our decisions as a group. Identical decisions were retained and points of divergence were discussed together until we decided whether or not the difficult structures were able to be kept or reassigned. In this way we were able to better recognize any bias and minimize its effect on our study.
3. Results and Discussion 3.1 Crystal Structures We obtained crystals using (R)-MPD , (S)-MPD, and (RS)-MPD solutions and were also able to crystallize the protein without any MPD additive. The crystals grown with MPD are shown below in figure 3. A
B
C
Figure 3. Crystals grown in A) (R)-MPD B) (S)-MPD and C) (RS)-MPD.
The lysozyme crystals all formed tetragonal crystals with space group P43212, regardless of whether or not any MPD was added. These findings are consistent with similar studies on the crystallization of lysozyme in Tris buffer at pH=8.00.30-34 The crystal structure data is found in table 4 in the appendix, but the most important pieces of information for this study are in table 1 below.
Axelbaum, 2014
Investigator
10
MPD Added
Maximum
Mosaicity (°)
Average B-factor (Å2, protein)
Resolution (Å) This paper
(R)-MPD
1.00
0.26
11.9
This paper
None
1.15
0.32
12.9
This paper
(S)-MPD
1.20
0.46
15.2
This paper
(RS)-MPD
1.25
0.43
13.7
Weiss et al.
(RS)-MPD
1.64
not reported
17.4
Michaux et al.
(RS)-MPD
1.75
not reported
14.6
Table 1. Summary of the crystal structure data from our experiments and the experiments of Weiss et al. and Michaux et al.
Note the higher resolution, B-factor, and mosaicity of the structures grown from pure (R)-MPD than the structures grown in pure (S)-MPD, the racemic (RS)-MPD, or no MPD at all. As mentioned in the introduction, resolution can be seen as a measure of the uniformity of the crystals because higher resolution is a product of higher uniformity in the crystal. A lower value corresponds to higher resolution, indicating a higher level of order in the crystals grown in (R)-MPD. Similarly, mosaicity is a measure of the order of the unit cells in a crystal. A perfectly ordered crystal has a mosaicity value of 0°.9 Additionally, B-factor is a measure of the mean-square deviation of the atoms in the structure, and a lower value corresponds to more uniform atoms throughout the crystal structure. All the data shows that there is a difference in the uniformity of the crystals grown in (R)-MPD from any of the other crystals. The (R)-MPD crystals grow more orderly and pack better than the other crystals. In order to understand this difference, we examined the protein structures to see if there are recognizable differences in the binding of the protein with the MPD.
3.2 Lysozyme-MPD Binding Sites The structure maps from the racemic and enanatiomerically pure crystals gave us the ability to analyze the MPD binding spots in lysozyme from a perspective that was never available before. We could now analyze the way that lysozyme interacts with each enantiomer when it is the only precipitant present as well as when both are present. From our structures we found three MPD-lysozyme interaction sites: tryptophan 63, tryptophan 123, and phenylalanine 34. These sites are shown in figure 4. The tryptophan 123 site is a new site discovered by our group, but the other two had been seen in previous structures.
Axelbaum, 2014
11
Figure 4. Overall structures of lysozyme with the enantiomers of MPD. A ribbon diagram of the Cα backbone is shown for crystals grown with (R)-MPD (blue) and (S)-MPD (green). The MPD molecules associated with each structure are shown in the same color as the protein backbone.
It is clear from our data, that (R)-MPD and (S)-MPD do not bind with lysozyme in the same way. The enantiomers each bind in one site where the other is not found, though they also share one binding site. This can explain why we find different degrees of order in the crystals grown with different enantiomers. However, it must still be explained why (R)-MPD produces more ordered crystals than crystals grown in (S)-MPD or (RS)-MPD. In order to explain this we will look at each binding site of MPD using the data from our crystals alongside the data from the structures published by the Michaux and Weiss groups, and study the interactions between the molecules and the protein. Differences in the enantiomers’ binding with lysozyme will be used to explain the better packing of crystals grown in (R)MPD. A table with the data from the MPD molecules at each interaction site (table 4) can be found in the appendix.
3.2.1 Tryptophan (W) 63 From our structures this site seemed the most straightforward for analysis. Both our (R)-MPD and (RS)-MPD structures showed very good density for an unambiguous assignment of (R)-MPD. The (R)MPD molecule makes hydrogen bonds with W63 and N59. Additionally, our (S)-MPD structure did not show any real density for an MPD molecule in the same site; this site was assigned with water molecules instead. Weiss et al. do not have enough density to support the assignment of an MPD molecule, and they do not assign one for this reason. However, our data seemed to be contradicted by the structure presented
Axelbaum, 2014
12
by the Michaux group. They assigned an (S)-MPD molecule that seemed to fit its density quite well. But once we analyzed their structure more thoroughly, we discovered an error in their assignment. While their assignment seemed to fit the density well, we noticed that the enantiomer that they had assigned occupied nearly the exact same position as our (R)-MPD molecule. The methyl group was in the same position as it was in our molecule and the intramolecular hydrogen bond was unbroken. We recognized that the only way this was possible was if the chiral end of the molecule was the inverted version of our molecule. Upon closer inspection of the needed chiral inversion of their molecule, we noticed that they had made a mistake in fitting their density. They had flipped their molecule so that the (S)-MPD molecule could make the same bond that our (R)-MPD molecule made, but in order to do so they had accidentally placed their C4 atom outside their electron density (figure 5b). The assignment of (S)-MPD was a mistake and in reality they too had an (R)-MPD molecule in their structure. A
B
Figure 5. W63 binding site of MPD from the Michaux structure. A) The (S)-MPD molecule assigned to the site seems to fit the density well and makes two hydrogen bonds (shown by the dashed lines) with the protein. B) The chiral end of the assigned molecule sticks outside the density as indicated by the white arrow. The chiral end should be inverted, thus making the molecule an (R)-MPD molecule.
