Development of selective artificial metalloenzymes by combining the biological concepts for selective recognition with those of transition metal catalysis has received much attention during the last decade. Targeting covalent incorporation of organometallic catalysts into proteins, we explored site-selective covalent coupling of phosphane and N–containing ligands. The successful approach for incorporation of phosphane ligands we report herein consists of site-specific covalent coupling of a maleimide functionalized hydrazide into proteins, followed by coupling of aldehyde functionalized phosphanes via a hydrazone linkage. Site selective incorporation of N–containing ligands was obtained by coupling maleimide functionalized N–ligands to proteins via Michael addition to the maleimide double bond. These two methods can be easily applied to virtually any protein displaying a single reactive cysteine and allows a wide range of possibilities in terms of cofactor design. Site-specific covalent incorporation of transition metal complexes of phosphane ligands into proteins was successfully obtained. The success of the approach is influenced by several factors like the metal precursor, the phosphane type and the protein scaffold. Metal complexes of 5–maleimido–1,10–phenanthroline modified proteins were formed in situ, via addition of a metal precursor to the phenanthroline modified proteins or by coupling preformed metal complexes to proteins via Michael addition of the thiol group from a cysteine residue to the maleimide double bond of the N-ligand. These successful coupling methods enable the use of a wide range of protein structures as templates for the preparation of artificial transition metalloenzymes, which opens the way to full exploration of the power of selective molecular recognition of proteins in transition metal catalysis.

This thesis describes the design, synthesis and application of artificial metalloenzymes for transition metal catalysed reactions not performed by natural enzymes. Unique cysteine containing protein templates were covalently modified with transition metal ligand complexes that generate catalytic activity, which allows for the use of virtually any protein template. SCP-2L was selected as template for the linear hydrophobic tunnel that traverses the protein, which has high affinity for linear aliphatic molecules. The use of catalysts based on this protein to induce increased activity in the biphasic hydroformylation of linear α-olefins is investigated in this work. For this purpose, unique cysteine containing mutants of SCP-2L were modified with phosphine ligands by application of a novel bioconjugation procedure. Application of rhodium adducts of the phosphine modified protein constructs led to up to a 100 fold increase of the turn over numbers was measured compared to a Rh/TPPTS model system which is used in industry. Furthermore, good selectivity towards the linear product was observed. If it can be confirmed that the found catalytic results truly are the result of substrate encapsulation by the protein scaffold, this system represents the first rationally designed artificial metalloenzyme which exploits the shape selectivity of the protein scaffold to direct the outcome of a catalytic reaction. In addition, a study was performed for the development of enantioselective artificial metalloenzymes. Nitrogen ligands were covalently introduced in SCP-2L and the obtained conjugates were applied in the copper catalysed Diels-Alder and Michael addition reaction. A promising 25% ee was found for the Diels-Alder reaction between azachalcone and cyclopentadiene using one of the created constructs. Further development of these catalyst systems with the use of both synthetic (e.g. optimisation of ligand structure) and biomolecular tools (e.g. optimisation of protein environment) for optimisation can lead to very efficient and enantioselective conversions in the future.

