Controlled Ligand Exchange Between Ruthenium Organometallic Cofactor Precursors and a Naïve Protein Scaffold Generates Artificial Metalloenzymes Catalysing Transfer Hydrogenation

Abstract Many natural metalloenzymes assemble from proteins and biosynthesised complexes, generating potent catalysts by changing metal coordination. Here we adopt the same strategy to generate artificial metalloenzymes (ArMs) using ligand exchange to unmask catalytic activity. By systematically testing RuII(η6‐arene)(bipyridine) complexes designed to facilitate the displacement of functionalised bipyridines, we develop a fast and robust procedure for generating new enzymes via ligand exchange in a protein that has not evolved to bind such a complex. The resulting metal cofactors form peptidic coordination bonds but also retain a non‐biological ligand. Tandem mass spectrometry and 19F NMR spectroscopy were used to characterise the organometallic cofactors and identify the protein‐derived ligands. By introduction of ruthenium cofactors into a 4‐helical bundle, transfer hydrogenation catalysts were generated that displayed a 35‐fold rate increase when compared to the respective small molecule reaction in solution.


Introduction
Creating artificial metalloenzymes from transition metal cofactors embedded in proteins allows for expansion of naturesc atalytic repertoire,b ringing about new-to-nature reactivity. [1][2][3][4][5][6][7][8][9][10] Catalysts of remarkable activity and selectivity have been obtained and evolved with naturally-biosynthesised metal cofactors and bare metal ions by varying the protein component of ArMs. [11][12][13][14][15][16][17][18] However,c omparable levels of evolutionary rate enhancement have yet to be achieved when using organometallic cofactors. [19][20][21][22][23][24][25][26][27][28] By expanding the reaction scope of enzymes to include catalytic activities normally exclusive to small molecules,A rMs have the potential to achieve efficient transformations without the need for activated substrates bearing directing groups,a s commonly encountered in classic asymmetric small molecule catalysis.A dditionally,A rMs are fully functional in aqueous solution, making it possible to avoid toxic organic solvents, which is ak ey advance towards more sustainable,g reen chemistry. [29,30] Already,"biofoundries" can be envisioned for synthesising or modifying essentially any conceivable organic molecule by pathway engineering using natural enzymes. [31] Thea ddition of orthogonal transition metal activity to the synthetic biologists toolbox would open up alternative metabolic routes with fewer steps,a nd better carbon and energy efficiency. [32] Strategies for generating ArMs containing non-natural metal cofactors include:m etal substitution, [25,26,[33][34][35][36] supramolecular assembly, [5,24,[37][38][39] covalent attachment, [22,[40][41][42][43][44] and coordinative metal-protein bonding. [13,15,[45][46][47][48][49][50] However,a n additional distinction can be made based on whether the ArM'sm etal cofactor contains am etal-protein coordination bond in the holoprotein or not. In the majority of ArMs published to date,s uch ac oordination bond is not formed, which crucially affects function and evolvability. [19,27,51] Without ametal-protein coordination bond, the first coordination sphere of the protein-bound metal cofactor (i.e.t he coordinated ligands) is unchanged from the precursor complex in solution. While this does not preclude as uitable protein scaffold from being evolved by improved binding arrangements vs.t he target substrates,t he core catalytic moiety will only be affected indirectly as any changes to the protein are restricted to the second coordination sphere.
Systems containing metal-peptide bonds,ordative ArMs, have been described, but mainly contain bare metal ions. [13,15,50] This has both biological and chemical consequences.F irstly,c ells tend to enact stringent control over bare metal ions in solution, making aconjugation system reliant on uptake of solvated ions difficult to translate into al iving host. [52] Secondly,t he limited ligand set available to natural proteins may restrict the type of catalysis that can be achieved, although this can be circumvented by the use of unnatural amino acids.B oth issues could be alleviated by using an organometallic complex as ap recursor to the reactive,protein-bound cofactor.
