FRAGSITE: A Fragment-Based Approach for Virtual Ligand Screening.

To reduce time and cost, virtual ligand screening (VLS) often precedes experimental ligand screening in modern drug discovery. Traditionally, high-resolution structure-based docking approaches rely on experimental structures, while ligand-based approaches need known binders to the target protein and only explore their nearby chemical space. In contrast, our structure-based FINDSITEcomb2.0 approach takes advantage of predicted, low-resolution structures and information from ligands that bind distantly related proteins whose binding sites are similar to the target protein. Using a boosted tree regression machine learning framework, we significantly improved FINDSITEcomb2.0 by integrating ligand fragment scores as encoded by molecular fingerprints with the global ligand similarity scores of FINDSITEcomb2.0. The new approach, FRAGSITE, exploits our observation that ligand fragments, e.g., rings, tend to interact with stereochemically conserved protein subpockets that also occur in evolutionarily unrelated proteins. FRAGSITE was benchmarked on the 102 protein DUD-E set, where any template protein whose sequence identify >30% to the target was excluded. Within the top 100 ranked molecules, FRAGSITE improves VLS precision and recall by 14.3 and 18.5%, respectively, relative to FINDSITEcomb2.0. Moreover, the mean top 1% enrichment factor increases from 25.2 to 30.2. On average, both outperform state-of-the-art deep learning-based methods such as AtomNet. On the more challenging unbiased set LIT-PCBA, FRAGSITE also shows better performance than ligand similarity-based and docking approaches such as two-dimensional ECFP4 and Surflex-Dock v.3066. On a subset of 23 targets from DEKOIS 2.0, FRAGSITE shows much better performance than the boosted tree regression-based, vScreenML scoring function. Experimental testing of FRAGSITE's predictions shows that it has more hits and covers a more diverse region of chemical space than FINDSITEcomb2.0. For the two proteins that were experimentally tested, DHFR, a well-studied protein that catalyzes the conversion of dihydrofolate to tetrahydrofolate, and the kinase ACVR1, FRAGSITE identified new small-molecule nanomolar binders. Interestingly, one new binder of DHFR is a kinase inhibitor predicted to bind in a new subpocket. For ACVR1, FRAGSITE identified new molecules that have diverse scaffolds and estimated nanomolar to micromolar affinities. Thus, FRAGSITE shows significant improvement over prior state-of-the-art ligand virtual screening approaches. A web server is freely available for academic users at http:/sites.gatech.edu/cssb/FRAGSITE.

Differential kinase activity of ACVR1 G328V and R206H mutations with implications to possible TßRI cross-talk in diffuse intrinsic pontine glioma.

Diffuse intrinsic pontine glioma (DIPG) is a lethal pediatric brain cancer whose median survival time is under one year. The possible roles of the two most common DIPG associated cytoplasmic ACVR1 receptor kinase domain mutants, G328V and R206H, are reexamined in the context of new biochemical results regarding their intrinsic relative ATPase activities. At 37 °C, the G328V mutant displays a 1.8-fold increase in intrinsic kinase activity over wild-type, whereas the R206H mutant shows similar activity. The higher G328V mutant intrinsic kinase activity is consistent with the statistically significant longer overall survival times of DIPG patients harboring ACVR1 G328V tumors. Based on the potential cross-talk between ACVR1 and TßRI pathways and known and predicted off-targets of ACVR1 inhibitors, we further validated the inhibition effects of several TßRI inhibitors on ACVR1 wild-type and G328V mutant patient tumor derived DIPG cell lines at 20-50 µM doses. SU-DIPG-IV cells harboring the histone H3.1K27M and activating ACVR1 G328V mutations appeared to be less susceptible to TßRI inhibition than SF8628 cells harboring the H3.3K27M mutation and wild-type ACVR1. Thus, inhibition of hidden oncogenic signaling pathways in DIPG such as TßRI that are not limited to ACVR1 itself may provide alternative entry points for DIPG therapeutics.

MEDICASCY: A Machine Learning Approach for Predicting Small-Molecule Drug Side Effects, Indications, Efficacy, and Modes of Action.

