Advanced exploitation of unmerged reflection data during processing and refinement with autoPROC and BUSTER.

The validation of structural models obtained by macromolecular X-ray crystallography against experimental diffraction data, whether before deposition into the PDB or after, is typically carried out exclusively against the merged data that are eventually archived along with the atomic coordinates. It is shown here that the availability of unmerged reflection data enables valuable additional analyses to be performed that yield improvements in the final models, and tools are presented to implement them, together with examples of the results to which they give access. The first example is the automatic identification and removal of image ranges affected by loss of crystal centering or by excessive decay of the diffraction pattern as a result of radiation damage. The second example is the `reflection-auditing' process, whereby individual merged data items showing especially poor agreement with model predictions during refinement are investigated thanks to the specific metadata (such as image number and detector position) that are available for the corresponding unmerged data, potentially revealing previously undiagnosed instrumental, experimental or processing problems. The third example is the calculation of so-called F(early) - F(late) maps from carefully selected subsets of unmerged amplitude data, which can not only highlight the location and extent of radiation damage but can also provide guidance towards suitable fine-grained parametrizations to model the localized effects of such damage.

Crystal Structure of the Marburg Virus VP35 Oligomerization Domain.

Marburg virus (MARV) is a highly pathogenic filovirus that is classified in a genus distinct from that of Ebola virus (EBOV) (genera Marburgvirus and Ebolavirus, respectively). Both viruses produce a multifunctional protein termed VP35, which acts as a polymerase cofactor, a viral protein chaperone, and an antagonist of the innate immune response. VP35 contains a central oligomerization domain with a predicted coiled-coil motif. This domain has been shown to be essential for RNA polymerase function. Here we present crystal structures of the MARV VP35 oligomerization domain. These structures and accompanying biophysical characterization suggest that MARV VP35 is a trimer. In contrast, EBOV VP35 is likely a tetramer in solution. Differences in the oligomeric state of this protein may explain mechanistic differences in replication and immune evasion observed for MARV and EBOV. IMPORTANCE: Marburg virus can cause severe disease, with up to 90% human lethality. Its genome is concise, only producing seven proteins. One of the proteins, VP35, is essential for replication of the viral genome and for evasion of host immune responses. VP35 oligomerizes (self-assembles) in order to function, yet the structure by which it assembles has not been visualized. Here we present two crystal structures of this oligomerization domain. In both structures, three copies of VP35 twist about each other to form a coiled coil. This trimeric assembly is in contrast to tetrameric predictions for VP35 of Ebola virus and to known structures of homologous proteins in the measles, mumps, and Nipah viruses. Distinct oligomeric states of the Marburg and Ebola virus VP35 proteins may explain differences between them in polymerase function and immune evasion. These findings may provide a more accurate understanding of the mechanisms governing VP35's functions and inform the design of therapeutics.

XModeScore: a novel method for accurate protonation/tautomer-state determination using quantum-mechanically driven macromolecular X-ray crystallographic refinement.

Gaining an understanding of the protein-ligand complex structure along with the proper protonation and explicit solvent effects can be important in obtaining meaningful results in structure-guided drug discovery and structure-based drug discovery. Unfortunately, protonation and tautomerism are difficult to establish with conventional methods because of difficulties in the experimental detection of H atoms owing to the well known limitations of X-ray crystallography. In the present work, it is demonstrated that semiempirical, quantum-mechanics-based macromolecular crystallographic refinement is sensitive to the choice of a protonation-state/tautomer form of ligands and residues, and can therefore be used to explore potential states. A novel scoring method, called XModeScore, is described which enumerates the possible protomeric/tautomeric modes, refines each mode against X-ray diffraction data with the semiempirical quantum-mechanics (PM6) Hamiltonian and scores each mode using a combination of energetic strain (or ligand strain) and rigorous statistical analysis of the difference electron-density distribution. It is shown that using XModeScore it is possible to consistently distinguish the correct bound protomeric/tautomeric modes based on routine X-ray data, even at lower resolutions of around 3 Å. These X-ray results are compared with the results obtained from much more expensive and laborious neutron diffraction studies for three different examples: tautomerism in the acetazolamide ligand of human carbonic anhydrase II (PDB entries 3hs4 and 4k0s), tautomerism in the 8HX ligand of urate oxidase (PDB entries 4n9s and 4n9m) and the protonation states of the catalytic aspartic acid found within the active site of an aspartic protease (PDB entry 2jjj). In each case, XModeScore applied to the X-ray diffraction data is able to determine the correct protonation state as defined by the neutron diffraction data. The impact of QM-based refinement versus conventional refinement on XModeScore is also discussed.