Here we set out to generate protein-metal conjugates by attaching small-molecule ruthenium complexes to protein scaffolds through ligand exchange,forming acofactor which is embedded within astructured environment. By systematically enhancing the ability of chelating ligands on the metal to be displaced upon binding to the protein, organometallic cofactors containing multiple protein-metal coordination bonds were successfully formed in ac ontrolled fashion. Ther esulting metal-protein conjugates that combine cytochrome b 562 variants with asuite of Ru II (h 6 -arene)(bipyridine) complexes ( Figure 1) show activity as catalysts for transfer hydrogenation.

Controlled Protein Conjugation via Ligand Exchange
We have previously used 19 FNMR spectroscopy to directly report the behaviour in aqueous solution of organometallic complexes carrying fluorinated ligands. [53] Specifically,t he ligand exchange between as uite of [Ru II -(arene)(bipyridine)Cl] + complexes and mixtures of amino acids established the cysteine thiol as the thermodynamicallypreferred replacement for the labile monodentate chloride ligand. Expanding upon the work with small molecules we explored the binding of these complexes to proteins and monitored their speciation through acombination of LC-MS and 19 FNMR spectroscopy.Ubiquitin K63C and cytochrome b 562 L10C/H102M have accessible cysteine residues for anchoring ruthenium complexes via ligand exchange and thus were examined initially.I ndeed, when the single cysteine-containing ubiquitin mutant, Ubq K63C,w as incubated with [Ru II (h 6 -cymene)(5,5'-difluorobipyridine)Cl] + , [2], as ingle species was observed by LC-MS,w here the protein was modified with the metal fragment [Ru(Cym)(FBipy)] (highlighted blue in Figure 2).
In contrast, after incubation with complex [5],a na dditional species was observed corresponding to Ubq K63C + [Ru(HMB)] (highlighted green in Figure 2), indicating loss of the bipyridyl ligand. Subsequent incubation of Ru-Ubq hybrids with N-ethyl maleimide (NEM) did not lead to any further changes in mass,confirming that the thiol was not free to react and verifying that these metal complexes prefer to bind to thiols,F igure S1. As the mass of the Ubq K63C + [Ru(HMB)] adduct did not indicate the coordination of further small molecules for example,s olvent, or ions to the metal, these observations suggest that the newly-assembled holoprotein complex had been formed by cysteine coordination, but with additional peptidic ligands replacing bipyridine in the first coordination sphere of the metal ion. The bipyridine-metal coordination linkage is stable in water and  amino acid solutions,b ut dissociation from the metal can clearly be facilitated in aprotein context favouring additional exchange.I mportantly,t he newly generated complex had to be formed by relatively weakly coordinating peptide-derived ligands (protein backbone,orsidechain carboxylates,amides, alcohols,p henols,a mines,i midazoles). This may be key for catalysis as,w ith more labile ligands,t he metal centre can undergo subsequent ligand exchange with as ubstrate to initiate ac atalytic cycle (vide infra).
Theo bserved loss of bipyridine ligand upon protein conjugation demonstrated the potential for [5] and similar complexes to undergo extensive ligand exchange with the protein beyond the single,m onodentate exchange observed using amino acids in solution. Further use of ubiquitin was thought to be unlikely to provide ap romising scaffold for aputative ArM given its small size and lack of awell-defined or nascent hydrophobic pocket leading to ah igh probability of the metal cofactor being bound to the surface.T his would inhibit an ArM to selectively bind and organise substrates. Furthermore,asitwas not possible to isolate the Ubq K63C + [Ru(HMB)] conjugate from the bipyridine-coordinated analogue for characterisation and further catalytic studies,amore detailed study was undertaken using as uite of related complexes [1]- [6] to explore the speciation within the more promising context of cytochrome b 562 .The speciation patterns on incubation of these complexes with the four-helix bundle protein cytochrome b 562 ,L10C/H102M are shown in Figure 3. Thec hoice of organometallic cofactor with general formula [Ru II (h 6 -arene)(bipyridine)Cl] + can be rationalised in terms of ab alance of aqueous solubility and stability as well as latent ligand lability.T he monodentate chloride ligand is readily replaceable by water and, depending on pH, generates aRu-OH 2 /OH bond which is in turn labilised on formation of aprimary coordination link to the protein. [54,55] Thearene component of the complex confers stability on the Ru II oxidation state and can be easily tuned to modulate 1Hr, x-axis:mass between 11 500 and 14 000 mass units, y-axis:s ignal intensity as %o fmax). The structures of the singly charged complexes (PF 6 removed) added at the start of the incubation are overlaid, and the assignments of adducts are given in the tables below the spectra.