It is thus clear that (R)-MPD is favored for the W63 binding site according to both our data and the data from the Michaux group.
3.2.2 Tryptophan (W) 123 This interaction site (figure 6) was discovered by our group and not seen in any previous structures. It was found only in our (S)-MPD structure and even then the density could only be seen at low rmsd values. However, the density could not be explained by water molecules and supports the assignment of an additional (S)-MPD molecule. At first glance, it is not clear what role this molecule
Axelbaum, 2014
13
plays in the stabilization of the protein. The (S)-MPD molecule is bound to water molecules that are bound to the symmetry-related protein and to other water molecules, but it has no direct hydrogen bonds with either protein molecules. However, it is important to note that the MPD backbone is favorable located near the hydrophobic aromatic rings of the tryptophan residue and could be acting as a bridge between the residue and the polar water molecules. This site is also intriguing because it is not clear why we do not find density for an MPD molecule in any of our other structures. An (R)-MPD molecule could seemingly fit the site just as well as an (S)-MPD molecule, yet we do not see any density in this site to justify an assignment of an (R)-MPD molecule. Additionally, it would seem difficult to understand why no MPD molecules are seen in this site when (RS)-MPD is used for crystallization-since there are (S)-MPD molecules in the racemic solution. However, this latter point is explained by the decreased concentration of each enantiomer in the solution that formed the (RS)-MPD crystals. The concentration of (S)-MPD molecules in the pure (S)-MPD crystals was 30% MPD (v/v), but in the (RS)-MPD solution there was only 15% (S)-MPD because the racemic solution is made of equal concentrations of each enantiomer. Since the molecule has a low occupancy at this site (0.5), the density is not very strong and a reduced concentration may make this density too weak to be detected.
Figure 6. An (S)-MPD molecule in the W123 interaction site.
Axelbaum, 2014
14
3.2.3 Phenylalanine (F) 34 This interaction site is the point of controversy between the Michaux and Weiss groups. It is a site where the MPD molecules make a crystal contact (it binds to two different protein molecules) through hydrogen bonding with F34 and G22ʹ. (The prime notation in G22ʹ refers to the fact that this residue is found in a symmetry-related protein molecule.) The Weiss group finds an (R)-MPD molecule while the Michaux group finds an (S)-MPD molecule (figure 7). Both groups used (RS)-MPD and in each case their density justifies its assignment. While the Michaux structure does show some additional density above C4 on the chiral end of the molecule, the methyl density (i.e. the density around CM) is stronger and thus justifies their assignment of (S)-MPD. A
B
Figure 7. F34 binding site. A) Structure found by Weiss group. B) Structure found by Michaux group. In both structures the MPD molecule makes intermolecular hydrogen bonds with F34 and G22ʹ.
Without any additional data, a preference for one enantiomer over the other could not be determined for this interaction site. However, with the data from our structures we are able to analyze the individual contributions of each enantiomer to the observed density in the (RS)-MPD structure. Our (R)-MPD shows well-defined MPD density at the F34 site. While the structure (figure 8A) shows a decent amount of density above C4 when the rmsd value for the electron density is low, this is indicative of rotation of the chiral end of the molecule and not of the presence of the second enantiomer. We know that this because we crystallized the protein in the presence of (R)-MPD molecules only.
Axelbaum, 2014
15
A
B1
C
B2
Figure 8. MPD molecules at the F34 binding site. A) (R)-MPD B1) (S)-MPD. There is extra density visible inside the white circle. The density indicates shared occupancy with H2O molecules. B2) (S)-MPD from a different perspective. The white arrow points at C3, where the density is absent. C) (RS)-MPD. The density fits an assignment of (R)-MPD.
However, the (S)-MPD structure (figures 8B1 and 8B2) is different from the (R)-MPD structure. The density is quite poor and does not immediately justify the assignment of an MPD molecule at all. It is only at an electron density rsmd value of 0.5σ that any density appears where C3 is meant to be. At that low of an rmsd value, the density can either imply that the molecule takes many different conformations inside the crystals or that the (S)-MPD molecule is not found in all of the crystal unit cells. In reality, both factors seem to play a role in this structure because the density has different characteristics that work better with each possibility. On one hand, the density around either end of the molecule is fairly strong and thus the molecule is likely present in a large portion of the crystal unit cells. This would imply that the missing density around C3 is from the molecule’s assuming multiple conformations within the crystal.