Detection of secondary binding sites in proteins using fragment screening.

Proteins need to be tightly regulated as they control biological processes in most normal cellular functions. The precise mechanisms of regulation are rarely completely understood but can involve binding of endogenous ligands and/or partner proteins at specific locations on a protein that can modulate function. Often, these additional secondary binding sites appear separate to the primary binding site, which, for example for an enzyme, may bind a substrate. In previous work, we have uncovered several examples in which secondary binding sites were discovered on proteins using fragment screening approaches. In each case, we were able to establish that the newly identified secondary binding site was biologically relevant as it was able to modulate function by the binding of a small molecule. In this study, we investigate how often secondary binding sites are located on proteins by analyzing 24 protein targets for which we have performed a fragment screen using X-ray crystallography. Our analysis shows that, surprisingly, the majority of proteins contain secondary binding sites based on their ability to bind fragments. Furthermore, sequence analysis of these previously unknown sites indicate high conservation, which suggests that they may have a biological function, perhaps via an allosteric mechanism. Comparing the physicochemical properties of the secondary sites with known primary ligand binding sites also shows broad similarities indicating that many of the secondary sites may be druggable in nature with small molecules that could provide new opportunities to modulate potential therapeutic targets.

Structure of the BTB domain of Keap1 and its interaction with the triterpenoid antagonist CDDO.

The protein Keap1 is central to the regulation of the Nrf2-mediated cytoprotective response, and is increasingly recognized as an important target for therapeutic intervention in a range of diseases involving excessive oxidative stress and inflammation. The BTB domain of Keap1 plays key roles in sensing environmental electrophiles and in mediating interactions with the Cul3/Rbx1 E3 ubiquitin ligase system, and is believed to be the target for several small molecule covalent activators of the Nrf2 pathway. However, despite structural information being available for several BTB domains from related proteins, there have been no reported crystal structures of Keap1 BTB, and this has precluded a detailed understanding of its mechanism of action and interaction with antagonists. We report here the first structure of the BTB domain of Keap1, which is thought to contain the key cysteine residue responsible for interaction with electrophiles, as well as structures of the covalent complex with the antagonist CDDO/bardoxolone, and of the constitutively inactive C151W BTB mutant. In addition to providing the first structural confirmation of antagonist binding to Keap1 BTB, we also present biochemical evidence that adduction of Cys 151 by CDDO is capable of inhibiting the binding of Cul3 to Keap1, and discuss how this class of compound might exert Nrf2 activation through disruption of the BTB-Cul3 interface.

Statistical quality indicators for electron-density maps.

The commonly used validation metrics for the local agreement of a structure model with the observed electron density, namely the real-space R (RSR) and the real-space correlation coefficient (RSCC), are reviewed. It is argued that the primary goal of all validation techniques is to verify the accuracy of the model, since precision is an inherent property of the crystal and the data. It is demonstrated that the principal weakness of both of the above metrics is their inability to distinguish the accuracy of the model from its precision. Furthermore, neither of these metrics in their usual implementation indicate the statistical significance of the result. The statistical properties of electron-density maps are reviewed and an improved alternative likelihood-based metric is suggested. This leads naturally to a χ(2) significance test of the difference density using the real-space difference density Z score (RSZD). This is a metric purely of the local model accuracy, as required for effective model validation and structure optimization by practising crystallographers prior to submission of a structure model to the PDB. A new real-space observed density Z score (RSZO) is also proposed; this is a metric purely of the model precision, as a substitute for other precision metrics such as the B factor.

Fragment screening using X-ray crystallography.