Depuis la révolution industrielle, la chimie ne cesse de prospérer en développant des procédés de plus en plus performants souvent aux dépens de l’environnement. Dans le cadre du développement d’une chimie durable, des procédés catalytiques dans le domaine de la chimie d’oxydation sont mis en place en utilisant des métaux physiologiques et des oxydants doux. En combinant les avantages de la catalyse homogène et de la biocatalyse, de nouveaux catalyseurs bio-inspirés ont émergé, les métalloenzymes artificielles. Elles sont constituées d’un complexe inorganique, choisi en fonction de la réaction visée, qui est ancré au sein d’une protéine, qui apporte la sélectivité de la réaction. Au cours des travaux de cette thèse, de nouvelles métalloenzymes artificielles ont été créées par ancrage de divers complexes de Fe ou de Ru au sein de la protéine NikA. Dans un premier temps, l’hybride NikA/Ru-bpza a été synthétisé pour réaliser l’hydroxychloration d’alcènes en présence d’un iode hypervalent. Bien que d’excellentes propriétés catalytiques aient été obtenues, l’amélioration de la stabilité de ce type de catalyseurs, en particulier pour des réactions d’oxydation, reste un challenge important à relever pour leur utilisation au niveau industriel. Une des solutions originale est basée sur le développement de la catalyse hétérogène, en utilisant de cristaux de métalloenzymes artificielles grâce à la technologie CLEC (Cross-Linked Enzyme Crystals). Cette technologie permet, d’une part, d’améliorer la stabilité et la recyclabilité des catalyseurs, et d’autre part, d’élargir les conditions réactionnelles utilisées (solvants, pH, températures). Trois réactivités ont été développées à base de CLEC NikA/FeL : (i) la sulfoxydation de thioéthers, (ii) l’hydroxychloration d’alcènes en présence d’Oxone® et de chlore et (iii) la coupure oxydante d’alcènes par activation d’O2. Ces résultats ont permis d’explorer de nouvelles réactivités en chimie cascade soit en combinant les CLEC mis au point, soit en combinant différents catalyseurs homogènes. / Since the industrial revolution, chemistry has continually thriven by developing new efficient processes at the expense of the environment. As an example, oxidation reactions are performed under harsh conditions with the use of toxic oxidants. With the emergence of green chemistry, catalytic processes using physiological metals and soft oxidants are privileged. Combining the advantages of biocatalysis and homogeneous catalysis, the design of novel bioinspired catalysts, consisting on the synthesis of artificial enzymes has recently emerged. These hybrids are composed of an inorganic complex, driving the reactivity of the enzyme, inserted into a protein, which drives the reaction selectivity. The thesis described new developments in original artificial metalloenzymes, based on the use of the NikA protein and Fe or Ru catalysts. First, a new hybrid has been developed by anchoring the Ru-bpza complex to NikA to catalyze alkene hydroxychloration with hypervalent iodine. Although excellent catalytic efficiencies were obtained, the stability improvement remains a major challenge for the industrial use of these catalytic processes, especially when oxidation chemistry is concerned. One possible strategy is based on the development of heterogeneous catalysis, by using a crystal/solution version of the artificial metalloenzymes thank to the cross-linked enzyme crystals (CLEC) technology. On the one hand, this technology allows to increase the stability and the recyclability of the catalysts. On the other hand, catalysis can be performed under a various reactions conditions (organic solvent, temperature, pH). Three reactivities have been developed with NikA/FeL-CLEC catalysts: (i) thioether sulfoxidation with NaOCl, (ii) alkene hydroxychloration with Oxone® and chloride source and (iii) oxidative cleavage of alkenes by O2 activation. To go further, new reactivities in cascade reactions have been explored combining either NikA-based CLEC developed, or different homogenous catalysts.

Les protéines à radical S-adénosyl-L-méthionine (SAM) utilisent un centre [Fe4S4] réduit pour initier le clivage réductive homolytique de la SAM et la formation d'une espèce hautement réactive - le radical 5'-déoxyadénosyl ou 5'-dA•. Dans la quasi-totalité de cas ce radical alkyl va arracher un atome d'hydrogène sur le substrat et déclencher ainsi sa conversion en produit. On trouve ces enzymes au niveau d'étapes clé de la synthèse de certaines vitamines, antibiotiques, précurseurs de l'ADN ou encore cofacteurs protéiques où elles sont souvent impliquées dans le clivage ou la formation des liaisons C-C, C-N, C-S ou encore C-P. Les travaux réalisés au cours de cette thèse ont été focalisés sur l'étude structurale et fonctionnelle de la protéine HydE ; une enzyme à radical SAM, qui intervient dans la biosynthèse du site actif organométallique de l'hydrogénase à [FeFe]. L'objectif principal était d'identifier le substrat de HydE et d'étudier les détails du fonctionnement d'une protéine à radical SAM. Nous avons réussi à identifier un groupe de molécules, dérivées de la cystéine, contentant un cycle thiazolidine avec un ou deux groupements carboxylates, qui ont une très bonne affinité pour le site actif de