High‐resolution crystal structures leverage protein binding affinity predictions

Predicting protein binding affinities from structural data has remained elusive, a difficulty owing to the variety of protein binding modes. Using the structure‐affinity‐benchmark (SAB, 144 cases with bound/unbound crystal structures and experimental affinity measurements), prediction has been undertaken either by fitting a model using a handfull of predefined variables, or by training a complex model from a large pool of parameters (typically hundreds). The former route unnecessarily restricts the model space, while the latter is prone to overfitting. We design models in a third tier, using 12 variables describing enthalpic and entropic variations upon binding, and a model selection procedure identifying the best sparse model built from a subset of these variables. Using these models, we report three main results. First, we present models yielding a marked improvement of affinity predictions. For the whole dataset, we present a model predicting K_d within 1 and 2 orders of magnitude for 48% and 79% of cases, respectively. These statistics jump to 62% and 89% respectively, for the subset of the SAB consisting of high resolution structures. Second, we show that these performances owe to a new parameter encoding interface morphology and packing properties of interface atoms. Third, we argue that interface flexibility and prediction hardness do not correlate, and that for flexible cases, a performance matching that of the whole SAB can be achieved. Overall, our work suggests that the affinity prediction problem could be partly solved using databases of high resolution complexes whose affinity is known. Proteins 2016; 84:9–20. © 2015 Wiley Periodicals, Inc.     

High‐resolution crystal structures reveal a mixture of conformers of the Gly61‐Asp62 peptide bond in an oxidized flavodoxin from Bacillus cereus

Flavodoxins (Flds) are small proteins that shuttle electrons in a range of reactions in microorganisms. Flds contain a redox‐active cofactor, a flavin mononucleotide (FMN), and it is well established that when Flds are reduced by one electron, a peptide bond close to the FMN isoalloxazine ring flips to form a new hydrogen bond with the FMN N5H, stabilizing the one‐electron reduced state. Here, we present high‐resolution crystal structures of Flavodoxin 1 from Bacillus cereus in both the oxidized (ox) and one‐electron reduced (semiquinone, sq) state. We observe a mixture of conformers in the oxidized state; a 50:50 distribution between the established oxidized conformation where the peptide bond is pointing away from the flavin, and a conformation where the peptide bond is pointing toward the flavin, approximating the conformation in the semiquinone state. We use single‐crystal spectroscopy to demonstrate that the mixture of conformers is not caused by radiation damage to the crystal. This is the first time that such a mixture of conformers is reported in a wild‐type Fld. We therefore carried out a survey of published Fld structures, which show that several proteins have a pronounced conformational flexibility of this peptide bond. The degree of flexibility seems to be modulated by the presence, or absence, of stabilizing interactions between the peptide bond carbonyl and its surrounding amino acids. We hypothesize that the degree of conformational flexibility will affect the Fld ox/sq redox potential.     

Sub-atomic resolution crystal structure of cholesterol oxidase: what atomic resolution crystallography reveals about enzyme mechanism and the role of the FAD cofactor in redox activity

The crystal structure of cholesterol oxidase, a 56kDa flavoenzyme was anisotropically refined to 0.95A resolution. The final crystallographic R-factor and R(free) value is 11.0% and 13.2%, respectively. The quality of the electron density maps has enabled modeling of alternate conformations for 83 residues in the enzyme, many of which are located in the active site. The additional observed structural features were not apparent in the previous high-resolution structure (1.5A resolution) and have enabled the identification of a narrow tunnel leading directly to the isoalloxazine portion of the FAD prosthetic group. The hydrophobic nature of this narrow tunnel suggests it is the pathway for molecular oxygen to access the isoalloxazine group for the oxidative half reaction. Resolving the alternate conformations in the active site residues provides a model for the dynamics of substrate binding and a potential oxidation triggered gating mechanism involving access to the hydrophobic tunnel. This structure reveals that the NE2 atom of the active site histidine residue, H447, critical to the redox activity of this flavin oxidase, acts as a hydrogen bond donor rather than as hydrogen acceptor. The atomic resolution structure of cholesterol oxidase has revealed the presence of hydrogen atoms, dynamic aspects of the protein and how side-chain conformations are correlated with novel structural features such as the oxygen tunnel. This new structural information has provided us with the opportunity to re-analyze the roles played by specific residues in the mechanism of the enzyme.     