Angewandte Chemie
Forschungsartikel electron density and steric pressure on the metal. [56] TheHMB ligand is more electron donating and agreater steric presence than Cym. Incorporation of fluorine atoms into the bipyridine ligand also enables electronic tuning with the electron withdrawing ability increasing from Bipy to FBipy to TFMBipy.T he loss of either arene or bipyridine is expected to be slow for these complexes,g iven their chelating nature and rutheniumss low metal-ligand exchange rates. [57] Complexes can therefore be predicted to be stable to ligand exchange in aqueous solution but become more labile upon equilibration with the protein, via the formation of coordination links and/or non-covalent interactions within the protein scaffold. These organometallic complexes were designed to promote displacement of the bipyridine by the protein in ac ontrolled manner,t herefore finding ab alance between ArM yield and non-specifically modified protein as an effect of the sequentially more electron withdrawing 5,5'-substituents on the bipyridine and the steric and electronic contributions of the arene.Cytochrome b 562 was chosen for its folding dynamics as well as its nascent haem cofactor binding site.In the apoprotein form, this four-helical-bundle protein is in dynamic,p artially folded states,w ith complete folding only being initiated by association of the heme cofactor. [58][59][60] This makes the four-helix-bundle protein highly versatile and promiscuous towards accommodating different cofactors and al ogical starting point for inclusion of an organometallic cofactor. [61] Themutant four-helix bundle protein cytochrome b 562 ,L10C/H102M (henceforth referred to as "Cyt b 562 Cys") had been historically designed for the covalent attachment of heme at the cysteiner esidue. [62] Ther eactions of complexes [1]- [6] with Cyt b 562 Cys showed that tuning the reactivity of the organometallic complexes had an effect on speciation. Whereas no bipyridine dissociation was observed using 5,5'-H-bipyridine (complexes [1] and [4],h ighlighted blue in Figure 3), increasing lability was observed across the series,culminating in much increased bipyridine dissociation when the substituents were -CF 3 ,that is,complexes [3] and [6].Inaddition to the mono-metallated conjugates,d i-metallated species were also observed for [1], [3], [ 4] and [6].S ingle modification of the protein was favoured by reducing the metal complex to protein ratio and incubation time,t wo further means of influencing metalprotein speciation.
Those metal-protein conjugates which contained ac ofactor carrying ab ipyridine ligand were isolated via anion exchange chromatography.P roducts obtained from incubations of Cyt b 562 Cys with [2], [3], [5] & [6] did not react with N-ethyl maleimide and all contained fluorinated ligands, enabling structural assignment via 19 FNMR spectroscopy. Thec hemical shift values for the fluorine atoms were consistent with cysteine coordination ( Figure S2). [53] Additionally,the spectra displayed two distinct 19 Fresonances with different linewidths,w ith each peak corresponding to one of the two fluorine atoms on the bipyridine ligand. Thedifferent shifts and relaxation properties attributed to the individual atoms suggested the fluorine atoms were situated in distinct chemical environments within the protein, potentially distinguished by buried or solvent-exposed positionings,t hus highlighting the ability of the scaffold to create asymmetric conditions for an unnatural cofactor, ap rincipal demand for stereoselective catalysis.
After establishing speciation, the catalytic potential of these hybrids was explored via atransfer hydrogenation assay. This reaction is known to be Ru II catalysed, has aw ell understood mechanism and was chosen due to its wide range of potential applications. [63][64][65] Them echanism of transfer hydrogenation does not necessitate ac hange in metal oxidation state hence the rate is dependent on ligand exchange kinetics,r ather than redox chemistry.F or catalysis to occur at the metal centre,aligand must exchange with ahydrogen donor (e.g.formate) to form aruthenium-hydride species,w hich can then hydrogenate the substrate.H ydrogenation was monitored via the reduction of apre-fluorescent quinolinium substrate 1 developed by Ward et al. (Figure S3 &S 4). [66] Fort hose hybrids where bipyridine dissociation had not occurred (highlighted blue in Figure 3), there was no observable catalytic activity.I nt hese cases,t he metal cofactor appeared to maintain as table first coordination sphere, suggesting that its stability precludes activity.W here protein metallation led to cysteinec oordination with loss of bipyridine (highlighted green in Figure 3), the potential for the protein to provide multiple peptidic ligands opened up the possibility for catalysis.U nfortunately,i solating the desired species,the 1:1adducts of Cyt b 562 Cys and [Ru(arene)],from the incubation mixtures was not readily achieved.
