Beyond the Second Coordination Sphere: Engineering Dirhodium

Mar 4, 2019 - Beyond the Second Coordination Sphere: Engineering Dirhodium Artificial Metalloenzymes To Enable Protein Control of Transition Metal ...
0 downloads 0 Views 5MB Size
Article Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX

pubs.acs.org/accounts

Beyond the Second Coordination Sphere: Engineering Dirhodium Artificial Metalloenzymes To Enable Protein Control of Transition Metal Catalysis Published as part of the Accounts of Chemical Research special issue “Artificial Metalloenzymes and Abiological Catalysis of Metalloenzymes”. Jared C. Lewis* Downloaded via LUND UNIV on March 5, 2019 at 06:56:18 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Department of Chemistry, Indiana University, Bloomington, Indiana 47405, United States

CONSPECTUS: Transition metal catalysis is a powerful tool for chemical synthesis, a standard by which understanding of elementary chemical processes can be measured, and a source of awe for those who simply appreciate the difficulty of cleaving and forming chemical bonds. Each of these statements is amplified in cases where the transition metal catalyst controls the selectivity of a chemical reaction. Enantioselective catalysis is a challenging but well-established phenomenon, and regio- or siteselective catalysis is increasingly common. On the other hand, transition-metal-catalyzed reactions are typically conducted under highly optimized conditions. Rigorous exclusion of air and water is common, and it is taken for granted that only a single substrate (of a particular class) will be present in a reaction, a desired site selectivity can be achieved by installing a directing group, and undesired reactivity can be blocked with protecting groups. These are all reasonable synthetic strategies, but they also highlight limits to catalyst control. The utility of transition metal catalysis could be greatly expanded if catalysts possessed the ability to regulate which molecules they encounter and the relative orientation of those molecules. The rapid and widespread adoption of stoichiometric bioorthogonal reactions illustrates the utility of robust reactions that proceed with high selectivity and specificity under mild reaction conditions. Expanding this capability beyond preprogrammed substrate pairs via catalyst control could therefore have an enormous impact on molecular science. Many metalloenzymes exhibit this level of catalyst control, and directed evolution can be used to rapidly improve the catalytic properties of these systems. On the other hand, the range of reactions catalyzed by enzymes is limited relative to that developed by chemists. The possibility of imparting enzyme-like activity, selectivity, and evolvability to reactions catalyzed by synthetic transition metal complexes has inspired the creation of artificial metalloenzymes (ArMs). The increasing levels of catalyst control exhibited by ArMs developed to date suggest that these systems could constitute a powerful platform for bioorthogonal transition metal catalysis and for selective catalysis in general. This Account outlines the development of a new class of ArMs based on a prolyl oligopeptidase (POP) scaffold. Studies conducted on POP ArMs containing a covalently linked dirhodium cofactor have shown that POP can impart enantioselectivity to a range of dirhodium-catalyzed reactions, increase reaction rates, and improve the specificity for reaction of dirhodium carbene intermediates with targeted organic substrates over components of cell lysate, including bulk water. Several design features of these ArMs enabled their evolution via random mutagenesis, which revealed that mutations throughout the POP scaffold, beyond the second sphere of the dirhodium cofactor, were important for ArM activity and selectivity. While it was anticipated that the POP scaffold would be capable of encapsulating and thus controlling the selectivity of bulky cofactors, molecular dynamics studies also suggest that POP conformational dynamics plays a role in its unique efficacy. These advances in scaffold selection, bioconjugation, and evolution form the basis of our ongoing efforts to control transition metal reactivity using protein scaffolds with the goal of enabling unique synthetic capabilities, including bioorthogonal catalysis.