The fragment-based approach is now well established as an important component of modern drug discovery. A key part in establishing its position as a viable technique has been the development of a range of biophysical methodologies with sufficient sensitivity to detect the binding of very weakly binding molecules. X-ray crystallography was one of the first techniques demonstrated to be capable of detecting such weak binding, but historically its potential for screening was under-appreciated and impractical due to its relatively low throughput. In this chapter we discuss the various benefits associated with fragment-screening by X-ray crystallography, and describe the technical developments we have implemented to allow its routine use in drug discovery. We emphasize how this approach has allowed a much greater exploitation of crystallography than has traditionally been the case within the pharmaceutical industry, with the rapid and timely provision of structural information having maximum impact on project direction.

A new generation of crystallographic validation tools for the protein data bank.

This report presents the conclusions of the X-ray Validation Task Force of the worldwide Protein Data Bank (PDB). The PDB has expanded massively since current criteria for validation of deposited structures were adopted, allowing a much more sophisticated understanding of all the components of macromolecular crystals. The size of the PDB creates new opportunities to validate structures by comparison with the existing database, and the now-mandatory deposition of structure factors creates new opportunities to validate the underlying diffraction data. These developments highlighted the need for a new assessment of validation criteria. The Task Force recommends that a small set of validation data be presented in an easily understood format, relative to both the full PDB and the applicable resolution class, with greater detail available to interested users. Most importantly, we recommend that referees and editors judging the quality of structural experiments have access to a concise summary of well-established quality indicators.

Crystal structure of human CDK4 in complex with a D-type cyclin.

The cyclin D1-cyclin-dependent kinase 4 (CDK4) complex is a key regulator of the transition through the G(1) phase of the cell cycle. Among the cyclin/CDKs, CDK4 and cyclin D1 are the most frequently activated by somatic genetic alterations in multiple tumor types. Thus, aberrant regulation of the CDK4/cyclin D1 pathway plays an essential role in oncogenesis; hence, CDK4 is a genetically validated therapeutic target. Although X-ray crystallographic structures have been determined for various CDK/cyclin complexes, CDK4/cyclin D1 has remained highly refractory to structure determination. Here, we report the crystal structure of CDK4 in complex with cyclin D1 at a resolution of 2.3 A. Although CDK4 is bound to cyclin D1 and has a phosphorylated T-loop, CDK4 is in an inactive conformation and the conformation of the heterodimer diverges from the previously known CDK/cyclin binary complexes, which suggests a unique mechanism for the process of CDK4 regulation and activation.

PDB_REDO: automated re-refinement of X-ray structure models in the PDB.

Structural biology, homology modelling and rational drug design require accurate three-dimensional macromolecular coordinates. However, the coordinates in the Protein Data Bank (PDB) have not all been obtained using the latest experimental and computational methods. In this study a method is presented for automated re-refinement of existing structure models in the PDB. A large-scale benchmark with 16 807 PDB entries showed that they can be improved in terms of fit to the deposited experimental X-ray data as well as in terms of geometric quality. The re-refinement protocol uses TLS models to describe concerted atom movement. The resulting structure models are made available through the PDB_REDO databank (http://www.cmbi.ru.nl/pdb_redo/). Grid computing techniques were used to overcome the computational requirements of this endeavour.

Experimental determination of optimal root-mean-square deviations of macromolecular bond lengths and angles from their restrained ideal values.

A number of inconsistencies are apparent in the recent research paper by Jaskolski et al. [(2007), Acta Cryst. D63, 611-620] concerning their recommendations for the values of the magnitude and resolution-dependence of the root-mean-square deviations (RMSDs) of bond lengths and angles from their restrained ideal values in macromolecular refinement, as well as their suggestions for the use of variable standard uncertainties dependent on atomic displacement parameters (ADPs) and occupancies. Whilst many of the comments and suggestions in the paper regarding updates for the ideal geometry values proposed by Engh and Huber are entirely reasonable and supported by the experimental evidence, the recommendations concerning the optimal values of RMSDs appear to be in conflict with previous experimental and theoretical work in this area [Tickle et al. (1998), Acta Cryst. D54, 243-252] and indeed appear to be based on a misunderstanding of the distinction between RMSD and standard uncertainty (SU). In contrast, it is proposed here that the optimal values of all desired weighting parameters, in particular the weighting parameters for the ADP differences and for the diffraction terms, be estimated by the purely objective procedure of maximizing the experiment-based log(free likelihood). In principle, this allows all weighting parameters that are not known accurately a priori to be scaled globally, relative to those that are known accurately, for an optimal refinement. The RMS Z score (RMSZ) is recommended as a more satisfactory statistic than the RMSD to assess the extent to which the geometry deviates from the ideal values and a theoretical rationale for the results obtained is presented in which the optimal RMSZ is identified as the calculated versus true Z-score correlation coefficient, the latter being a monotonic function of the resolution cutoff of the data. Regarding the proposal to use variable standard uncertainties, it is suggested that any departure from the current practice of using fixed weights for geometric restraints based on experimental values of standard uncertainties be subject to the same experiment-based validation.