Axelbaum, 2014
16
Presumably, the methyl group on the (S)-MPD molecule destabilizes the interaction between the protein and the (S)-MPD molecule so that it must rotate to make the intermolecular bond. On the other hand, the (S)-MPD structure has density that is very close to the MPD density but is outside the boundaries of a normal MPD molecule. These cannot be explained by the (S)-MPD molecule, but can be explained by water molecules. Together, the (S)-MPD molecules and the water molecules bond with and stabilize the protein molecules so that it crystallizes. A figure showing the difference between the interactions (R)-MPD and (S)-MPD with lysozyme at this F34 interaction site is shown below. Note the limited density in the (S)-MPD structure which implies a reduced occupancy, which is due to rotation and the presence of water molecules in many of the unit cells. For this reason, the occupancy of the (R)-MPD molecule in this site was assigned as 1.0 while the occupancy of the (S)-MPD was assigned as 0.5 (see table 4 in the appendix).
Figure 9. Interaction of A) (R)-MPD and B) (S)-MPD with lysozyme at the F34 interaction site. The protein molecule is shown in green and the symmetry-related protein is shown in blue. The density in gray is contoured at 1.5σ.
Using this data, we can better analyze the (RS)-MPD site (figure 8c), which would otherwise be very difficult to understand. The density favors the assignment of (R)-MPD, but when the rmsd value is lowered there is extra density that appears above C4 on the chiral end of the molecule that could be explained either by the rotation of the molecule about the C3-C4 bond or by the presence of (S)-MPD molecules. However, since our data shows that (S)-MPD cannot explain the density of the backbone on its own, we have chosen the assignment of an (R)-MPD molecule to better explain the data. This assignment is also favored because the methyl density is stronger than the extra density over C4. The Michaux group, however, has methyl density that more strongly implies an (S)-MPD molecule. In order to best assign the (RS)-MPD structure of the Michaux group, it is important to again
Axelbaum, 2014
17
point out the contribution of each enantiomer. In our (RS)-MPD structure, we have extra density on C4 that indicates occupancy of the site by (S)-MPD molecules. But the overall density for the molecule in this interaction site has a significant density for the MPD backbone, which the pure (S)-MPD structure does not. This density must come from occupancy of site by (R)-MPD molecules, whose structure shows pronounced backbone density. Remember that the density of any structure is an average of the density of that site in every unit cell in the crystal. In an (RS)-MPD structure, if some proteins are bound to (R)MPD and some to (S)-MPD, the density of the MPD molecule seen in that site will resemble a hybrid of the conformations taken by the two molecules. Our crystallization of lysozyme with pure (S)-MPD shows that the (S)-MPD molecule cannot account for the density around C3 and thus the presence of density on the carbon backbone in the (RS)-MPD structure must be due to the presence of (R)-MPD molecule. When the average combines all of the different structures, we end up with a molecule that has both a defined backbone and density above C2 and C4. For this density, an assignment of an (S)-MPD molecule is not incorrect, as our structure also shows that (S)-MPD molecules are likely present in this interaction site, but it is misleading because we have shown that (S)-MPD cannot explain the density on its own. From our structures it is clear that an assignment of (R)-MPD better explains the density since it can account for both the backbone density and the methyl density above C2. The extra density above C4 is then explained by the rotation of the chiral end of the molecule about the C3-C4 bond. This rotation is seen in the structure from the crystals made with pure (R)-MPD. While an (S)-MPD molecule can also be assigned to explain this density, it must not replace the assignment of the (R)-MPD which can completely explain the density.
3.2.4 Summary of MPD Interaction Sites We used structures from crystals made with pure (R)-MPD or (S)-MPD to supplement the structures from crystals grown in (RS)-MPD to resolve the conflicts in the assignments of one MPD enantiomer over the other. Additionally, this method led to the discovery of a third, unknown MPD binding site. The interaction sites of MPD molecules with lysozyme are shown in figure 4. It is clear that only (R)-MPD is present in the W63 interaction site and only (S)-MPD is present in the W123 interaction site. Finally, (R)-MPD is the favored assignment for the F34 interaction site, though we have shown that the assignment of (S)-MPD, in addition to the (R)-MPD molecule, is also appropriate for structures grown from (RS)-MPD. Using this data, the differences in the packing of the crystals grown with the different solutions can be explained. The W63 binding site only binds with (R)-MPD molecules and the F34 site, where the MPD molecules make a crystal contact between protein molecules, favors the interaction with (R)-MPD
Axelbaum, 2014
18
molecules. These interactions apparently lead to better, tighter packing of the protein molecules that leads to higher resolution and lower mosaicity and B-factor values. (S)-MPD molecules bind with the protein at W123, but this site has a low occupancy value (0.5) and while this site may contribute to the packing of the protein, it does not benefit the crystal as much as the interaction of the (R)-MPD molecules. Additionally, (S)-MPD does not bind as well at the F34 crystal contact site and this gives the crystal less uniformity. In the (RS)-MPD crystal structure, the competition between the (R)- and (S)-MPD molecules leads to worse overall packing than if crystallized in the presence of pure (R)-MPD. Since (S)-MPD molecules occupy the F34 site in some of the protein interactions, this interferes with the bonding within the crystal and results in lower resolution. It is interesting to note that the crystals grown without any MPD additive have higher resolution than the crystals grown in (S)- or (RS)-MPD. This means that the (S)-MPD molecules actually stabilize the protein molecules less than the other molecules in the solution. It is only the pure (R)-MPD molecules that help give the protein molecules added stability.