To improve the drug discovery yield, a method which is implemented at the beginning of drug discovery that accurately predicts drug side effects, indications, efficacy, and mode of action based solely on the input of the drug's chemical structure is needed. In contrast, extant predictive methods do not comprehensively address these aspects of drug discovery and rely on features derived from extensive, often unavailable experimental information for novel molecules. To address these issues, we developed MEDICASCY, a multilabel-based boosted random forest machine learning method that only requires the small molecule's chemical structure for the drug side effect, indication, efficacy, and probable mode of action target predictions; however, it has comparable or even significantly better performance than existing approaches requiring far more information. In retrospective benchmarking on high confidence predictions, MEDICASCY shows about 78% precision and recall for predicting at least one severe side effect and 72% precision drug efficacy. Experimental validation of MEDICASCY's efficacy predictions on novel molecules shows close to 80% precision for the inhibition of growth in ovarian, breast, and prostate cancer cell lines. Thus, MEDICASCY should improve the success rate for new drug approval. A web service for academic users is available at http://pwp.gatech.edu/cssb/MEDICASCY.

Crystal Structure of α-Xylosidase from Aspergillus niger in Complex with a Hydrolyzed Xyloglucan Product and New Insights in Accurately Predicting Substrate Specificities of GH31 Family Glycosidases.

Glycoside hydrolase family 31 (GH31) enzymes show both highly conserved folds and catalytic residues. Yet different members of GH31 show very different substrate specificities, and it is not obvious how these specificities arise from the protein sequences. The fungal α-xylosidase, AxlA, was originally isolated from a commercial enzyme mixture secreted by Aspergillus niger and was reported to have potential as a catalytic component in biomass deconstruction in the biofuel industry. We report here the crystal structure of AxlA in complex with its catalytic product, a hydrolyzed xyloglucan oligosaccharide. On the basis of our new structure, we provide the structural basis for AxlA's role in xyloglucan utilization and, more importantly, a new procedure to predict and differentiate C5 vs C6 sugar specific activities based on protein sequences of the functionally diverse GH31 family enzymes.

Target highlights in CASP13: Experimental target structures through the eyes of their authors.

The functional and biological significance of selected CASP13 targets are described by the authors of the structures. The structural biologists discuss the most interesting structural features of the target proteins and assess whether these features were correctly reproduced in the predictions submitted to the CASP13 experiment.

Time-resolved x-ray crystallography capture of a slow reaction tetrahydrofolate intermediate.

Time-resolved crystallography is a powerful technique to elucidate molecular mechanisms at both spatial (angstroms) and temporal (picoseconds to seconds) resolutions. We recently discovered an unusually slow reaction at room temperature that occurs on the order of days: the in crystalline reverse oxidative decay of the chemically labile (6S)-5,6,7,8-tetrahydrofolate in complex with its producing enzyme Escherichia coli dihydrofolate reductase. Here, we report the critical analysis of a representative dataset at an intermediate reaction time point. A quinonoid-like intermediate state lying between tetrahydrofolate and dihydrofolate features a near coplanar geometry of the bicyclic pterin moiety, and a tetrahedral sp 3 C6 geometry is proposed based on the apparent mFo-DFc omit electron densities of the ligand. The presence of this intermediate is strongly supported by Bayesian difference refinement. Isomorphous Fo-Fo difference map and multi-state refinement analyses suggest the presence of end-state ligand populations as well, although the putative intermediate state is likely the most populated. A similar quinonoid intermediate previously proposed to transiently exist during the oxidation of tetrahydrofolate was confirmed by polarography and UV-vis spectroscopy to be relatively stable in the oxidation of its close analog tetrahydropterin. We postulate that the constraints on the ligand imposed by the interactions with the protein environment might be the origin of the slow reaction observed by time-resolved crystallography.

The crystal structure of a tetrahydrofolate-bound dihydrofolate reductase reveals the origin of slow product release.

Dihydrofolate reductase (DHFR) catalyzes the stereospecific reduction of 7,8-dihydrofolate (FH2) to (6s)-5,6,7,8-tetrahydrofolate (FH4) via hydride transfer from NADPH. The consensus Escherichia coli DHFR mechanism involves conformational changes between closed and occluded states occurring during the rate-limiting product release step. Although the Protein Data Bank (PDB) contains over 250 DHFR structures, the FH4 complex structure responsible for rate-limiting product release is unknown. We report to our knowledge the first crystal structure of an E. coli. DHFR:FH4 complex at 1.03 Å resolution showing distinct stabilizing interactions absent in FH2 or related (6R)-5,10-dideaza-FH4 complexes. We discover the time course of decay of the co-purified endogenous FH4 during crystal growth, with conversion from FH4 to FH2 occurring in 2-3 days. We also determine another occluded complex structure of E. coli DHFR with a slow-onset nanomolar inhibitor that contrasts with the methotrexate complex, suggesting a plausible strategy for designing DHFR antibiotics by targeting FH4 product conformations.