Thereactions with the cysteine variant had demonstrated that changing the electronic properties of the ligands influenced the speciation significantly.F or the metallation reactions presented, all approached completion rapidly and could also be achieved with only minor excess of metal complex. However,t he reaction yielded am ixture of protein-metal hybrids,with and without bipyridine dissociation which could not be individually isolated. Based on these preliminary findings it was hypothesised, that different, more dynamic speciation should be explored using aprotein scaffold without the very rapidly reacting cysteine,asthis could potentially be trapping the newly formed metalloprotein in aconformation where the bipyridine cannot be displaced. [37,39,67] Instead, as caffold with reduced positional dependencyo nasingle cysteine residue was chosen, allowing for equilibration of the cofactor and protein during the reaction. Therefore,the same set of complexes [1]- [6] were incubated with the cysteinefree,wild-type cytochrome b 562 (Cyt b 562 wt).
As work with amino acids had shown, [53] these same ruthenium complexes do undergo ligand exchange with Lewis basic amino acid functionalities other than thiol-albeit with amuch lower affinity-and therefore different speciation was expected from these incubations.Indeed, complexes [1, 2&4] did not produce any observable modified protein in the absence of the strong thiol ligand (Figure 4). However,t he more active complexes [3 &5 ]p roduced as ingle monosubstituted hybrid protein (identified by mass spectrometry, highlighted green in Figure 4), where loss of the bipyridine had occurred. Incubation with [6] resulted in complete conversion with considerable quantities of adoubly-modified protein also observed (highlighted cyan in Figure 4). By reducing the number of equivalents of ruthenium complexes Angewandte Chemie Forschungsartikel added from 20 to 2, full modification of Cyt b 562 wt was achieved within 2hours with as trong preference for the desired mono-substituted variant.
Having established as imple,y et highly efficient, method of generating hybrid proteins via ligand exchange,e fforts were made to isolate and characterise these novel metalloproteins.Anion-exchange chromatography proved to be an effective method, however,i tw as observed that, instead of purifying into as ingle fraction, the mono-substituted Cyt b 562 wt + [Ru(HMB)] adducts separated into two fractions ( Figure S5). These fractions contained ap rotein of the same mass but upon closer inspection showed clearly distinct ion series in their ESI-MS spectra. These individual fractions did not appear to interconvert in buffer after weeks nor after buffer exchange,r uling out dynamic causes such as different protonation states or interchangeable conformation. Thus,t he fractions were henceforth treated as separate variants,d enoted as Hybrid 1&2a nd Hybrids 3&4f or the products of incubation of the protein with [5] and [6] respectively,T able 1.

Catalytic Activity of Ruthenium-Cyt b 562 wt Hybrids
Satisfyingly,a ll hybrid variants,H ybrids 1-4, displayed significant activity in the transfer hydrogenation assay, 1Hr, x-axis:mass between 11 500 and 14 000 mass units, y-axis:s ignal intensity as %o fmax). The structures of the singly charged complexes (PF 6 removed) added at the start of the incubation are overlaid, and the assignments of adducts are given in the tables below the spectra.
Tabelle 1: Origin and analysis of the protein metal hybrids isolated from incubations of Cyt b 562 wt and complexes [ 5] and [6].R etention time is given for anion exchange, see Figure S5.