3.3 MPD Conformational Analysis So far we have assigned the same conformer, where the hydroxyl groups are parallel to one another, to our MPD molecules in our structures. Additionally, we have seen the same conformation in the structures shown by the Michaux and Weiss group. However, we have mentioned that the MPD molecule can, and does, rotate thus giving the molecule a different conformation. Additionally, the MPD molecule in the (S)-MPD structure was assumed to occupy many different conformations so that it could maintain the crystal contact for the proteins while still maintaining the intramolecular hydrogen bond. In order to understand rotations of MPD molecules, we must analyze the conformations that MPD can assume.
3.3.1 MPD Conformations The C2-C3 and the C3-C4 single bonds allow for rotations that affect the overall conformation of the molecule. The dihedral angles, or torsion angles, made up by atoms C1-C2-C3-C4 and C2-C3-C4-C5 are named ψ1 and ψ2 respectively (figure 10).
Axelbaum, 2014
19
Figure 10. The torsion angles of (R)- and (S)-MPD.
If the consecutive substituents on the MPD backbone are to assume a stable staggered conformation, there are a limited number of conformations for the molecule. For simplicity, we will only consider these locally stable conformations. With each rotation between staggered conformations being approximately 120° this allows for three rotations per bond giving nine possible conformations. These are shown in figure 11. We named the conformers as follows: we chose 1a to refer to the conformation where the torsion angles are each 180° (the backbone of the MPD molecule has an M or W shape to it). This backbone conformation has been shown as favorable.35 Additionally, in this conformation, the C-O bonds are both parallel to one another which would allow for a strong intramolecular hydrogen bond. The other eight conformations are labelled in relation to this 1a conformation.
Axelbaum, 2014
20
Figure 11. The nine locally stable conformations of MPD. The conformation assumed to be the most stable is labeled 1a. Rotations clockwise about the C3-C2 angle, when looking down the bond from C3 to C2, change the number of the conformation. Rotations clockwise about the C4-C3 angle, when looking down the bond from C4 to C3, change the letter of the conformation.
3.3.2 Relative Free Energies To measure the strength of the intramolecular hydrogen bond, we studied the conformations of MPD molecules found in structures found in the Protein Data Bank (PDB). Our analysis started with 117 PDB structures containing MPD molecules (see materials and methods). During our analysis of the torsion angles, we began to recognize the difficulty in assigning specific MPD conformations.
Axelbaum, 2014
21
One of the difficulties that we faced with our current work, which we did not face before, is the similarity of the density of the enantiomers of MPD. Tartrate, the precipitant used by our group to crystallize thaumatin,14 was a chiral molecule with two chiral centers and the enantiomers were sufficiently different such that there was little ambiguity when it came to assignment. MPD is different from tatrate in that respect. The MPD molecule is fairly small and while the enantiomers can never take the exact same conformation in 3-dimensional space, the molecules can get very close to resembling one another. This was unacceptable for our study of the conformations because a misassignment of enantiomers would result in a mix-up of the values of ψ1 and ψ2. Note that assignment of chirality was not an issue for the structures of our crystals which were made with pure (R)-MPD or (S)-MPD because we knew that only one enantiomer was present. But for our analysis of our structure made with (RS)-MPD and for our analysis of MPD structures from the Protein Data Bank, where all structures were made with the racemic (RS)-MPD, the similarity of the MPD enantiomers presented quite a difficulty. Our analysis led us to recognize two signs that allowed us to distinguish between enantiomers so that we could confidently assign ψ1 and ψ2. These are signs that are necessary for the determination of chirality. The first piece of density we needed was the methyl density. MPD has a methyl group on its non-chiral end and this density must be seen to determine the chirality of the molecule. Without being able to determine the location of the methyl density, the density can be explained by either enantiomer. The determination of the methyl density would seem straightforward, but we discovered that quite often there was also density above C4 (figure 12). This is density that is less pronounced than the methyl density, but appears on the chiral end of the MPD molecule. This density complicated our analysis because it could have resulted from the presence of both chiral molecules or from rotation of the chiral end of the MPD molecule. This is because the density that we analyzed is not a snapshot of the protein structure, but rather represents an average of the individual protein orientations throughout the entire crystal. For instance, if exactly half of the unit cells had (R)-MPD bound to the protein and exactly half of the unit cells had (S)-MPD the density would have shown that with weak methyl density on both halves of the molecule. Additionally, if some of the molecules had the chiral end in one conformation and others had the chiral end in a different rotation, the density would have reflected this and we would have seen density that would correspond to both conformations.