Received: December 7, 2018

© XXXX American Chemical Society

A

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research



INTRODUCTION Designing catalysts that control reaction selectivity is difficult.1 Designing such catalysts that provide high turnover numbers and turnover frequencies at ambient temperature and pressure in aqueous solution in the absence of protecting groups, directing groups, and harsh stoichiometric reagents is extremely difficult.2 Other considerations further emphasize the challenge of selective catalysis even if these reaction criteria are relaxed. For example, while enantioselective catalysis of reactions in which a pro-stereogenic element is proximal to the reaction site is relatively common, catalysts that differentiate pro-stereogenic elements distal to the reaction site3 or, more generally, functional groups of similar reactivity in complex molecules (site-selective catalysis) are far less so. Substrate scope can also be problematic for selective catalysts. Catalysts recognize and activate substrates via specific intermolecular interactions with particular functional groups in appropriate orientations, often distal to the reaction site,4 so poor activity or selectivity may be observed on substrates lacking these features. From a practical perspective, it is not obvious that the extreme catalyst control outlined above would be attractive on the basis of standard figures of merit.5 Why create a catalyst compatible with water when a suitable organic solvent can be identified? Why create a catalyst that can activate a substrate to react at room temperature when a reaction can be heated? The answers depend on your ultimate goals. What if you cannot exclude water from or heat your reaction because you want conduct it in a living cell? What if you want to avoid directing or protecting groups to improve synthetic efficiency or must avoid them because you want to modify a specific compound in a complex pool of metabolites? The utility of click chemistry and bioorthogonal reactions in general6 suggests that the ability to catalyze synthetic reactions under similar conditions could be highly valuable, and progress has been made toward this end.7 From a fundamental perspective, however, general approaches for developing catalysts that conform to the above criteria require a level of catalyst control beyond current capabilities and thus have the potential to significantly deepen our understanding of catalytic processes. Enzymes can teach us much about the benefits of extreme catalyst control and how it could be achieved in synthetic systems. These catalysts operate at ambient temperatures and pressures within narrow pH and redox potential ranges in aqueous solution and can be readily interfaced with one another in vivo or in vitro.8 They do not typically rely on directing or protecting groups, yet they can have exquisite site selectivity. In general, the selectivity of enzymes results from their ability to bind substrates within their active sites.9 Most catalysts bind their substrates to some extent, but the large three-dimensional active sites of enzymes give them greater control over substrate orientation than typical small molecules can achieve. A range of factors have been shown to contribute to the catalytic proficiency of enzymes,10 and conformational dynamics can play a large role in mediating enzyme−small molecule interactions.11 These same effects also apply to cofactors,12 including a range of structurally intriguing transition metal complexes. Indeed, metal complexes mediate some of the most challenging reactions in nature as a result of unique reactivity conferred to them by their protein hosts.12,13 In the context of a cytochrome P450, for example (Figure 1A),14 heme is capable of hydroxylating unactivated sp3 C−H bonds with a second-order

Figure 1. (A) Structure of heme B in cytochrome P450 BM3 showing the roles of selected residues. (B) Cofactor removal from natural metalloenzymes and incorporation of synthetic catalysts into protein scaffolds to generate artificial metalloenzymes (ArMs).

rate constant on the order of 107 M−1 s−1 at 4 °C15 and of selectively functionalizing C−H bonds at room temperature with extremely high rates (>17 000 min−1).14 These remarkable feats result from the ability of P450s to specifically bind substrates and modulate the reduction potential and basicity of ferryl−oxo heme species.16 Appropriately substituted metalloporphyrins can catalyze a range of reactions,17 but these lack the unique selectivity and mild reaction conditions that make the enzymatic transformations so remarkable (Figure 1B). Given the huge number of synthetic organometallic catalysts that have been developed, we asked an obvious question: Can a protein scaffold improve the reactivity and selectivity of a synthetic catalyst to the same levels as P450s do to heme (Figure 1B)? If so, could this strategy be used to develop bioorthogonal catalysts for site-selective reaction of otherwise inert fragments, such as C−H bonds?18 Before these questions could be answered, however, more immediate issues needed to be addressed. What would a protein scaffold for an arbitrary metal catalyst look like? How would synthetic complexes be linked to it to create an artificial metalloenzyme (ArM)? Most importantly, how would ArM function be optimized? Over the past seven years, my group has sought answers to these fundamental design questions, complementing related efforts by an increasingly vibrant field of researchers.19 The solutions we have identified touch on aspects of enzyme evolution, conformational dynamics, supramolecular catalysis, and secondary-sphere effects in transition metal catalysis that together enabled the development of a unique platform for addressing broader issues of catalyst control and bioorthogonality.



