Kynurenine 3-monooxygenase is a FAD-dependent aromatic hydroxylase (FAH) which is a widely suggested therapeutic target for controlling the balance of bioactive metabolite levels produced by the mammalian kynurenine pathway (KP). Prior to starting this work no structural information was known for the enzyme, with studies of the human form complicated by the presence of a C-terminal transmembrane helix. The bacterial Pseudomonas fluorescens enzyme (PfKMO) lacks the transmembrane region and has been previously characterised by Crozier-Reabe and Moran [1, 2]. Therefore PfKMO, which shares 32 % sequence identity with the human enzyme, was selected as a target for structure solution. Initial substrate bound PfKMO crystals showed poor X-ray diffraction. Subsequent growth optimisation and the generation of a C252S/C461S PfKMO mutant (dm2) yielded crystals suitable for structure solution. Selenomethioninelabelled substrate bound dm2 crystals were used to solve the first structure to a resolution of 3.40 Å. With just one protein molecule per asymmetric unit, a high solvent content was responsible for the poor diffraction properties of this crystal form. The overall fold resembled that of other FAH enzymes with a Rossmann-fold based FADbinding domain above a buried substrate binding pocket. Interestingly PfKMO possesses an additional, novel C-terminal domain that caps the back of the substrate-binding pocket on the opposite side to the flavin. Residues proposed to be involved in substrate binding were identified and shown to be highly conserved among mammalian KMO sequences. Subsequently single crystals of substrate-free dm2 PfKMO were obtained and showed significantly stronger diffraction due to new lattice packing in an orthorhombic space group bearing four molecules per asymmetric unit. The structure was solved to a resolution of 2.26 Å and revealed a clear conformational change of the novel C-terminal domain. This movement opens a potential route of substrate/product exchange between bulk solvent and the active site. The investigation of a set of C-terminal mutants further explored the relevance and mechanics of the conformational change. In addition the presence of chloride ions in the substrate-free crystal growth solution caused a small number of localised subtle alterations to the structure, with a potential chloride binding site identified adjacent to the flavin cofactor. This may have relevance for the observed inhibition of PfKMO activity by monovalent anions – a feature widely common to FAH enzymes [3]. The first discovered KMO inhibitors were analogues of the substrate L-Kyn, however one such compound (m-NBA) was recently shown to instigate uncoupled NADPH oxidation leading to the release of cytotoxic hydrogen peroxide [1]. A set of substrate analogues were tested and characterised for inhibition of PfKMO. The picture was shown to be complex as some substrate analogues completely inhibited the enzyme whilst the binding of some still stimulated low-levels of NADPH oxidation. Crystallographic studies with m-NBA and 3,4-dichlorobenzoylalanine (3,4-CBA) bound revealed indistinguishable structures from that of substrate-bound PfKMO. These studies suggest that the analogue 3,4CBA is a potent PfKMO inhibitor whose therapeutic potential may be re-visited. The previous most potent KMO inhibitor whose structure was not analogous to the substrate was Ro 61-8048 [4], which unfortunately did not pass pre-clinical safety tests. A novel series of 1,2,4-oxadiazole amides based on the structure of Ro 61-8048 was created by Gavin Milne (PhD, University of St Andrews) and tested on PfKMO. Rounds of refinement led to the discovery and refinement of low nanomolar competitive inhibitors of the bacterial enzyme. PfKMO was co-crystallised with each of the four most potent compounds forming a third different lattice arrangement, which yielded structures to resolutions of 2.15-2.40 Å. The structures displayed conformational changes resembling the substrate-free fold possibly caused by displacement of a crucial substrate-binding residue, R84. Overall the wealth of structural data obtained may be transferable to predictions about the structural features of human KMO and to the rational design of therapeutic inhibitors. The potent novel inhibitors tested may additionally present a new exciting development for the therapeutic inhibition of human KMO.