Axelbaum, 2014
22
Figure 12. (S)-MPD molecule (yellow) with extra density above C4, shown inside the white circle. This density can be explained by rotation of the chiral end of the (S)-MPD molecule, around the C3-C4 bond, or by the presence of an (R)-MPD molecule (green). The oxygen atoms are pink for both molecules.
The second piece of density we needed was around chiral end of our MPD molecules. This also sounds straightforward, but more than just needing density around the chiral end of the molecule, we needed to be able to determine the exact shape of the chiral end (around C4). If the density around the chiral end wasn’t clear enough or wasn’t complete, we couldn’t definitively assign the chirality of the molecule. An inversion of any group bonded to C4, the chiral end of the molecule, results in the opposite enantiomer, so we had to exclude the possibility of this “chiral flip.” This was found in many of the structures analyzed for our Protein Data Bank analysis of MPD molecules, including the MPD molecule near the W63 interaction site assigned in the Michaux group’s lysozyme structure discussed in section 3.2.1.
Recognizing these difficulties, our group worked through the many MPD molecules contained in the 117 PDB structures and measured the torsion angles of the final assignments. We then binned the data from the torsion angles into nine bins, each corresponding to one conformation that the MPD molecules can take. These torsion angles and bins can be seen in figure 13 below. Notice the concentration of points in the bin for the 180°, 180° corresponding to the 1a conformation, which was presumed to be the preferred conformation.
Axelbaum, 2014
23
Figure 13. The torsion angles of MPD molecules found in the Protein Data Bank. Red squares correspond to (R)MPD molecules and blue squares to (S)-MPD molecules.
Using the number of data points in each bin we were able to calculate the relative Gibbs free energies:
Gi RT ln
Ni N 1a
Here N1a is the number of MPD molecules found in the 1a conformation and Ni is the number of MPD molecules found in any bin, i. The free energies are in the table 2.
Axelbaum, 2014 Conformer
24 G (kJ/mol) 300K
1a
0.00
2a
3.70
3a
4.71
3c
6.99
1b
7.33
1c
7.33
2c
7.33
2b
7.71
3b
8.72
Table 2. Relative free energies calculated from the conformations found in the Protein Data Bank.
From the table it is clear that there is a noticeable energy gap between the 1a conformation and the next lowest two lowest conformations, 2a and 3a. The energy difference is larger than kT, which at 300K is ~2.4kJ/mol. This energy gap nearly doubles when considering the difference between the 1a conformation and the rest of the conformations. This is an interesting result as 2a, and 3a are obtained by the rotation about the C2-C3 bond, which is a rotation of the non-chiral end of the molecule. Even though the intramolecular bond is broken, these rotations are more thermodynamically stable than rotations of the chiral end of the molecule. We also note that 3c, the next most stable conformer, corresponds to a conformation where the intramolecular hydrogen bond is still maintained. It is for steric reasons having to do with the carbon backbone that this conformation is relatively unfavored. However, it is more favorable than the remaining conformers where no intramolecular hydrogen bond is made. These findings are supported by quantum chemical calculations and molecular dynamics simulations by our collaborators. (Mark Stauber, Jean Jakoncic, Jacob Berger, Jerome Karp, Ariel Axelbaum, Dahniel Sastow, Sergey V. Buldyrev, Bruce J. Hrnjez and Neer Asherie; unpublished results.) From our study of the PDB it is clear that 1a is the favored conformer. This result is important because it shows that 1a is the favored conformer even in real solutions where many hydrogen bonds can be made to other molecules. This means that when crystallographers are considering the assumed
Axelbaum, 2014
25
conformation in which they should assign their MPD molecules, they should initially assume the 1a conformation, unless their data shows that the MPD molecule has taken a different conformation. This finding is also useful for our discussion about (S)-MPD in lysozyme. We mentioned in section 3.2.3 that (S)-MPD in the F34 interaction site has very little density for the backbone, but the ends of the molecule still have strong density. It seems that the (S)-MPD molecule rotates in space to accommodate its less desirable methyl group in order to maintain this intramolecular hydrogen bond. The molecule rotates, without changing its conformation, and this allows it to bond with the protein and make the crystal contact, but results in very little apparent density for C3. This result is also relevant to the extra density above C4 that we saw in our structure that was grown from pure (R)-MPD. We explained that this density was explained by rotation of the chiral end of the molecule, but our data indicates that rotations of the chiral end of the molecule around the C3-C4 bond would not be expected. If we did not know that our protein had been crystallized with only one enantiomer, it would be better to assume the presence of the second enantiomer instead of assuming rotation of the chiral end of the molecule. Another interesting point is the potential chiral effect that we might expect to see in the analysis of the PDB data. It is possible that one enantiomer would show a stronger preference for the 1a conformation when interacting with proteins in comparison to the other enantiomer. Any difference we see, however, is not statistically significant. This shows that while we have shown that chirality matters in the interaction of MPD with lysozyme, thermodynamically both enantiomers are equally stable.