Angewandte Chemie
Forschungsartikel establishing these proteins as ArMs,F igure 5. Noticeably, Hybrids 1and 3had the same activity,even though they were derived from different complexes,asdid Hybrids 2and 4(see Figure S5), with the earlier eluting species,H ybrids 1a nd 3, being slightly more active.Compared to the well-established, small molecule catalyst, dichloro(p-cymene)ruthenium(II) dimer,[ Ru(Cym)Cl 2 ] 2 ,t he rates of transfer hydrogenation were ten-fold greater for Hybrids 1and 3. Importantly,ligand exchange with the protein yielded an up to 35-fold increase in rate compared to the small-molecule ruthenium complexes, [5] and [6] respectively. In order to gain am ore direct measure of the rate of metallation, the same assay was employed to monitor the metal-protein conjugation reaction in situ by measuring the transfer hydrogenation activity over time after addition of one equivalent of complex [6] to Cyt b 562 wt. Thek inetic traces obtained initially followed the same rate as the free metal complex before gradually increasing in activity as the ArM is formed in situ until full conversion into hybrid protein was reached and ac onstant rate is observed, as indicated by al inear trace ( Figure S6). Using equimolar amounts,c omplete conjugation of Cyt b 562 wt was achieved in less than two hours at 37 8 8C. Mass spectrometry analysis confirmed that the mono-substituted hybrid variants were the dominant species with minor contamination by some di-substituted protein.
Further,the maximum rate observed was in accordance with those obtained from the purified samples.T hus,s imilar activities to that from isolated species can be achieved in as ingle reaction vessel without purification, demonstrating the viability of "one-pot" ArMs generated by ligand exchange reactions,tobeused in techniques where individual purification may be limiting.

Characterisation of Ruthenium-Cyt b 562 wt Hybrids
Theorigin of the observed rate increase can be attributed to the new ligand environment in the activated complex, although tighter substrate binding,n on-covalent protein interactions stabilising the transition state or ac ombination of all of these could also contribute.T oi dentify the new ligands comprising the first coordination sphere of active ArMs (Hybrids 1-4), we performed tandem MS/MS.T his revealed that in all cases the ruthenium was bound by the same amino acids,n amely the N-terminal Alanine and Asn6 on a-helix 1a nd His63 on a-helix 3, Figure 6. Them etal is therefore held between two helices positioned diagonally in the apoprotein, implying that the active complex is embedded within the helical bundle and not at the surface.T he coordinating residues must have changed their orientation

Angewandte Chemie
Forschungsartikel compared to the average NMR apoprotein structure. [68] Besides inducing rearrangement, association of the metal complex may be favoured for ac ertain subset of the interchanging apoprotein conformations, [59,69] thus potentially explaining the generation of multiple holoprotein isomers following ligand exchange,asseen with the hybrid pairs 1/3 & 2/4. Tw oofthe peptidic ligands,the N-terminal amino group and the amide group of Asn6 are likely to be weakly coordinating,y et the complexes are stable enough to be detected by mass spectrometry.N evertheless,t hese weakly coordinating ligands would potentially lead to faster ligand exchange facilitating transfer hydrogenation activity.Further structural studies are ongoing to elucidate if the complex is distorted from the ideal piano-stool, pseudo-octahedral geometry.