Kynurenine 3-monooxygenase (KMO) is a flavoprotein which has been implicated in Huntington’s disease, Alzheimer’s disease and acute pancreatitis. Recently there has been important research published about this enzyme including the structure of a truncated Saccharomyces cerevisiae KMO enzyme and KMO inhibition studies in animal models of disease. In previous work from this research group the complete Pseudomonas fluorescens KMO enzyme has been successfully crystallised both with and without the substrate, L-kynurenine, from which significant insights were gained into function and the potential role of domain movement. To examine substrate binding in KMO and to consolidate previous structural studies, key residues in the active site have been investigated using site directed mutagenesis, crystallography and kinetic analysis using steady-state techniques. This analysis has identified the interactions between the enzyme and the substrate and provides a basis for inhibitor design. The residues implicated in substrate binding are N369, Y404 and R84. For N369 and Y404, minor changes to the amino acid in the mutations N369S and Y404F were shown to cause a decrease in binding affinity of the substrate but the enzyme remained active. For the mutations Y404A and R84K enzyme activity was significantly affected. Crystal structures of N369S, Y404F and R84K were also obtained. Another residue in the active site studied was H320 which is the only amino acid to differ in the active sites of the human and Pseudomonas fluorescens enzymes. This residue was therefore of interest to determine whether the bacterial enzyme used in this work is likely to be a good model for the human enzyme, which has not yet been successfully isolated in significant quantities for in vitro research. Modifying this residue to obtain H320F KMO revealed that this residue does not have a significant role in substrate binding. Potent inhibitor molecules have been studied with this enzyme and shown in kinetic assays to have nanomolar Ki values. These inhibitors are the most potent inhibitors studied with Pseudomonas fluorescens to date and continue previous inhibitor studies carried out with this enzyme. This group of inhibitors contain different substituents in the part of the molecule shown to bind closest to the C-terminal domain of the protein. These novel inhibitors do not allow the flavin to be reduced by NADPH (which results in unwanted peroxide production) unlike a number of previously studied molecules and therefore have the potential to be clinically useful. This research therefore answers many questions about this enzyme, in particular about the role of particular residues in the active site, substrate recognition and inhibition of this important drug target.

Kynurenine 3-monooxygenase (KMO) is an enzyme central to the kynurenine pathway of tryptophan degradation. KMO is emerging as an increasingly important target for drug development. The enzyme is implicated in the development and progression of several neurodegenerative disorders, in the regulation of the immune response and in sterile systemic inflammation. Production of recombinant human enzyme is challenging due to the presence of transmembrane domains, which localise KMO to the outer mitochondrial membrane and render KMO insoluble in many in vitro expression systems. Although several in vitro KMO assay techniques have been reported in the literature these methods are typically insensitive or require purified protein for use in high-throughput screening assays of human KMO enzyme. The first report of bacterial expression of soluble active human KMO enzyme is described here. Fusion protein tags were used to optimise soluble expression and enable characterisation and partial purification of the active protein constructs. Functional enzyme was used to develop several novel high-throughput drug screening techniques for the discovery of inhibitors specifically targeting human KMO. These screening techniques were fully characterised and validated using known KMO inhibitors from the patent literature. One of the novel KMO assay techniques was implemented for compound screening and several hit compounds were identified, validated and their in vitro DMPK characteristics determined. In addition to assay development, KMO was characterised at the cellular level when overexpressed in HEK293 cells. These experiments indicated that KMO overexpressing cells undergo bidirectional adaptation via alteration of kynurenine pathway homeostasis. As a result, these cells are protected from cytotoxicity mediated by 3-hydroxykynurenine (3-HK), the toxic product of KMO catalysis. The development of novel high throughput screening techniques targeting KMO has enabled screening of potential new inhibitors specifically targeting the human enzyme. Implementation of these screening assays will allow accelerated and improved discovery and development of novel KMO inhibitors for the potential treatment of numerous disease states.

BACKGROUND: Acute pancreatitis (AP) is a sterile initiator of systemic inflammation that can trigger multiple organ dysfunction syndrome (MODS). In the acute phase of AP, the kynurenine pathway of tryptophan metabolism plays an important role in the genesis of AP-MODS in experimental animal models, but it is unknown whether the pathway is activated in human AP. Human data are required to support the rationale for kynurenine 3- monooxygenase (KMO) inhibition as a treatment for AP-MODS and reinforce the translational potential. Additionally, as respiratory dysfunction is frequent in severe AP, the role of lung ultrasonography in severity stratification deserves investigation. Furthermore, the effect of AP-MODS on long-term survival is unknown. OBJECTIVES: My objectives were to: 1) Define the temporal and quantitative relationship of kynurenine metabolites with the onset and severity of APMODS, 2) Investigate the value of lung ultrasonography in the early diagnosis of respiratory dysfunction in human AP-MODS, and 3) Examine whether early AP-MODS impacts on long-term survival. METHODS: 1) A prospective, observational, clinical experimental medicine study titled “Inflammation, Metabolism, and Organ Failure in Acute Pancreatitis” (IMOFAP) was performed. For 90 days, consecutive patients with a potential diagnosis of AP were recruited and venous blood was sampled at 0, 3, 6, 12, 24, 48, 72 and 168 hours post-recruitment. Kynurenine metabolite concentrations were measured by liquid chromatography–tandem mass spectrometry (LC-MS/MS) and analysed in the context of clinical data, disease severity indices, and cytokine profiles. 2) In a nested cohort within IMOFAP, 41 participants underwent lung ultrasonography to evaluate whether this imaging modality can detect respiratory dysfunction in AP. 3) Survival data for a prospectively maintained database of patients with AP was analysed after accounting for in-hospital deaths. RESULTS: 1) During the IMOFAP study, 79 patients were recruited with an elevated serum amylase, of which 57 patients met the diagnostic criteria for AP; 9 had severe disease. Temporal profiling revealed early tryptophan depletion and contemporaneous elevation of plasma concentrations of 3- hydroxykynurenine, which paralleled systemic inflammation and AP severity. 2) Lung ultrasonography findings correlated with respiratory dysfunction. 3) 694 patients were followed up for a median of 8.8 years. AP-MODS conferred a deleterious effect on overall survival which persisted after the exclusion of inhospital deaths (10.0 years, 95% C.I. = 9.4-10.6 years) compared to AP without MODS (11.6 years, 95% C.I. = 11.2-11.9 years; P = 0.001). This effect was independent of age. CONCLUSIONS: In the acute phase of AP, metabolic flux through KMO is elevated and proportionate to AP severity. Lung ultrasonography may be a useful technique for evaluating AP-MODS. AP-MODS is an independent predictor of long-term mortality. Together, this work reinforces the rationale for investigating early phase KMO inhibition as a therapeutic strategy in humans.