4. Conclusion Our crystallization of lysozyme with pure (R)- and (S)-MPD, along with the racemic (RS)-MPD, allowed us to generalize our previous findings of a chiral effect in the crystallization of thaumatin and tartrate. (R)-MPD forms crystals with higher resolution and lower mosaicity and B-factor than crystals grown in (S)-MPD or (RS)-MPD. These are the result of a higher level of order in the packing of the proteins in the crystals grown in (R)-MPD. We showed that this packing is the result of the different interactions with each enantiomer in specific binding sites. Additionally, we were able to use our structures from crystals grown in enantiomerically pure MPD to analyze the individual contribution of each enantiomer to the electron density seen in the structure grown with the racemic MPD. This allowed us to resolve the inconsistency of the data from the previously published structures. These findings point
Axelbaum, 2014
26
to the importance of a recognition of the chirality of precipitants as an aid in crystallization and assignment in structures. Finally, we were able to use the conformations of MPD molecules in structures in the Protein Data Bank to measure the relative Gibbs free energy between the nine locally stable conformations of MPD. We established that 1a, the conformation where the intramolecular hydrogen bond is maintained while the carbon backbone takes an anti-anti conformation, is the favored conformation for MPD molecules. These findings again help in the interpretation of electron density and the assignment of MPD molecules in structures. The 1a conformation should be assumed for MPD unless the density confirms that another confirmation is favored.
5. Acknowledgments I’d like to thank Professor Neer Asherie for his continual guidance and mentorship. I would also like to thank the other members of the group, Dahniel Sastow, Mark Stauber, Jacob Berger, and Jerome Karp, as well as our collaborators, Jean Jakoncic, Sergey Buldyrev, and Bruce Hrnjez. I thank the National Science Foundation (DMR 1206416) for providing summer stipends and funding for a trip to the ACA conference in Hawaii where I was able to present my work as a poster and receive helpful feedback. Lastly, I thank Yeshiva University and the Jay and Jeanie Schottenstein Honors Program for their contributions to this project.
6. References 1. Fersht, A. (1999). Structure and mechanism in protein science: a guide to enzyme catalysis and protein folding. Macmillan. 2. Selkoe, D. J. (2004). Cell biology of protein misfolding: the examples of Alzheimer's and Parkinson's diseases. Nature cell biology, 6(11), 1054-1061. 3. Li, S. H., & Li, X. J. (2004). Huntingtin–protein interactions and the pathogenesis of Huntington's disease. TRENDS in Genetics, 20(3), 146-154. 4. Kendrew, J. C., Bodo, G., Dintzis, H. M., Parrish, R. G., Wyckoff, H., & Phillips, D. C. (1958). A three-dimensional model of the myoglobin molecule obtained by x-ray analysis. Nature, 181(4610), 662-666. 5. Blake, C. C. F., Koenig, D. F., Mair, G. A., North, A. C. T., Phillips, D. C., & Sarma, V. R. (1965). Structure of hen egg-white lysozyme: a three-dimensional fourier synthesis at 2 Å resolution. Nature, 206(4986), 757-761. 6. Raines, R. T. (1998). Ribonuclease a. Chemical reviews, 98(3), 1045-1066. Wyckoff, H. W., Hardman, K. D., Allewell, N. M., Inagami, T., Tsernoglou, D., Johnson, L. N., & Richards, F. M. (1967). The structure of ribonuclease-S at 6 A resolution. Journal of Biological Chemistry, 242(16), 3749-3753. 7. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., & Bourne, P. E. (2000). The protein data bank. Nucleic acids research, 28(1), 235-242.