On interpretation of protein X‐ray structures: Planarity of the peptide unit

Pauling's mastery of peptide stereochemistry—based on small molecule crystal structures and the theory of chemical bonding—led to his realization that the peptide unit is planar and then to the Pauling–Corey–Branson model of the α‐helix. Similarly, contemporary protein structure refinement is based on experimentally determined diffraction data together with stereochemical restraints. However, even an X‐ray structure at ultra‐high resolution is still an under‐determined model in which the linkage among refinement parameters is complex. Consequently, restrictions imposed on any given parameter can affect the entire structure. Here, we examine recent studies of high resolution protein X‐ray structures, where substantial distortions of the peptide plane are found to be commonplace. Planarity is assessed by the ω‐angle, a dihedral angle determined by the peptide bond (C? N) and its flanking covalent neighbors; for an ideally planar trans peptide, ω = 180°. By using a freely available refinement package, Phenix [Afonine et al. (2012) Acta Cryst. D, 68:352–367], we demonstrate that tightening default restrictions on the ω‐angle can significantly reduce apparent deviations from peptide unit planarity without consequent reduction in reported evaluation metrics (e.g., R‐factors). To be clear, our result does not show that substantial non‐planarity is absent, only that an equivalent alternative model is possible. Resolving this disparity will ultimately require improved understanding of the deformation energy. Meanwhile, we urge inclusion of ω‐angle statistics in new structure reports in order to focus critical attention on the usual practice of assigning default values to ω‐angle constraints during structure refinement. Proteins 2015; 83:1687–1692. © 2015 Wiley Periodicals, Inc.     

The 0.83 A resolution crystal structure of alpha-lytic protease reveals the detailed structure of the active site and identifies a source of conformational strain

The crystal structure of the extracellular bacterial serine protease α-lytic protease (αLP) has been solved at 0.83 Å resolution at pH 8. This ultra-high resolution structure allows accurate analysis of structural elements not possible with previous structures. Hydrogen atoms are visible, and confirm active-site hydrogen-bonding interactions expected for the apo enzyme. In particular, His57 N^δ1 participates in a normal hydrogen bond with Asp102 in the catalytic triad, with a hydrogen atom visible 0.83(±0.06) Å from the His N^δ1. The catalytic Ser195 occupies two conformations, one corresponding to a population of His57 that is doubly protonated, the other to the singly protonated His57. Based on the occupancy of these conformations, the pK_a of His57 is calculated to be ∼8.8 when a sulfate ion occupies the active site. This 0.83 Å structure has allowed critical analysis of geometric distortions within the structure. Interestingly, Phe228 is significantly distorted from planarity. The distortion of Phe228, buried in the core of the C-terminal domain, occurs at an estimated energetic cost of 4.1 kcal/mol. The conformational space for Phe228 is severely limited by the presence of Trp199, which prevents Phe228 from adopting the rotamer observed in many other chymotrypsin family members. In αLP, the only allowed rotamer leads to the deformation of Phe228 due to steric interactions with Thr181. We hypothesize that tight packing of co-evolved residues in this region, and the subsequent deformation of Phe228, contributes to the high cooperativity and large energetic barriers for folding and unfolding of αLP. The kinetic stability imparted by the large, cooperative unfolding barrier plays a critical role in extending the lifetime of the protease in its harsh environment.     

A high quality nuclear magnetic resonance solution structure of peptide deformylase from Escherichia coli: Application of an automated assignment strategy using GARANT