The kynurenine pathway produces NAD+ from L-tryptophan. Metabolites known as the kynurenines are produced within the pathway. The effects of the kynurenines have been associated with a number of diseases including cancer, Alzheimer’s disease, Huntington’s disease, and acute pancreatitis. Kynurenine monooxygenase (KMO) is an enzyme that catalyses the conversion of L-kynurenine to 3-hydroxy-L-kynurenine, the downstream product of which is the neurotoxic quinolinic acid. L-kynurenine is positioned at a branching point within the pathway. Metabolism via KMO leads to quinolinic acid production whereas conversion via kynurenine aminotransferase (KAT) produces the neuroprotective kynurenic acid. Inhibition of KMO leads to an increase in kynurenic acid concentration. This has also been shown to ameliorate the symptoms of neurological diseases in a number of animal models as well as to protect against multiple organ dysfunction caused by acute pancreatitis in rodent models. These findings present KMO as a promising drug target. Due to the hydrophobic nature of human KMO (hKMO) it has been necessary to utilise other forms of KMO as models. Past studies have produced crystal structures of a truncated Saccharomyces cerevisiae KMO and of Pseudomonas fluorescens KMO (PfKMO). Previous work in this research group has resulted in the structure of variants of PfKMO bound to either inhibitor molecules or substrate. These structures identified residues involved in substrate binding and the presence of a highly mobile section of the C-terminus, giving rise to open and closed conformations. It was surmised the movement of the C-terminus was dependent upon the presence of substrate and an interactive network between the C-terminus and the rest of the protein. Using improved crystallising conditions high-resolution structures of PfKMO have been produced that allow for further study of residues involved in substrate binding and the interactive network within the C-terminus. The mutants R84K and Y404F showed severely decreased enzyme activity. Crystal structures of these proteins showed disrupted interactions between substrate and active site. These findings underline the importance of residues R84 and Y404 in substrate binding. An H320F mutation gives an analogous active site to hKMO. Crystallographic and kinetic study of this mutant proved very similar to PfKMO, supporting the use of PfKMO as a model for hKMO. Throughout the work each structure had a P21221 space group with two molecules in the asymmetric unit. The presence of an open and closed molecule within each structure, including substrate-free molecules refuted the connection between C-terminus and substrate. R386K and E372T mutations were separately introduced in order to interrupt the interactive network. The presence of both open and closed conformations in the structures of R386K and E372T refutes the necessity for the interactive network in C-terminus movement. The data analysed throughout the project suggest simple mobility and thermal motion as the cause of the movement of the C-terminus. This work, in conjunction with kinetic data from the thesis of Helen Bell, presents structural data to characterise the role of binding residues within the active site of KMO as well as the mechanistic role of the C-terminus. It also highlights the importance of certain binding residues and countered the previously held hypotheses surrounding the significance of the C-terminus. The mechanistic role of the C-terminus therefore remains unclear and requires further study.

京都大学 / 0048 / 新制・課程博士 / 博士(人間健康科学) / 甲第20296号 / 人健博第44号 / 新制||人健||4(附属図書館) / 京都大学大学院医学研究科人間健康科学系専攻 / (主査)教授 高桑 徹也, 教授 三谷 章, 教授 浅野 雅秀 / 学位規則第4条第1項該当 / Doctor of Human Health Sciences / Kyoto University / DFAM