Axelbaum, 2014
27
8. This number was determined by searching for structures that are labeled as protein and removing homologous structures with a 90% sequence similarity. URL (March 2014): http://www.rcsb.org/pdb/search/advSearch.do 9. McPherson, A. (2004). Introduction to protein crystallization. Methods, 34(3), 254-265. Dumetz, A. C., Chockla, A. M., Kaler, E. W. & Lenhoff, A. M. (2009). Cryst. Growth Des. 9, 682-691. 10. Gabanyi, M. J., Adams, P. D., Arnold, K., Bordoli, L., Carter, L. G., Flippen-Andersen, J., Gifford, L., Haas, J., Kouranov, A., McLaughlin, W. A., Micallef, D. I., Minor, W., Shah, R., Schwede, T., Tao, Y. P., Westbrook, J. D., Zimmerman, M. & Berman, H. M. (2011). J. Struct. Funct. Genomics 12, 45-54. 11. Retreived from URL (May 2013): http://targetdb.sbkb.org/statistics/TargetStatistics.html#general 12. Chayen, N. E. (2002). Trends Biotechnol. 20, 98-98. Chayen, N. E. & Saridakis, E. (2008). Nat. Methods 5, 147-153. 13. Pusey, M. L., Liu, Z. J., Tempel, W., Praissman, J., Lin, D. W., Wang, B. C., Gavira, J. A. & Ng, J. D. (2005). Prog. Biophys. Mol. Biol. 88, 359-386. 14. Asherie, N., Ginsberg, C., Blass, S., Greenbaum, A. & Knafo, S. (2008). Cryst. Growth Des. 8, 1815-1817. 15. Asherie, N., Ginsberg, C., Greenbaum, A., Blass, S. & Knafo, S. (2008). Cryst. Growth Des. 8, 4200-4207. 16. Asherie, N., Jakoncic, J., Ginsberg, C., Greenbaum, A., Stojanoff, V., Hrnjez, B. J., Blass, S. & Berger, J. (2009). Cryst. Growth Des. 9, 4189-4198. 17. Chayen, N. E. & Saridakis, E. (2001). J. Cryst. Growth 232, 262-264. Magay, E. & Yoon, T. S. (2011). J. Appl. Crystallogr. 44, 252-253. 18. Anand, K., Pal, D. & Hilgenfeld, R. (2002). Acta Crystallogr D 58, 1722-1728. 19. Sigma-Aldrich. Hexylene Glycol [Material Safety Data Sheet]. Retrieved from URL (May 2014): http://www.sigmaaldrich.com/MSDS/MSDS/DisplayMSDSPage.do?country=US&language=en& productNumber=112100&brand=ALDRICH&PageToGoToURL=http%3A%2F%2Fwww.sigma aldrich.com%2Fcatalog%2Fproduct%2Faldrich%2F112100%3Flang%3Den 20. Weiss, M. S., Palm, G. J. & Hilgenfeld, R. (2000). Acta Crystallogr D 56, 952-958. 21. Michaux, C., Pouyez, J., Wouters, J., & Privé, G. G. (2008). Protecting role of cosolvents in protein denaturation by SDS: a structural study. BMC structural biology, 8(1), 29. 22. Aune, K. C. & Tanford, C. (1969). Biochemistry 8, 4579-4585. 23. Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307-326. 24. Vagin, A. & Teplyakov, A. (1997). J. Appl. Crystallogr. 30, 1022-1025. 25. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Acta Crystallogr D 53, 240-255. 26. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Crystallogr D 66, 486-501. 27. Laskowski, R. A., Macarthur, M. W., Moss, D. S. & Thornton, J. M. (1993). J. Appl. Crystallogr. 26, 283-291. 28. Kleywegt, G. J., Harris, M. R., Zou, J. Y., Taylor, T. C., Wahlby, A. & Jones, T. A. (2004). Acta Crystallogr D 60, 2240-2249. 29. Jeffrey, G. A., & Jeffrey, G. A. (1997). An introduction to hydrogen bonding (Vol. 12). New York: Oxford University Press. 30. Bujacz, G., Wrzesniewska, B. & Bujacz, A. (2010). Acta Crystallogr D 66, 789-796. 31. Helliwell, J. R. & Tanley, S. W. M. (2013). Acta Crystallogr D 69, 121-125. 32. Michaux, C., Pouyez, J., Wouters, J., & Privé, G. G. (2008). Protecting role of cosolvents in protein denaturation by SDS: a structural study. BMC structural biology, 8(1), 29. 33. Weiss, M. S., Palm, G. J. & Hilgenfeld, R. (2000). Acta Crystallogr D 56, 952-958. 34. Judge, R. A., Jacobs, R. S., Frazier, T., Snell, E. H. & Pusey, M. L. (1999). Biophys. J. 77, 15851593. 35. Salam, A. & Deleuze, M. S. (2002). J. Chem. Phys. 116, 1296-1302.
Axelbaum, 2014
28
7. Appendix Accession Codes Our structures: Lysozyme without MPD, 4B49; lysozyme with (R)-MPD, 4B4E; lysozyme with (S)-MPD, 4B4I; lysozyme with (RS)-MPD, 4B4J. Michaux et al., 3B72; Weiss et al., 1DPW. (Both are lysozyme with (RS)-MPD.)