Circular dichroism spectra of all these hybrids were very similar to that of the wild-type protein ( Figure S7). Thermal stability measurements of these proteins showed that Hybrids 2&4had a T m for denaturation close to the wild type protein but the melting temperature of Hybrids 1&3w as raised significantly in comparison, Figure S8. By comparing mass spectrometry data, melting temperature and the transfer hydrogenation activities it can be confidently concluded that Hybrids 1&3a re the same species,a sa re Hybrids 2&4. Thus,the same products can be obtained by incubation of one apoprotein with different small molecule complexes, [5] & [6] respectively.T he differences between Hybrids 1&3a nd Hybrids 2&4m ust result from different topologies around the metal-protein interactions,p ossibly different chirality at the metal centre. [70,71] Thehigher melting point together with the observation that more folded proteins tend to produce ion series of lower charge state distribution in the mass spectrometer indicates that the protein scaffold has adopted afold with more extensive intramolecular bonding in Hybrids 1&3 than in Hybrids 2&4. [72] Conclusion By exploiting the enhanced lability of ruthenium arene bipyridyl complexes in ap rotein context, protein derived coordination bonds can replace an otherwise inert chelating ligand. This observation allows us to report astraightforward and reliable incubation protocol for the generation of some novel artificial metalloproteins.T hese functional conjugates contained cofactors displaying multiple protein-metal coordination bonds;t he protein therefore exerts ad irect, firstsphere influence on the metal. Specifically,the products of the reaction of [Ru II (h 6 -arene)(Bipy)Cl] + complexes with Cyt b 562 wt, yielded artificial metalloenzymes capable of catalysing the reduction of aq uinolone substrate 1 via transfer hydrogenation from formate.U sing an aive protein scaffold, a3 5fold rate increase was achieved, when compared to the smallmolecule complex in water. Further, this ArM displayed ar ate of reaction in water approximately an order of magnitude higher than the known small-molecule catalyst with the residues for ruthenium coordination highlighted, N-terminal alanine (red), Asn6 (magenta)a nd His63 (cyan). [68] [Ru(Cym)Cl 2 ] 2 alone.I nt he future it may be possible to leverage the interdependence afforded by metal-protein coordination bonds for further optimization by directed evolution, to aim at catalytic rates eventually matching and rivalling the vast field of small-molecule hydrogenation catalysts but under more benign aqueous conditions and with the high specificity intrinsic to enzymes. [13,28,29,66] Akey feature of the method of metal-protein conjugation presented in this report is that, as ac onsequence of ligand exchange,the first coordination spheres of the organometallic species in the ArM and the small molecule precursor are distinct in both geometry and ligand identity and therefore chemical properties,i ncluding further ligand lability.I n particular,l igand exchange upon protein binding allows for using synthetic organometallic complexes that are stable and unreactive in aqueous solution as precursors to active ArMs. Thedesired catalytic activity can then be unmasked in situ by subsequent reaction with aprotein. This has clear advantages for implementing ArMs in living systems,a si tr educes potential background activity from the small molecule and allows for protection of the metal centre from the manifold of catalyst poisons present in cells. [67] Furthermore,this strategy still enables the metal centre to carry non-natural ligands into the enzyme which, besides from activating the metal centre, provides extended functionality that can be recognised by the protein. Such as ystem has parallels with the apparent mechanism in naturally evolved metalloenzymes.F or example,v itamin B12 is catalytically inactive in solution, yet undergoes ligand exchange upon binding to as pecific apoprotein, unmasking activity. [73,74] In contrast to ArMs where no ligand exchange occurs,reactivity is not imported" based on an intrinsic property of the metal complex and brought to bear in ap rotein environment, but generated as part of the ArM formation which proceeds via the unprecedented exchange of ab identate ligand to bring about ar eactive complex. Of the few ArM studies that have been reported that make use of as trategy where ligand exchange is central to formation of the artificial holoprotein, to the best of our knowledge,n one make use of the potential benefits listed above,asthey involve either toxic, water-sensitive or already very active precursor molecules. [45][46][47][48] With multiple direct coordination bonds between the metal and the protein scaffold, these ArMs can be expected to evolve like their natural counterparts,asprotein structure will directly impact the metal properties.S mall molecule metal complexes in solution will spontaneously adopt the nearest accessible lowest energy geometry,a ssuming the ligands can move freely.H owever,i fo ne or more of the ligands are constituted by the protein, these ligands cannot freely arrange,a st hey are an integral part of the peptide macromolecule and thus linked cooperatively.Aprotein scaffold can therefore distort the metal complex to an (in terms of the metal) energetically less favourable geometry,p otentially placing the metal into an activated or entatic state. [75] This distortion is possible by compensating the energetic cost of strain and low metal coordination energy with the many other non-covalent bonding interactions that form the three-dimensional structure of the protein. Thus,w hen subjugating the protein to evolutionary pressures,the core metal complex can constitute part of the evolutionary response and therefore an influence can be exerted over the fundamental chemically transforming step.W hen entire metal complexes are conjugated to protein hosts,asoften found in supramolecular and covalent methods of generating ArMs,this is not possible,and may be af actor limiting maximum activity in previous directed evolution campaigns. [19,27,28] With ArMs promising to play as ignificant role in achieving greener catalytic chemistry,o vercoming such limits using simple and easily available methodology will be key in fulfilling these expectations.