FINDSITEcomb2.0: A New Approach for Virtual Ligand Screening of Proteins and Virtual Target Screening of Biomolecules.

Computational approaches for predicting protein-ligand interactions can facilitate drug lead discovery and drug target determination. We have previously developed a threading/structural-based approach, FINDSITEcomb, for the virtual ligand screening of proteins that has been extensively experimentally validated. Even when low resolution predicted protein structures are employed, FINDSITEcomb has the advantage of being faster and more accurate than traditional high-resolution structure-based docking methods. It also overcomes the limitations of traditional QSAR methods that require a known set of seed ligands that bind to the given protein target. Here, we further improve FINDSITEcomb by enhancing its template ligand selection from the PDB/DrugBank/ChEMBL libraries of known protein-ligand interactions by (1) parsing the template proteins and their corresponding binding ligands in the DrugBank and ChEMBL libraries into domains so that the ligands with falsely matched domains to the targets will not be selected as template ligands; (2) applying various thresholds to filter out falsely matched template structures in the structure comparison process and thus their corresponding ligands for template ligand selection. With a sequence identity cutoff of 30% of target to templates and modeled target structures, FINDSITEcomb2.0 is shown to significantly improve upon FINDSITEcomb on the DUD-E benchmark set by increasing the 1% enrichment factor from 16.7 to 22.1, with a p-value of 4.3 × 10-3 by the Student t-test. With an 80% sequence identity cutoff of target to templates for the DUD-E set and modeled target structures, FINDSITEcomb2.0, having a 1% ROC enrichment factor of 52.39, also outperforms state-of-the-art methods that employ machine learning such as a deep convolutional neural network, CNN, with an enrichment of 29.65. Thus, FINDSITEcomb2.0 represents a significant improvement in the state-of-the-art. The FINDSITEcomb2.0 web service is freely available for academic users at http://pwp.gatech.edu/cssb/FINDSITE-COMB-2 .

Structure and specificity of a permissive bacterial C-prenyltransferase.

This study highlights the biochemical and structural characterization of the L-tryptophan C6 C-prenyltransferase (C-PT) PriB from Streptomyces sp. RM-5-8. PriB was found to be uniquely permissive to a diverse array of prenyl donors and acceptors including daptomycin. Two additional PTs also produced novel prenylated daptomycins with improved antibacterial activities over the parent drug.

OleD Loki as a Catalyst for Tertiary Amine and Hydroxamate Glycosylation.

We describe the ability of an engineered glycosyltransferase (OleD Loki) to catalyze the N-glycosylation of tertiary-amine-containing drugs and trichostatin hydroxamate glycosyl ester formation. As such, this study highlights the first bacterial model catalyst for tertiary-amine N-glycosylation and further expands the substrate scope and synthetic potential of engineered OleDs. In addition, this work could open the door to the discovery of similar capabilities among other permissive bacterial glycosyltransferases.

Crystal Structures of SgcE6 and SgcC, the Two-Component Monooxygenase That Catalyzes Hydroxylation of a Carrier Protein-Tethered Substrate during the Biosynthesis of the Enediyne Antitumor Antibiotic C-1027 in Streptomyces globisporus.

C-1027 is a chromoprotein enediyne antitumor antibiotic produced by Streptomyces globisporus. In the last step of biosynthesis of the (S)-3-chloro-5-hydroxy-ß-tyrosine moiety of the C-1027 enediyne chromophore, SgcE6 and SgcC compose a two-component monooxygenase that hydroxylates the C-5 position of (S)-3-chloro-ß-tyrosine. This two-component monooxygenase is remarkable for two reasons. (i) SgcE6 specifically reacts with FAD and NADH, and (ii) SgcC is active with only the peptidyl carrier protein (PCP)-tethered substrate. To address the molecular details of substrate specificity, we determined the crystal structures of SgcE6 and SgcC at 1.66 and 2.63 Å resolution, respectively. SgcE6 shares a similar ß-barrel fold with the class I HpaC-like flavin reductases. A flexible loop near the active site of SgcE6 plays a role in FAD binding, likely by providing sufficient space to accommodate the AMP moiety of FAD, when compared to that of FMN-utilizing homologues. SgcC shows structural similarity to a few other known FADH2-dependent monooxygenases and sheds light on some biochemically but not structurally characterized homologues. The crystal structures reported here provide insights into substrate specificity, and comparison with homologues provides a catalytic mechanism of the two-component, FADH2-dependent monooxygenase (SgcE6 and SgcC) that catalyzes the hydroxylation of a PCP-tethered substrate.

