Bio-hybrid catalysts in asymmetric catalysis

Kasper van den Hurk

Van ’t Hoff Institute for Molecular Sciences

Supervisor: Joost N.H. Reek

Daily supervisor: FrédéricTerrade

Abstract

Bio-hybrid catalysts result from combining a catalytically active transition metal complex with a biomolecular host and lie at the interface between transition metal catalysis and biocatalysis.Until recently, transition metal catalysis and biocatalysishave developed in parallel as separate approaches towards achieving enantioselective transformations. The field of Bio-hybrid catalysis aims at harnessing second coordination sphere interactions to create transition metal complexes that display enzyme-like activities and selectivities.

In this review, the design strategies and the catalytic scope of the reported bio-hybrid catalysts are discussed.Many examples of enantioselective catalysis with bio-hybrid catalysts have been reported, showing promising results. However, for all reactions reported to date, good alternatives using conventional approaches are available. The main advantage that bio-hybrid catalysts have is that they allow for asymmetric catalysis in water. In the long term, it is imperative that the field starts to address some of the challenges in enantioselective catalysis, for which there are no alternatives available in conventional methods.

Introduction

The catalytic activity and selectivity of conventional transition metal catalysts are almost exclusively controlled by the first coordination sphere provided by the ligands. The dominant strategy towards obtaining enantiomerically pure compounds involves forcing an incoming reagent to approach from one prochiral face of a substrate, by sterically blocking the other side. The growing insight in organometallic chemistry and the development of optimisation techniques has provided us with very effective transition metal catalysts. However, transition metal catalystsremain inadequate for numerous chemical conversionsfor which high activity and selectivity cannot be reached.

In contrast, enzymes often outperform synthetic catalysts by employing highly efficient second-sphere substrate interactions provided by the biomolecular scaffold, such as hydrogen bonding and hydrophobic interactions. The second coordination sphere of an enzyme is an important contributor to the activity and selectivity by recognising and positioning substrates and by stabilizing intermediates.Enzymes are increasingly applied in synthetic processes and have proven to be very effective. Natural enzymes, however, evolved to catalyse very specific reactions with specific substrates and their scope could not always be expanded to the desired conversions. It is not always possible to produce sufficient amount of enzyme to use them for synthesis purpose and they are not always stable outside of their original organism.

Until recently, transition metal catalysis and biocatalysishave developed in parallel as separate approaches towards achieving enantioselective transformations. Over the last decade, however, significant efforts have been made to merge the attractive properties of natural and synthetic catalysts by designing transition metal catalyst systems based on the building blocks of nature. This field seeks to create transition metal catalysts with an enzyme-like second coordination sphere for reactions not found in nature.

In this review, various bio-hybrid transition metal catalysts and their applications in asymmetric catalysis are highlighted and the role of the second coordination sphere in the activity and the selectivity is discussed. There will be special attention for proteins acting as the biomolecular scaffold and supramolecular anchoring strategies. In addition, the potential of this field of research is evaluated.

Biomolecular scaffold

Catalysis in nature relies primarily on proteins. Nevertheless, oligonucleotides may offer an attractive scaffold for the incorporation of catalytically competent organometallic moieties. The chemical compositions of RNA and DNA lack the functional groups needed for catalysis and instead fully depend upon a catalytically active moiety to act as a cofactor. Various functional ligands can be rationally incorporated into the matrices of nucleic acids and the phosphate backbone provides a source of chirality. This field was first explored by Roelfes et al. by introducing a [Cu(diimine)]2+ complex within double stranded salmon testes DNA.1)

The concept of DNA-based asymmetric catalysis has mainly found applications in enantioselective Lewis acid-catalysed C-C bond forming reactions, including Diels-Alder cycloaddition, Friedel-Crafts alkylation and Michael addition.2) High enantiomeric excess (ee) values (>80%) were reported for these reactions using a 2,2’-bipyridine ligand,in which the DNA and metal binding is at the same position in the molecule, meaning that the substrates are in closer proximity of the DNA (Scheme 1). These and other approaches were reviewed by Roelfes and co-workers in 2010.2)