Table 3. Crystallographic data and refinement statistics. PDB entry
4B49
4B4E
4B4I
4B4J
MPD added
none
(R)
(S)
(RS)
Data Collection Space group
P43212
P43212
P43212
P43212
Unit cell dimensions (Å)
76.84
77.53
77.44
77.67
a = b, c (Å); (α = β = γ = 90°)
38.69
37.89
37.94
37.70
Resolution (Å) a
20.00-1.15
15.00-1.00
30.00-1.20
30.00-1.25
(1.17-1.15)
(1.02-1.00)
(1.22-1.20)
(1.27-1.25)
Total reflections
562743
853125
512028
456741
Unique reflectionsa
41718 (2040)
62397 (3076)
36597 (1806)
32449 (1584)
Rmerge (%)a
0.08 (0.76)
0.07 (0.82)
0.08 (0.79)
0.06 (0.56)
I/σIa
41.5 (3.2)
43.3 (2.5)
45.1 (5.6)
45.9 (5.5)
Completeness (%)a
99.8 (99.1)
99.3 (99.0)
99.7 (100.0)
99.8 (99.6)
Redundancya
13.5 (10.9)
13.7 (11.2)
14.0 (12.7)
14.1 (13.2)
Mosaicity (°)
0.32
0.26
0.46
0.43
B factor (Å2) (Overall)
15.5
13.4
17.2
15.1
VM (Å3 /Da) / Solvent content (%)
2.03 / 39.6
2.03 / 39.6
2.03 / 39.4
2.03 / 39.3
Axelbaum, 2014
29
Refinement Resolution (Å) a
18.67-1.15
13.30-1.00
21.49-1.20
27.05-1.25
(1.18-1.15)
(1.03-1.00)
(1.23-1.20)
(1.28-1.25)
Reflections [Rcryst +Rfree (5%)]
39425 + 2089
59136 + 3155
34682 + 1825
30732 + 1643
Rcryst / Rfreea
12.7 / 15.1 (22.2 /
12.4 / 14.4 (25.0 /
12.9 / 16.7 (18.8 /
13.05 / 15.57 (18.1 /
25.6)
25.7)
22.1)
22.7)
Protein (no. of residues) b
1001 (129)
1000 (129)
1000 (129)
1001 (129)
MPD c
—
16 (2; R)
16 (2; S)
16 (2; R)
Ions / Ligands (no. of molecules)
11 (4)
3 (3)
2 (2)
2 (2)
Water
267
196
188
162
Protein
12.9
11.9
15.2
13.7
MPD
—
12.3
21.4
13.1
Ions / Ligands
13.1
14.2
17.7
15.4
Water
26.2
21.6
28.3
24.6
Bond length (Å)
0.009
0.008
0.007
0.007
Bond angles (°)
1.349
1.306
1.214
1.272
Most favored (%)
90.3
90.3
88.5
87.6
Additionally favored (%)
9.7
9.7
11.5
12.4
Disallowed (%)
0.0
0.0
0.0
0.0
Number of atoms
B factors (Å2)
RMS deviation from ideal
Ramachandran plot
a
Number in parentheses refers to the highest resolution shell. Atom OXT (residue L129) was omitted during the refinement for structures 4B4E and 4B4J. c In parentheses: (no. of molecules in the structure; enantiomer). b
Axelbaum, 2014
30
Table 4. MPD molecule data from lysozyme-MPD interaction sites.
Interaction Site
Tryptophan 63
PDB ID: Molecule Label
Atom
B-factor
Average B- Occupancy
(Å )
factor (Å2)
C1
76.86
34.2275
C2
44.06
1
O2
21.81
1
CM
15.5
1
C3
31.84
1
C4
34.87
1
O4
23.53
1
C5
25.35
1
4B4E: A 1131 MRD
C1
10.81
(R)-MPD
C2
11.06
0.6
O2
9.94
0.6
CM
14.82
0.6
C3
13.35
0.6
C4
13.85
0.6
O4
13.75
0.6
C5
17.01
0.6
4B4J: A 1132 MRD
C1
12.47
(RS)-MPD
C2
12.17
0.5
O2
10.85
0.5
CM
15.87
0.5
C3
14.39
0.5
C4
15.67
0.5
O4
15.2
0.5
C5
18.86
0.5
MPD Additive
Label
3B72: A 1 MPD (RS)-MPD
2
13.0738
14.435
1
0.6
0.5
Axelbaum, 2014 Tryptophan 123
Phenylalanine 34
31 4B4I: A 1132 MPD
C1
20.53
(S)-MPD
C2
20.55
0.5
O2
23.35
0.5
CM
22.94
0.5
C3
21.15
0.5
C4
22.01
0.5
O4
25.32
0.5
C5
22.42
0.5
C1
13.44
C2
33.55
1
O2
25.46
1
CM
53.77
1
C3
22.42
1
C4
30.19
1
O4
30.14
1
C5
28.32
1
C1
21.48
C2
21.71
1
O2
21.81
1
CM
21.16
1
C3
21.16
1
C4
20.69
1
O4
20.32
1
C5
20.45
1
4B4E: A 1130 MRD
C1
12.35
(R)-MPD
C2
10.66
1
O2
10.76
1
CM
13.31
1
C3
11.14
1
C4
11.39
1
O4
10.53
1
C5
12.17
1
3B72: A 5 MPD (RS)-MPD
1DPW: A 400 MRD (RS)-MPD
22.2838
29.6613
21.0975
11.5388
0.5
1
1
1
Axelbaum, 2014
32
4B4I: A 1133 MPD (S)-MPD
C1
19.63
20.6286
C2
19.63
0.5
O2
20.45
0.5
CM
20.16
0.5
C3
20.02
0.5
C4
26.63
0.5
O4
17.88
0.5
C5 4B4J: A 1133 MRD (RS)-MPD
0.5
0.5
C1
11.38
11.6813
0.55
C2
10.92
0.55
O2
12.07
0.55
CM
13.73
0.55
C3
12.13
0.55
C4
11.37
0.55
O4
10.12
0.55
C5
11.73
0.55