Structure of the ent-Copalyl Diphosphate Synthase PtmT2 from Streptomyces platensis CB00739, a Bacterial Type II Diterpene Synthase.

Terpenoids are the largest and most structurally diverse family of natural products found in nature, yet their presence in bacteria is underappreciated. The carbon skeletons of terpenoids are generated through carbocation-dependent cyclization cascades catalyzed by terpene synthases (TSs). Type I and type II TSs initiate cyclization via diphosphate ionization and protonation, respectively, and protein structures of both types are known. Most plant diterpene synthases (DTSs) possess three α-helical domains (αßγ), which are thought to have arisen from the fusion of discrete, ancestral bacterial type I TSs (α) and type II TSs (ßγ). Type II DTSs of bacterial origin, of which there are no structurally characterized members, are a missing piece in the structural evolution of TSs. Here, we report the first crystal structure of a type II DTS from bacteria. PtmT2 from Streptomyces platensis CB00739 was verified as an ent-copalyl diphosphate synthase involved in the biosynthesis of platensimycin and platencin. The crystal structure of PtmT2 was solved at a resolution of 1.80 Å, and docking studies suggest the catalytically active conformation of geranylgeranyl diphosphate (GGPP). Site-directed mutagenesis confirmed residues involved in binding the diphosphate moiety of GGPP and identified DxxxxE as a potential Mg(2+)-binding motif for type II DTSs of bacterial origin. Finally, both the shape and physicochemical properties of the active sites are responsible for determining specific catalytic outcomes of TSs. The structure of PtmT2 fundamentally advances the knowledge of bacterial TSs, their mechanisms, and their role in the evolution of TSs.

Structural dynamics of a methionine γ-lyase for calicheamicin biosynthesis: Rotation of the conserved tyrosine stacking with pyridoxal phosphate.

CalE6 from Micromonospora echinospora is a (pyridoxal 5' phosphate) PLP-dependent methionine γ-lyase involved in the biosynthesis of calicheamicins. We report the crystal structure of a CalE6 2-(N-morpholino)ethanesulfonic acid complex showing ligand-induced rotation of Tyr100, which stacks with PLP, resembling the corresponding tyrosine rotation of true catalytic intermediates of CalE6 homologs. Elastic network modeling and crystallographic ensemble refinement reveal mobility of the N-terminal loop, which involves both tetrameric assembly and PLP binding. Modeling and comparative structural analysis of PLP-dependent enzymes involved in Cys/Met metabolism shine light on the functional implications of the intrinsic dynamic properties of CalE6 in catalysis and holoenzyme maturation.

Structural Basis for the Stereochemical Control of Amine Installation in Nucleotide Sugar Aminotransferases.

Sugar aminotransferases (SATs) are an important class of tailoring enzymes that catalyze the 5'-pyridoxal phosphate (PLP)-dependent stereo- and regiospecific installation of an amino group from an amino acid donor (typically L-Glu or L-Gln) to a corresponding ketosugar nucleotide acceptor. Herein we report the strategic structural study of two homologous C4 SATs (Micromonospora echinospora CalS13 and Escherichia coli WecE) that utilize identical substrates but differ in their stereochemistry of aminotransfer. This study reveals for the first time a new mode of SAT sugar nucleotide binding and, in conjunction with previously reported SAT structural studies, provides the basis from which to propose a universal model for SAT stereo- and regiochemical control of amine installation. Specifically, the universal model put forth highlights catalytic divergence to derive solely from distinctions within nucleotide sugar orientation upon binding within a relatively fixed SAT active site where the available ligand bound structures of the three out of four representative C3 and C4 SAT examples provide a basis for the overall model. Importantly, this study presents a new predictive model to support SAT functional annotation, biochemical study and rational engineering.

LucY: A Versatile New Fluorescent Reporter Protein.