The NMR structure of the peptide deformylase (PDF) (1–150) from Escherichia coli, which is an essential enzyme that removes the formyl group from nascent polypeptides and represents a potential target for drug discovery, was determined using 15N/13C doubly labeled protein. Nearly completely automated assignment routines were employed to assign three-dimensional triple resonance, 15N-resolved and 13C-resolved NOESY spectra using the program GARANT. This assignment strategy, demonstrated on a 17 kDa protein, is a significant advance in the automation of NMR data assignment and structure determination that will accelerate future work. A total of 2302 conformational constraints were collected as input for the distance geometry program DYANA. After restrained energy minimization with the program X-PLOR the 20 best conformers characterize a high quality structure with an average of 0.43 Å for the root-mean-square deviation calculated from the backbone atoms N, C and C, and 0.81 Å for all heavy atoms of the individual conformers relative to the mean coordinates for residues 1 to 150. The globular fold of PDF contains two -helices comprising residues 25–40, 125–138, six -strands 57–60, 70–77, 85–88, 98–101, 105–111, 117–123 and one 310 helix comprising residues 49–51. The C-terminal helix contains the HEXXH motif positioning a zinc ligand in a similar fashion to other metalloproteases, with the third ligand being cysteine and the fourth presumably a water. The three-dimensional structure of PDF affords insight into the substrate recognition and specificity for N-formylated over N-acetylated substrates and is compared to other PDF structures.     

Atomic resolution crystallography of a complex of triosephosphate isomerase with a reaction‐intermediate analog: New insight in the proton transfer reaction mechanism

Enzymes achieve their catalytic proficiency by precisely positioning the substrate and catalytic residues with respect to each other. Atomic resolution crystallography is an excellent tool to study the important details of these geometric active‐site features. Here, we have investigated the reaction mechanism of triosephosphate isomerase (TIM) using atomic resolution crystallographic studies at 0.82‐Å resolution of leishmanial TIM complexed with the well‐studied reaction‐intermediate analog phosphoglycolohydroxamate (PGH). Remaining unresolved aspects of the reaction mechanism of TIM such as the protonation state of the first reaction intermediate and the properties of the hydrogen‐bonding interactions in the active site are being addressed. The hydroxamate moiety of PGH interacts via unusually short hydrogen bonds of its N1? O1 moiety with the carboxylate group of the catalytic glutamate (Glu167), for example, the distance of N1(PGH)‐OE2(Glu167) is 2.69 ± 0.01 Å and the distance of O1(PGH)‐OE1(Glu167) is 2.60 ± 0.01 Å. Structural comparisons show that the side chain of the catalytic base (Glu167) can move during the reaction cycle in a small cavity, located above the hydroxamate plane. The structure analysis suggests that the hydroxamate moiety of PGH is negatively charged. Therefore, the bound PGH mimics the negatively charged enediolate intermediate, which is formed immediately after the initial proton abstraction from DHAP by the catalytic glutamate. The new findings are discussed in the context of the current knowledge of the TIM reaction mechanism. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

Water polygons in high‐resolution protein crystal structures

We have analyzed the interstitial water (ISW) structures in 1500 protein crystal structures deposited in the Protein Data Bank that have greater than 1.5 Å resolution with less than 90% sequence similarity with each other. We observed varieties of polygonal water structures composed of three to eight water molecules. These polygons may represent the time‐ and space‐averaged structures of “stable” water oligomers present in liquid water, and their presence as well as relative population may be relevant in understanding physical properties of liquid water at a given temperature. On an average, 13% of ISWs are localized enough to be visible by X‐ray diffraction. Of those, averages of 78% are water molecules in the first water layer on the protein surface. Of the localized ISWs beyond the first layer, almost half of them form water polygons such as trigons, tetragons, as well as expected pentagons, hexagons, higher polygons, partial dodecahedrons, and disordered networks. Most of the octagons and nanogons are formed by fusion of smaller polygons. The trigons are most commonly observed. We suggest that our observation provides an experimental basis for including these water polygon structures in correlating and predicting various water properties in liquid state.     