Structure of a CBS-domain pair from the regulatory gamma1 subunit of human AMPK in complex with AMP and ZMP.

AMP-activated kinase (AMPK) is central to sensing energy status in eukaryotic cells via binding of AMP and ATP to CBS (cystathionine beta-synthase) domains in the regulatory gamma subunit. The structure of a CBS-domain pair from human AMPK gamma1 in complex with the physiological activator AMP and the pharmacological activator ZMP (AICAR) is presented.

Automated protein-ligand crystallography for structure-based drug design.

An approach to automate protein-ligand crystallography is presented, with the aim of increasing the number of structures available to structure-based drug design. The methods we propose deal with the automatic interpretation of diffraction data for targets with known protein structures, and provide easy access to the results. Central to the system is a novel procedure that fully automates the placement of ligands into electron density maps. Automation provides an objective way to structure solution, whereas manual placement can be rather subjective, especially for data of low to medium resolution. Ligands are placed by docking into electron density, whilst taking care of protein-ligand interactions. The ligand fitting procedure has been validated on both public domain and in-house examples. Some of the latter deal with cocktails of low-molecular weight compounds, as used in fragment-based drug discovery by crystallography. For such library-screening experiments we show that the method can automatically identify which of the compounds from a cocktail is bound.

Fragment-based lead discovery using X-ray crystallography.

Fragment screening offers an alternative to traditional screening for discovering new leads in drug discovery programs. This paper describes a fragment screening methodology based on high throughput X-ray crystallography. The method is illustrated against five proteins (p38 MAP kinase, CDK2, thrombin, ribonuclease A, and PTP1B). The fragments identified have weak potency (>100 microM) but are efficient binders relative to their size and may therefore represent suitable starting points for evolution to good quality lead compounds. The examples illustrate that a range of molecular interactions (i.e., lipophilic, charge-charge, neutral hydrogen bonds) can drive fragment binding and also that fragments can induce protein movement. We believe that the method has great potential for the discovery of novel lead compounds against a range of targets, and the companion paper illustrates how lead compounds have been identified for p38 MAP kinase starting from fragments such as those described in this paper.

High-throughput protein crystallography and drug discovery.

Single crystal X-ray diffraction is the technique of choice for studying the interactions of small organic molecules with proteins by determining their three-dimensional structures; however the requirement for highly purified protein and lack of process automation have traditionally limited its use in this field. Despite these shortcomings, the use of crystal structures of therapeutically relevant drug targets in pharmaceutical research has increased significantly over the last decade. The application of structure-based drug design has resulted in several marketed drugs and is now an established discipline in most pharmaceutical companies. Furthermore, the recently published full genome sequences of Homo sapiens and a number of micro-organisms have provided a plethora of new potential drug targets that could be utilised in structure-based drug design programs. In order to take maximum advantage of this explosion of information, techniques have been developed to automate and speed up the various procedures required to obtain protein crystals of suitable quality, to collect and process the raw X-ray diffraction data into usable structural information, and to use three-dimensional protein structure as a basis for drug discovery and lead optimisation. This tutorial review covers the various technologies involved in the process pipeline for high-throughput protein crystallography as it is currently being applied to drug discovery. It is aimed at synthetic and computational chemists, as well as structural biologists, in both academia and industry, who are interested in structure-based drug design.