HydE. Certains de ces ligands se sont montrés d’être des substrats non physiologiques de l’enzyme. Grâce à ces substrats nous avons pu mettre en évidence un nouveau mécanisme d’attaque radicalaire dans les protéines à radical SAM. En effet, dans HydE nous avons observé une attaque directe du radical 5'-dA• sur l’atome soufre du thioéther appartenant au cycle thiazolidine. Cette réaction constitue un exemple pas comme les autres d’une insertion d’un atome de soufre (ou de sélénium) catalysée par une enzyme à radical SAM. Il s'agit également d'une première observation d'une réaction radicalaire dans les cristaux protéiques d'une enzyme à radical SAM et également un premier suivi en temps réel par la RMN du 13C et 1H de l'accumulation d'un des produits de la réaction catalysée par ces enzymes. Les résultats de calculs théoriques basés sur nos structures cristallographiques de haute résolution suggèrent que dans le cas de cette superfamille de protéines le radical 5'-dA• serait plutôt un état de transition et donc pas une espèce intermédiaire isolable. / Radical S-adenosyl-L-methionine (SAM) proteins use a reduced [Fe4S4] cluster to initiate homolytic reductive cleavage of SAM, which leads to the formation of highly reactive 5'-deoxyadenosyl radical species or 5'-dA•. In almost all cases this alkyl radical will abstract a hydrogen atom from the substrate and thus trigger its conversion into product. These enzymes are found in key steps of the synthesis of certain vitamins, antibiotics, DNA precursors or protein cofactors. They are often involved in the cleavage or formation of C-C, C-N, C-S or C-P bonds. The present thesis work has been focused on the structural and functional study of HydE protein; a radical SAM enzyme, involved in the biosynthesis of the organometallic active site of [FeFe]-hydrogenase. The main goal was to identify the substrate of HydE and to study details of how radical SAM proteins control the highly oxidizing 5'-dA• species. We managed to identify a group of molecules, derived from cysteine, containing a thiazolidine ring with one or two carboxylate groups, which have a very good affinity for the active site of HydE. We have demonstrated some of these ligands are non-physiological substrates of the enzyme. With these substrates we could highlight a new radical attack mechanism in radical SAM proteins. Indeed, in HydE we observed a direct attack on the 5'-dA • radical on the sulfur atom of the thioether belonging to the thiazolidine ring. This is an unprecedented reaction that contrasts with sulfur (or selenium) atom insertion reactions catalysed by some radical SAM enzymes. This is also the first observation of a radical reaction in the protein crystal of a radical SAM enzyme and also the first real-time monitoring by 1H- & 13C-NMR spectroscopy of the accumulation of products of the reaction catalysed by these enzymes. Theoretical calculations based on our high-resolution crystal structures suggest that in the case of this protein superfamily the 5'-dA• radical, which triggers the reaction in radical SAM enzymes, is a transition state and therefore not an isolable intermediate species.

This thesis describes the synthesis, characterisation and application of artificial metalloenzymes as catalysts. The focus was on two mutants of SCP-2L (SCP-2L A100C and SCP-2L V83C) both of which possess a hydrophobic tunnel in which apolar substrates can accumulate. The crystal structure of SCP-2L A100C was determined and discussed with a special emphasis on its hydrophobic tunnel. The SCP-2L mutants were covalently modified at their unique cysteine with two different N-ligands (phenanthroline or dipicolylamine based) or three different phosphine ligands (all based on triphenylphosphine) in order to increase their binding capabilities towards metals. The metal binding capabilities of these artificial proteins towards different transition metals was determined. Phenanthroline modified SCP-2L was found to be a promising scaffold for Pd(II)-, Cu(II)-, Ni(II)- and Co(II)-enzymes while dipicolylamine-modified SCP-2L was found to be a promising scaffold for Pd(II)-enzymes. The rhodium binding capacity of two additional phosphine modified protein scaffolds was also investigated. Promising scaffolds for Rh(I)- and Ir(I)-enzymes were identified. Rh-enzymes of the phosphine modified proteins were tested in the aqueous-organic biphasic hydroformylation of linear long chain 1-alkenes and compared to the Rh/TPPTS reference system. Some Rh-enzymes were found to be several orders of magnitude more active than the model system while yielding comparable selectivities. The reason for this remarkable reactivity increase could not be fully elucidated but several potential modes of action could be excluded. Cu-, Co-, and Ni-enzymes of N-ligand modified SCP-2L A100C were tested in the asymmetric Diels-Alder reaction between cyclopentadiene and trans-azachalcone. A promising 29% ee for the exo-product was found for the phenanthroline modified protein in the presence of nickel. Further improvement of these catalyst systems by chemical means (e.g. optimisation of ligand structure) and bio-molecular tools (e.g. optimisation of protein environment) can lead to even more active and (enantio)selective catalysts in the future.