Atomic resolution structures of oxidized [4Fe-4S] ferredoxin from Bacillus thermoproteolyticus in two crystal forms: systematic distortion of [4Fe-4S] cluster in the protein

Diffraction data of two crystal forms (forms I and II) of [4Fe-4S] ferredoxin from Bacillus thermoproteolyticus have been collected to 0.92 A and 1.00 A resolutions, respectively, at 100 K using synchrotron radiation. Anisotropic temperature factors were introduced for all non-hydrogen atoms in the refinement with SHELX-97, in which stereochemical restraints were applied to the protein chain but not to the [4Fe-4S] cluster. The final crystallographic R-factors are 9.8 % for 7.0-0.92 A resolution data of the form I and 11.2 % for the 13.3-1.0 A resolution data of the form II. Many hydrogen atoms as well as multiple conformations for several side-chains have been identified. The present refinement has revised the conformations of several peptide bonds and side-chains assigned previously at 2.3 A resolution; the largest correction was that the main-chain of Pro1 and the side-chain of Lys2 were changed by rotating the C(alpha)-C bond of Lys2. Although the overall structures in the two crystal forms are very similar, conformational differences are observed in the two residues at the middle (Glu29 and Asp30) and the C-terminal residues, which have large temperature factors. The [4Fe-4S] cluster is a distorted cube with non-planar rhombic faces. Slight but significant compression of the four Fe-S bonds along one direction is observed in both crystal forms, and results in the D(2d) symmetry of the cluster. The compressed direction of the cluster relative to the protein is conserved in the two crystal forms and consistent with that in one of the clusters in Clostridium acidurici ferredoxin.     

Five atomic resolution structures of endothiapepsin inhibitor complexes: implications for the aspartic proteinase mechanism

Endothiapepsin is derived from the fungus Endothia parasitica and is a member of the aspartic proteinase class of enzymes. This class of enzyme is comprised of two structurally similar lobes, each lobe contributing an aspartic acid residue to form a catalytic dyad that acts to cleave the substrate peptide bond. The three-dimensional structures of endothiapepsin bound to five transition state analogue inhibitors (H189, H256, CP-80,794, PD-129,541 and PD-130,328) have been solved at atomic resolution allowing full anisotropic modelling of each complex. The active sites of the five structures have been studied with a view to studying the catalytic mechanism of the aspartic proteinases by locating the active site protons by carboxyl bond length differences and electron density analysis. In the CP-80,794 structure there is excellent electron density for the hydrogen on the inhibitory statine hydroxyl group which forms a hydrogen bond with the inner oxygen of Asp32. The location of this proton has implications for the catalytic mechanism of the aspartic proteinases as it is consistent with the proposed mechanism in which Asp32 is the negatively charged aspartate. A number of short hydrogen bonds (approximately 2.6 A) with ESD values of around 0.01 A that may have a role in catalysis have been identified within the active site of each structure; the lengths of these bonds have been confirmed using NMR techniques. The possibility and implications of low barrier hydrogen bonds in the active site are considered.     

Veni,vidi, vici - atomic resolution unravelling the mysteries of protein function

Atomic resolution macromolecular crystallography has become a powerful and versatile tool in structural biology; the number of atomic resolution structures is steadily increasing. Novel techniques are being developed and the use of complementary methods that span the field from sample preparation to validation and analysis of the resulting models has emerged. These allow the fuller exploitation of the information stored in crystal structures and reveal a depth of structural detail that was unattainable in the recent past.     

Centrosymmetric bilayers in the 0.75 A resolution structure of a designed alpha-helical peptide, D,L-Alpha-1.

We report the 0.75 A crystal structure of a racemic mixture of the 12-residue designed peptide "Alpha-1" (Acetyl-ELLKKLLEELKG), the L-enantiomer of which is described in the accompanying paper. Equivalent solutions of the centrosymmetric bilayers were determined by two direct phasing programs in space groups P1 and P1bar. The unit cell contains two L-alpha-helices and two D-alpha-helices. The columnar-sheet bilayer motif seen in L-Alpha-1 is maintained in the D,L-Alpha-1 structure except that each sheet of head-to-tail helices is composed of one enantiomer and is related to its neighboring sheets by inversion symmetry. Comparison to the L-Alpha-1 structure provides further insight into peptide design. The high resolution and small asymmetric unit allowed building an intricate model (R = 13.1%, Rfree = 14.5%) that incorporates much of the discrete disorder of peptide and solvent. Ethanolamine and 2-methyl-2,4-pentanediol (MPD) molecules bind near helix termini. Rigid body analysis identifies sites of restricted displacements and torsions. Side-chain discrete disorder propagates into the backbone of one helix but not the other. Although no side chain in Alpha-1 is rigid, the environments in the crystal restrict some of them to no or only one active torsion.     