Scheme 1: Scope of DNA/Cu-dmbipy catalyzed C–C bond forming reactions.3)

Another, perhaps more obvious approach to combining transition metal catalysts with natural moieties is the incorporation of amino acids, for they constitute an easily accessible chiral pool. Furthermore, amino acids are the source of chirality in most chemical conversions performed by nature. After the modification of a diphosphane moiety with leucine by Hayashi et al. in 1982 to obtain a chiral diphosphane ligand4), many examples of connecting ligands to small peptide structures were reported.5)

Longer oligopeptide ligands were first synthesised by Gilbertson and co-workers.6) Various phosphane-modified amino acids were introduced at different positions in synthetic oligopeptides containing β-turn and helical structures. Inspired by this work, several other groups have created phosphane-containing oligopeptides as ligands for various transition metal catalysed reactions.5)

The purest and most challenging form of mimicking enzymes is de novo design of the polypeptide scaffold from the 20 natural amino acids. This involves constructing a polypeptide sequence which is not directly related to any natural protein and folds precisely into a well-defined three-dimensional structure capable of binding a metal ion. In theory, this would allow all the structural features required for highly selective catalysis to be present from the start. α helical bundles were among the first de novo designed proteins, for they are known to be common scaffolds for heme containing proteins in nature.7)Unfortunately, our current knowledge of protein folding limits the number of de novo designed polypeptides to only a few types.8)For this reason, the design of bio-hybrid catalyst using proteins as the biomolecular scaffold has focused on native proteins.

When using a native protein, either an existing active site and/or binding pocket can be reengineered or a new active site can be created. Pioneering work in this field of research was reported by Wilson and Whitesides and makes use of an existing binding pocket in avidin.9)The binding pocket of the protein must be large enough to accommodate both the transition metal catalyst and the substrates in order to be suitable for catalysis. Examples of proteins that have either been used or investigated include, apart from avidin and streptavidin, the oxygen transport protein Myoglobin10)11), bovine serum albumin12)and papain13), a cysteine protease present in papaya. More recently, bovine β-lactoglobulin was applied in this way.14)

Avidin and streptavidin are the most successful hosts for artificial metalloproteins used in catalysis to date. The strategy used to incorporate metals into the binding pocket is based on the exceptional affinity of the protein for biotin and biotinylated catalysts.15) This strategy will be discussed in more detail later. The thermal stability, tolerance to high concentrations of organic solvents, well-defined structure and coordination site and the suitability for optimization techniques make avidin and streptavidin attractive candidates for the biomolecular scaffold.16)

Another Class of possible scaffolds consists of proteins like ferritin, containing a large vacant space that allows for the incorporation of unnatural metal cofactors.17) These proteins have proven to be very useful in the selection of substrates based on their shape and size and have been successfully applied as a reactor for nanoparticles. The catalysis, however, did not yield any significant enantioselectivity.

An alternative is creating an active site in a polypeptide that does not yet have a binding pocket. This method can be applied to a greater number of proteins, but it is less clear how the structure and stability will be affected, for it may disrupt some important intramolecular interactions. An example of this approach is the binding of a CuII ion in the peptide hormone bovine pancreatic polypeptide.18)

In a recent study, a protein dimer was used as the biomolecular scaffold. Protein dimerization is the result of a combination of intermolecular, mostly hydrophobic, interactions. The chirality of the dimer interface, acting as the second coordination sphere, provides enantioselectivity while the hydrophobic nature facilitates the binding of the organic substrates and thus may improve the catalytic activity. Although the introduction of the transition metal catalyst into the scaffold may cause disruption of the structure and loss of dimerization affinity, a suitable dimeric protein was found and this might expand the number of scaffolds that can be used in catalysis.19)

Anchoring strategies

There are different strategies for the incorporation of the catalytic transition metal centres into the polypeptide scaffolds: dative,supramolecular and covalent anchoring (Figure 1).