Crystal structures of human cytochrome P450 3A4 bound to metyrapone and progesterone.

Cytochromes P450 (P450s) metabolize a wide range of endogenous compounds and xenobiotics, such as pollutants, environmental compounds, and drug molecules. The microsomal, membrane-associated, P450 isoforms CYP3A4, CYP2D6, CYP2C9, CYP2C19, CYP2E1, and CYP1A2 are responsible for the oxidative metabolism of more than 90% of marketed drugs. Cytochrome P450 3A4 (CYP3A4) metabolizes more drug molecules than all other isoforms combined. Here we report three crystal structures of CYP3A4: unliganded, bound to the inhibitor metyrapone, and bound to the substrate progesterone. The structures revealed a surprisingly small active site, with little conformational change associated with the binding of either compound. An unexpected peripheral binding site is identified, located above a phenylalanine cluster, which may be involved in the initial recognition of substrates or allosteric effectors.

Estimation of weights and validation: a marginal likelihood approach.

The estimation of weights is quite an important aspect of the restrained refinement of macromolecular structures and related procedures such as the estimation of coordinate errors and structure validation using geometrical criteria. In principle, the method of maximum likelihood can be used for estimation of both atomic and weighting parameters. However, the low observation-to-parameter ratio in macromolecular refinement makes this kind of estimate of weighting parameters seriously biased; thus, the weighting parameters have traditionally been estimated separately from atomic parameters using a special technique, such as minimizing the free R factor. However, the variance of the latter estimate is large, as only a small portion of all data is used. In this work, an estimator of weights is proposed which is based on an approximation of a marginal likelihood function of the weighting parameters and which uses all the X-ray data. There is a known true value for the overall scaling coefficient for distance variances in restrained refinement and therefore the (maximum-likelihood) estimator for this coefficient may be used as a validation statistic.

Engineering of an intersubunit disulfide bridge in the iron-superoxide dismutase of Mycobacterium tuberculosis.

With the aim of enhancing interactions involved in dimer formation, an intersubunit disulfide bridge was engineered in the superoxide dismutase enzyme of Mycobacterium tuberculosis. Ser-123 was chosen for mutation to cysteine since it resides at the dimer interface where the serine side chain interacts with the same residue in the opposite subunit. Gel electrophoresis and X-ray crystallographic studies of the expressed mutant confirmed formation of the disulfide bond under nonreducing conditions. However, the mutant protein was found to be less stable than the wild type as judged by susceptibility to denaturation in the presence of guanidine hydrochloride. Decreased stability probably results from formation of a disulfide bridge with a suboptimal torsion angle and exclusion of solvent molecules from the dimer interface.

Sensitivity of molecular docking to induced fit effects in influenza virus neuraminidase.

Many proteins undergo small side chain or even backbone movements on binding of different ligands into the same protein structure. This is known as induced fit and is potentially problematic for virtual screening of databases against protein targets. In this report we investigate the limits of the rigid protein approximation used by the docking program, GOLD, through cross-docking using protein structures of influenza neuraminidase. Neuraminidase is known to exhibit small but significant induced fit effects on ligand binding. Some neuraminidase crystal structures caused concern due to the bound ligand conformation and GOLD performed poorly on these complexes. A 'clean' set, which contained unique, unambiguous complexes, was defined. For this set, the lowest energy structure was correctly docked (i.e. RMSD < 1.5 A away from the crystal reference structure) in 84% of proteins, and the most promiscuous protein (1mwe) was able to dock all 15 ligands accurately including those that normally required an induced fit movement. This is considerably better than the 70% success rate seen with GOLD against general validation sets. Inclusion of specific water molecules involved in water-mediated hydrogen bonds did not significantly improve the docking performance for ligands that formed water-mediated contacts but it did prevent docking of ligands that displaced these waters. Our data supports the use of a single protein structure for virtual screening with GOLD in some applications involving induced fit effects, although care must be taken to identify the protein structure that performs best against a wide variety of ligands. The performance of GOLD was significantly better than the GOLD implementation of ChemScore and the reasons for this are discussed. Overall, GOLD has shown itself to be an extremely good, robust docking program for this system.