Label‐free ultra‐sensitive visualization of structure below the diffraction resolution limit

For both fundamental study of biological processes and early diagnosis of diseases, information about nanoscale changes in tissue and cell structure is crucial. Nowadays, almost all currently known nanoscopy methods rely upon the contrast created by fluorescent stains attached to the object or molecule of interest. This causes limitations due to the impact of the label on the object and its environment, as well as its applicability in vivo, particularly in humans. In this paper, a new label‐free approach to visualize small structure with nano‐sensitivity to structural alterations is introduced. Numerically synthesized profiles of the axial spatial frequencies are used to probe the structure within areas whose size can be beyond the diffraction resolution limit. Thereafter, nanoscale structural alterations within such areas can be visualized and objects, including biological ones, can be investigated with sub‐wavelength resolution, in vivo, in their natural environment. Some preliminary results, including numerical simulations and experiments, which demonstrate the nano‐sensitivity and super‐resolution ability of our approach, are presented.      

Ultrahigh resolution drug design. II. Atomic resolution structures of human aldose reductase holoenzyme complexed with Fidarestat and Minalrestat: implications for the binding of cyclic imide inhibitors

The X-ray structures of human aldose reductase holoenzyme in complex with the inhibitors Fidarestat (SNK-860) and Minalrestat (WAY-509) were determined at atomic resolutions of 0.92 A and 1.1 A, respectively. The hydantoin and succinimide moieties of the inhibitors interacted with the conserved anion-binding site located between the nicotinamide ring of the coenzyme and active site residues Tyr48, His110, and Trp111. Minalrestat's hydrophobic isoquinoline ring was bound in an adjacent pocket lined by residues Trp20, Phe122, and Trp219, with the bromo-fluorobenzyl group inside the "specificity" pocket. The interactions between Minalrestat's bromo-fluorobenzyl group and the enzyme include the stacking against the side-chain of Trp111 as well as hydrogen bonding distances with residues Leu300 and Thr113. The carbamoyl group in Fidarestat formed a hydrogen bond with the main-chain nitrogen atom of Leu300. The atomic resolution refinement allowed the positioning of hydrogen atoms and accurate determination of bond lengths of the inhibitors, coenzyme NADP+ and active-site residue His110. The 1'-position nitrogen atom in the hydantoin and succinimide moieties of Fidarestat and Minalrestat, respectively, form a hydrogen bond with the Nepsilon2 atom of His 110. For Fidarestat, the electron density indicated two possible positions for the H-atom in this bond. Furthermore, both native and anomalous difference maps indicated the replacement of a water molecule linked to His110 by a Cl-ion. These observations suggest a mechanism in which Fidarestat is bound protonated and becomes negatively charged by donating the proton to His110, which may have important implications on drug design.     

Structural and mechanistic insight from high resolution structures of archaeal rhodopsins

Lipidic cubic phase-grown crystals yielded high resolution structures of a number of archaeal retinal proteins, the molecular mechanisms of which are being revealed as structures of photocycle intermediates become available. The structural basis for bacteriorhodopsin's mechanism of proton pumping is discussed, revealing a well-synchronized sequence of molecular events. Comparison with the high resolution structures of the halide pump halorhodopsin, as well as with the receptor sensory rhodopsin II, illustrates how small and localized structural changes result in functional divergence. Fundamental principles of energy transduction and sensory reception in the archaeal rhodopsins, which may have relevance to other systems, are discussed.     

The crystal structure of Z‐(Aib)10‐OH at 0.65 Å resolution: three complete turns of 310‐helix

The synthetic peptide Z‐(Aib)₁₀‐OH was crystallized from hot methanol by slow evaporation. The crystal used for data collection reflected synchrotron radiation to sub‐atomic resolution, where the bonding electron density becomes visible between the non‐hydrogen atoms. Crystals belong to the centrosymmetric space group P . Both molecules in the asymmetric unit form regular 3₁₀‐helices. All residues in each molecule possess the same handedness, which is in contrast to all other crystal structure determined to date of longer Aib‐homopeptides. These other peptides are C‐terminal protected by OtBu or OMe. In these cases, because of the missing ability of the C‐terminal protection group to form a hydrogen bond to the residue i‐3, the sense of the helix is reversed in the last residue. Here, the C‐terminal OH‐groups form hydrogen bonds to the residues i‐3, in part mediated by water molecules. This makes Z‐(Aib)₁₀‐OH an Aib‐homopeptide with three complete 3₁₀‐helical turns in spite of the shorter length it has compared with Z‐(Aib)₁₁‐OtBu, the only homopeptide to date with three complete turns.     