Figure 1: Representation of the different anchoring strategies for biohybrid catalysts: (a) supramolecular, (b) dative and (c) covalent. M denotes the catalytically active transition metal centre.3)

Dative anchoring involves direct coordination of the transition metal to the protein in existing cavities or on the outer sphere. Kaiser and co-workers were the first to modify the active site of an enzyme datively by substitution of zinc by copper in carboxypeptidase A, affording novel catalytic properties. More recently, zinc has been replaced by manganese20)and rhodium21) in the active site of carbonic anhydrase (Figure 2).

Figure 2: Application of dative anchoring by replacing zinc atoms in carbonic anhydrase.5)

Since proteins can have multiple residues present which can act as a metal binding site (His, Asp, Glu, Lys, etc.), it can be difficult to create a well-defined transition metal centre in the catalyst. Site-directed mutagenesis and chemical modification can lead to more precise positioning of the transition metal in the protein. For example, Reetz et al. have been able to increase the enantioselectivity of a copper-coordinated protein, by removing several native histidines, which could act as alternative copper binding sites.22)

The covalent anchoring approach involves the covalent incorporation of a transition metal complex via the ligand to a predetermined position in the target protein. Kaiser and co-workers were the first to create an artificial metalloprotein by covalent modification of an amino acid.23)Like in this seminal report, a cysteine residue is often used as the anchoring site. Many different metal complexes have been incorporated using the same covalent attachment procedure.5) Disappointingly however, the catalytic potential of the resulting artificial metalloproteins remains very modest and the optimization techniques are not straightforward.

The supramolecular approach makes use of the strong and highly specific non-covalent interactions between proteins and small molecules, like inhibitors.By exploiting the high affinity of an inhibitor for a protein, an organometallic moiety can be introduced within a protein environment, resulting in the formation of stable supramolecular complexes. The interaction between avidin and biotin is one of the strongest protein-substrate interactions known and it was used by Wilson and Whitesides to incorporate a rhodium-diphosphane complex into the protein.9) Moreover, this affinity is only moderately affected by derivatization of the valeric acid side chain of biotin, which functions as a spacer to connect the chelating ligand (Figure 3).24)

Figure 3: Schematic representation of artificial metalloproteins based on biotin-(strept)avidin technology.25)

Whitesides’ work has inspired several groups to exploit the biotin-avidin system for the creation of artificial metalloproteins . A great advantage of this technology is that both the protein host and the transition metal catalyst can be optimized separately and joined together afterwards. This allows one to improve the performance of the catalyst either by varying the spacer and the ligand, or by mutating the gene of the host protein.25) Ward and co-workers found that changing the host protein to streptavidin, which has a less cationic character and a deeper binding pocket, but similar affinity to biotin, improved the performance of the catalyst drastically.26)

The variety of possibilities to modify the catalyst both chemically and biologically allows for profound testing of combinations of different biotinylated metal complexes with (strept)avidin proteins, which have been modified at strategic positions. At these positions saturation mutagenesis was applied, which means that all possible mutations at a specific site were generated. Many streptavidin isoforms, resulting from saturation mutagenesis, were screened with a library of biotinylated ligands, either differing in the chelating bisphenylphosphino moiety or in the spacer.25) This so-called chemogenetic approach resulted in successful improvement of the catalyst. The biotin binding site of streptavidin later proved to be large enough to accommodate small M=O containing coordination compounds, such asvanadyl salts.27)

Additional supramolecular anchoring strategies have been applied. The transport proteins serum albumin display the ability to bind a variety of hydrophobic guests tightly. The crystal structure of human serum albumin indicates that it is capable of storing five substrates using several hydrophobic cavities at a time.17)Different serum albumins have been tested as hosts for organometallic moieties.28) Although the exact position ofanchoring in the proteins was not elucidated with absolutecertainty in all cases, X-ray crystal structures showed that binding probably occurs in the IB subdomain. Another example is myoglobin, which can include a metal complex or a modified heme instead of native heme in the active site. Myoglobin was successfully modified by the group of Watanabe to host chromium-salphen complexes.10)

Catalytic scope

From the early stage of development of artificial metalloproteins, catalytic enantioselective hydrogenations have been extensively investigated. In the seminal report of Wilson and Whitesides, in which the biotin-(strept)avidin technology was introduced, the rhodium-catalysed reduction of acetamidoacrylic acid was reported.9) Only modest enantioselectivity was achieved (41% ee), but optimization of the technology by Ward and co-workers resulted in >95%ee for this reaction and further chemogenetic improvements led to >95% ee for both enantiomers and a threefold rate increase.25) These so-called second generation hybrid catalysts showed improved tolerance towards organic solvents.