We report on the discovery, isolation, and use of a novel yellow fluorescent protein. Lucigen Yellow (LucY) binds one FAD molecule within its core, thus shielding it from water and maintaining its structure so that fluorescence is 10-fold higher than freely soluble FAD. LucY displays excitation and emission spectra characteristic of FAD, with 3 excitation peaks at 276 nm, 377 nm, and 460 nm and a single emission peak at 530 nm. These excitation and emission maxima provide the large Stokes shift beneficial to fluorescence experimentation. LucY belongs to the MurB family of UDP-N-acetylenolpyruvylglucosamine reductases. The high resolution crystal structure shows that in contrast to other structurally resolved MurB enzymes, LucY does not contain a potentially quenching aromatic residue near the FAD isoalloxazine ring, which may explain its increased fluorescence over related proteins. Using E. coli as a system in which to develop LucY as a reporter, we show that it is amenable to circular permutation and use as a reporter of protein-protein interaction. Fragmentation between its distinct domains renders LucY non-fluorescent, but fluorescence can be partially restored by fusion of the fragments to interacting protein domains. Thus, LucY may find application in Protein-fragment Complementation Assays for evaluating protein-protein interactions.

The reductive half-reaction of xanthine dehydrogenase from Rhodobacter capsulatus: the role of Glu232 in catalysis.

The kinetic properties of an E232Q variant of the xanthine dehydrogenase from Rhodobacter capsulatus have been examined to ascertain whether Glu(232) in wild-type enzyme is protonated or unprotonated in the course of catalysis at neutral pH. We find that kred, the limiting rate constant for reduction at high [xanthine], is significantly compromised in the variant, a result that is inconsistent with Glu(232) being neutral in the active site of the wild-type enzyme. A comparison of the pH dependence of both kred and kred/Kd from reductive half-reaction experiments between wild-type and enzyme and the E232Q variant suggests that the ionized Glu(232) of wild-type enzyme plays an important role in catalysis by discriminating against the monoanionic form of substrate, effectively increasing the pKa of substrate by two pH units and ensuring that at physiological pH the neutral form of substrate predominates in the Michaelis complex. A kinetic isotope study of the wild-type R. capsulatus enzyme indicates that, as previously determined for the bovine and chicken enzymes, product release is principally rate-limiting in catalysis. The disparity in rate constants for the chemical step of the reaction and product release, however, is not as great in the bacterial enzyme as compared with the vertebrate forms. The results indicate that the bacterial and bovine enzymes catalyze the chemical step of the reaction to the same degree and that the faster turnover observed with the bacterial enzyme is due to a faster rate constant for product release than is seen with the vertebrate enzyme.

Structure and substrate specificity of a eukaryotic fucosidase from Fusarium graminearum.

The secreted glycoside hydrolase family 29 (GH29) α-L-fucosidase from plant pathogenic fungus Fusarium graminearum (FgFCO1) actively releases fucose from the xyloglucan fragment. We solved crystal structures of two active-site conformations, i.e. open and closed, of apoFgFCO1 and an open complex with product fucose at atomic resolution. The closed conformation supports catalysis by orienting the conserved general acid/base Glu-288 nearest the predicted glycosidic position, whereas the open conformation possibly represents an unreactive state with Glu-288 positioned away from the catalytic center. A flexible loop near the substrate binding site containing a non-conserved GGSFT sequence is ordered in the closed but not the open form. We also identified a novel C-terminal ßγ-crystallin domain in FgFCO1 devoid of calcium binding motif whose homologous sequences are present in various glycoside hydrolase families. N-Glycosylated FgFCO1 adopts a monomeric state as verified by solution small angle x-ray scattering in contrast to reported multimeric fucosidases. Steady-state kinetics shows that FgFCO1 prefers α1,2 over α1,3/4 linkages and displays minimal activity with p-nitrophenyl fucoside with an acidic pH optimum of 4.6. Despite a retaining GH29 family fold, the overall specificity of FgFCO1 most closely resembles inverting GH95 α-fucosidase, which displays the highest specificity with two natural substrates harboring the Fucα1-2Gal glycosidic linkage, a xyloglucan-derived nonasaccharide, and 2'-fucosyllactose. Furthermore, FgFCO1 hydrolyzes H-disaccharide (lacking a +2 subsite sugar) at a rate 10(3)-fold slower than 2'-fucosyllactose. We demonstrated the structurally dynamic active site of FgFCO1 with flexible general acid/base Glu, a common feature shared by several bacterial GH29 fucosidases to various extents.

X-ray crystal structure of a xanthine oxidase complex with the flavonoid inhibitor quercetin.