Packed protein bilayers in the 0.90 A resolution structure of a designed alpha helical bundle.

A 12-residue peptide designed to form an alpha-helix and self-associate into an antiparallel 4-alpha-helical bundle yields a 0.9 A crystal structure revealing unanticipated features. The structure was determined by direct phasing with the "Shake-and-Bake" program, and contains four crystallographically distinct 12-mer peptide molecules plus solvent for a total of 479 atoms. The crystal is formed from nearly ideal alpha-helices hydrogen bonded head-to-tail into columns, which in turn pack side-by-side into sheets spanning the width of the crystal. Within each sheet, the alpha-helices run antiparallel and are closely spaced (9-10 A center-to-center). The sheets are more loosely packed against each other (13-14 A between helix centers). Each sheet is amphiphilic: apolar leucine side chains project from one face, charged lysine and glutamate side chains from the other face. The sheets are stacked with two polar faces opposing and two apolar faces opposing. The result is a periodic biomaterial composed of packed protein bilayers, with alternating polar and apolar interfaces. All of the 30 water molecules in the unit cell lie in the polar interface or between the stacked termini of helices. A section through the sheet reveals that the helices packed at the apolar interface resemble the four-alpha-helical bundle of the design, but the helices overhang parts of the adjacent bundles, and the helix crossing angles are less steep than intended (7-11 degrees rather than 18 degrees).     

The mechanism of vault opening from the high resolution structure of the N‐terminal repeats of MVP

Vaults are ubiquitous ribonucleoprotein complexes involved in a diversity of cellular processes, including multidrug resistance, transport mechanisms and signal transmission. The vault particle shows a barrel‐shaped structure organized in two identical moieties, each consisting of 39 copies of the major vault protein MVP. Earlier data indicated that vault halves can dissociate at acidic pH. The crystal structure of the vault particle solved at 8 Å resolution, together with the 2.1‐Å structure of the seven N‐terminal domains (R1–R7) of MVP, reveal the interactions governing vault association and provide an explanation for a reversible dissociation induced by low pH. The structural comparison with the recently published 3.5 Å model shows major discrepancies, both in the main chain tracing and in the side chain assignment of the two terminal domains R1 and R2.     

The structure of the soluble domain of an archaeal Rieske iron-sulfur protein at 1.1 A resolution

The first crystal structure of an archaeal Rieske iron-sulfur protein, the soluble domain of Rieske iron-sulfur protein II (soxF) from the hyperthermo-acidophile Sulfolobus acidocaldarius, has been solved by multiple wavelength anomalous dispersion (MAD) and has been refined to 1.1 A resolution. SoxF is a subunit of the terminal oxidase supercomplex SoxM in the plasma membrane of S. acidocaldarius that combines features of a cytochrome bc(1) complex and a cytochrome c oxidase. The [2Fe-2S] cluster of soxF is most likely the primary electron acceptor during the oxidation of caldariella quinone by the cytochrome a(587)/Rieske subcomplex. The geometry of the [2Fe-2S] cluster and the structure of the cluster-binding site are almost identical in soxF and the Rieske proteins from eucaryal cytochrome bc(1) and b(6)f complexes, suggesting a strict conservation of the catalytic mechanism. The main domain of soxF and part of the cluster-binding domain, though structurally related, show a significantly divergent structure with respect to topology, non-covalent interactions and surface charges. The divergent structure of soxF reflects a different topology of the soxM complex compared to eucaryal bc complexes and the adaptation of the protein to the extreme ambient conditions on the outer membrane surface of a hyperthermo-acidophilic organism.