De Vries and co-workers attached a rhodium complex covalently to papain for the same kind of hydrogenation reaction.13) This resulted in full conversion, but no enantiomeric discrimination. It was suggested that papain is an unsuitable host for application in asymmetric catalysis due to its conformational flexibility.

The biotin-(strept)avidin concept has also been successfully applied in the enantioselective transfer hydrogenation of ketones by the group of Ward, using a biotinylated aminosulfonamide as a ligand for the ruthenium η6-arenes.29)Based on the X-ray crystal structure of the catalyst, optimizations by designed evolution and mutagenesis were applied and excellent enantioselectivities were obtained for a broad range of substrates, including p-methylbenzophenone with 98% ee.30)

Artificial metalloproteins have also been successful in stereoselective sulfoxidation reactions, which are challenging with organometallic catalysis. Mn-salen complexes were incorporated into apo-Myoglobin by Watanabe and co-workers and 32% ee was obtained using H2O2 as oxidant.31) The importance of conformational rigidity was once again shown by the group of Lu, who improved the enantioselectivity of the catalyst by applying a dual anchoring technique resulting in up to 51% ee.11) Changing one of the attachment sites resulted in further increase of enantioselectivity to 60% ee.32) Moreover, the relatively apolar nature of the active site does not favour binding of the formed sulfoxide and therefore overoxidation does not occur (Scheme 2).

Scheme 2: a) Enantioselective and chemoselectivesulfoxidation reactions catalyzed by apo-Myoglobin/Mn-salen. b) Schematic representation of the protein-bound metal complex.3)

Serum albumins are another class of hosts for sulfoxidation catalysts. Gross et al. incorporated iron and manganese corroles into various serum albumins and up to 74% ee was obtained with Mn-corrole catalysts and H2O2 as the oxidant.33) More recently, vanadyl-loaded streptavidin was used by the group of Ward in asymmetric sulfoxidation reactions, with [VO(H2O)5]2+ions believed to be bound in the biotin binding pocket.27) The bio-hybrid catalyst displayed increased activity (and selectivity) compared to the protein-free salt.

Carbonic anhydrases, in which the active site zinc was replaced by manganese, was used as a catalyst for epoxidation reactions by the group of Kazlauskas, leading to enantioselectivities up to 67% ee.20) Recently, incorporation of manganese complexes into xylanase gave rise to 80% ee in the epoxidation of 4-methoxystyrene using KHSO5 as oxidant.34)

Most bio-hybrid catalysts for C-C bond forming reactions are based on DNA as the biomolecular scaffold. This has been extensively reviewed by Roelfes2) and will not be covered in this report. Using both dative and supramolecular anchoring, Reetz and co-workers introduced a copper complex into bovine serum albumin, producing an enantioselective catalyst for the Diels-Alder cycloaddition of aza-chalcone with cyclopentadiene that gave rise to ee values up to 98%.28) The binding of a CuII ion in the peptide hormone bovine pancreatic polypeptide resulted in a catalyst capable of asymmetric Diels-Alder and Michael additions.18)

Asymmetric alkylations have been achieved using the biotin-(strept)avidin system, using a biotinylated palladium complex.35) This reaction has, in contrast to other reactions catalysed by bio-hybrid catalysts, no equivalent in enzymatic catalysis. Using a chiral o-aminobenzoate spacer, up to 93% ee was achieved. The low to moderate ee values reported for the reaction in the absence of streptavidin, underlines the importance of the scaffold, although the ligand is chiral in this case. Recently, a biotinylated rhodium complex anchored to engineered streptavidin enabled asymmetric C-H activation, allowing for the coupling of benzamides and alkenes to access dihydroisoquinolones.36)The bio-hybrid catalyst gave rise to ee values up to 86% and the reaction proceeded with up to a 100-fold rate acceleration compared with the isolated rhodium complex.