Xanthine oxidase catalyzes the sequential hydroxylation of hypoxanthine to uric acid via xanthine as intermediate. Deposition of crystals of the catalytic product uric acid or its monosodium salt in human joints with accompanying joint inflammation is the major cause of gout. Natural flavonoids are attractive leads for rational design of preventive and therapeutic xanthine oxidase inhibitors due to their beneficial antioxidant, anti-inflammatory, and antiproliferative activities in addition to their micromolar inhibitory activities toward xanthine oxidase. We determined the first complex X-ray structure of mammalian xanthine oxidase with the natural flavonoid inhibitor quercetin at 2.0 Å resolution. The inhibitor adopts a single orientation with its benzopyran moiety sandwiched between Phe 914 and Phe 1009 and ring B pointing toward the solvent channel leading to the molybdenum active center. The favorable steric complementarity of the conjugated three-ring structure of quercetin with the active site and specific hydrogen-bonding interactions of exocyclic hydroxy groups with catalytically relevant residues Arg 880 and Glu 802 correlate well with a previously reported structure-activity relationship of flavonoid inhibitors of xanthine oxidase. The current complex provides a structural basis for the rational design of flavonoid-type inhibitors against xanthine oxidase useful for the treatment of hyperuricemia, gout, and inflammatory disease states.

Substrate orientation and specificity in xanthine oxidase: crystal structures of the enzyme in complex with indole-3-acetaldehyde and guanine.

Xanthine oxidase is a molybdenum-containing hydroxylase that catalyzes the hydroxylation of sp(2)-hybridized carbon centers in a variety of aromatic heterocycles as well as aldehydes. Crystal structures of the oxidase form of the bovine enzyme in complex with a poor substrate indole-3-acetaldehyde and the nonsubstrate guanine have been determined, both at a resolution of 1.6 Å. In each structure, a specific and unambiguous orientation of the substrate in the active site is observed in which the hydroxylatable site is oriented away from the active site molybdenum center. The orientation seen with indole-3-acetaldehyde has the substrate positioned with the indole ring rather than the exocyclic aldehyde nearest the molybdenum center, indicating that the substrate must rotate some 30° in the enzyme active site to permit hydroxylation of the aldehyde group (as observed experimentally), accounting for the reduced reactivity of the enzyme toward this substrate. The principal product of hydroxylation of indole-3-acetaldehyde by the bovine enzyme is confirmed to be indole-3-carboxylic acid based on its characteristic UV-vis spectrum, and the kinetics of enzyme reduction are reported. With guanine, the dominant orientation seen crystallographically has the C-8 position that might be hydroxylated pointed away from the active site molybdenum center, in a configuration resembling that seen previously with hypoxanthine (a substrate that is effectively hydroxylated at position 2). The ∼180° reorientation required to permit reaction is sterically prohibited, indicating that substrate (mis)orientation in the active site is a major factor precluding formation of the highly mutagenic 8-hydroxyguanine.

X-ray crystal structure of arsenite-inhibited xanthine oxidase: µ-sulfido,µ-oxo double bridge between molybdenum and arsenic in the active site.

Xanthine oxidoreductase is a molybdenum-containing enzyme that catalyzes the hydroxylation reaction of sp(2)-hybridized carbon centers of a variety of substrates, including purines, aldehydes, and other heterocyclic compounds. The complex of arsenite-inhibited xanthine oxidase has been characterized previously by UV-vis, electron paramagnetic resonance, and X-ray absorption spectroscopy (XAS), and the catalytically essential sulfido ligand of the square-pyrimidal molybdenum center has been suggested to be involved in arsenite binding through either a µ-sulfido,µ-oxo double bridge or a single µ-sulfido bridge. However, this is contrary to the crystallographically observed single µ-oxo bridge between molybdenum and arsenic in the desulfo form of aldehyde oxidoreductase from Desulfovibrio gigas (an enzyme closely related to xanthine oxidase), whose molybdenum center has an oxo ligand replacing the catalytically essential sulfur, as seen in the functional form of xanthine oxidase. Here we use X-ray crystallography to characterize the molybdenum center of arsenite-inhibited xanthine oxidase and solve the structures of the oxidized and reduced inhibition complexes at 1.82 and 2.11 Å resolution, respectively. We observe µ-sulfido,µ-oxo double bridges between molybdenum and arsenic in the active sites of both complexes. Arsenic is four-coordinate with a distorted trigonal-pyramidal geometry in the oxidized complex and three-coordinate with a distorted trigonal-planar geometry in the reduced complex. The doubly bridged binding mode is in agreement with previous XAS data indicating that the catalytically essential sulfur is also essential for the high affinity of reduced xanthine oxidoreductase for arsenite.