DIRHODIUM ARM CONSTRUCTION Our initial ArM efforts were designed to ultimately enable C−H functionalization reactions that proceed via outer-sphere C−H insertion or abstraction mechanisms since the selectivity of these reactions could be controlled by second-sphere interactions.20 Dirhodium paddlewheel complexes were selected for cofactor development because they catalyze carbene insertion into C−H bonds21 and a variety of other reactions22 using diazoacetate B

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research carbene precursors. Carbene addition to olefins to form cyclopropanes was selected for initial studies since this relatively facile reaction provided a best-case scenario for identifying active ArMs. Moreover, carbene addition to olefins23 catalyzed by peptide-based dirhodium complexes had been demonstrated, showing that second-sphere interactions could impart enantioselectivity to these transformations, albeit in organic solvent. Carbene addition to tryptophan residues on proteins catalyzed by rhodium(II) tetraacetate24 and selective peptide modification by peptide-based dirhodium catalysts25 had also been established. These reactions require a large excess of diazoacetate since it is consumed by formal insertion of carbene into the water O−H bond, a side reaction that would have to be mitigated to achieve high-yielding ArM catalysis in water. Dirhodium cofactors based on the tetradentate Esp ligand26 were synthesized to ensure that ligand dissociation would not lead to metal loss from the scaffold, and the primary coordination sphere was completed with acetate ligands to minimize cofactor bulk (Figure 2A). Complex 1 was substituted with a variety of covalent anchors, including a phosphonate, a

maleimide, and a bicyclo[6.1.0]nonyne (BCN). The resulting cofactors (2−4) were linked to serine proteases in the former case and to proteins containing genetically encoded cysteine or p-azido-L-phenylalanine (AzF, Z) residues in the latter two cases. The highest bioconjugation yields and rates were obtained via strain-promoted azide−alkyne cycloaddition (SPAAC) of 4 (Figure 2B),27 which proceeded smoothly in cell lysate and did not require removal of existing cysteine residues in the scaffold. Bioconjugation of 4 to AzF mutants of tHisF, which had been used in early ArM studies,28 was accomplished in good to excellent yields at all sites tested (Figure 2C). FRET studies confirmed that cofactors linked within the tHisF β-barrel protruded up through the barrel, where loops at the barrel surface could influence the ArM selectivity (Figure 2D). The resulting ArMs catalyzed carbene addition to styrenes and insertion into Si−H bonds, but in both cases racemic products were observed (Figure 2E), showing that improved scaffold− cofactor interactions would be required for selective catalysis.



ARM SCAFFOLD SELECTION Beginning in the late 1970s, researchers established that ArMs could be generated via covalent bioconjugation, biotin−avidin binding, non-native metal ion coordination, or non-native cofactor binding (Figure 3A−D).29 These systems established

Figure 3. (A−D) ArM formation methods. (E, F) Crystal structures of streptavidin- and myoglobin-based ArMs. Cofactors are shown in yellow.

many of the basic procedures used for ArM formation today and demonstrated that enantioselective ArM catalysis is possible. They also highlighted trade-offs between the generality of a cofactor linkage method and the degree of molecular recognition in the resulting ArM. For example, covalent attachment permits use of many different scaffolds and metal complexes, but flexible linkers between these fragments preclude complete cofactor encapsulation within the small protein cavities typically used (Figure 3A).30 Anchoring biotinylated complexes to (strept)avidin also allows for facile ArM formation19 and can largely avoid the linker problem (Figure 3B) since cofactors can be designed with a minimal number of rotatable bonds between the tightly bound biotin fragment and attached metal complexes. While many enantioselective ArMs have been developed using this approach,31 the metal complexes in these systems reside in a shallow cleft at one of the dimer interfaces of the streptavidin tetramer (Figure 3E).32,33 The selectivity of these ArMs also stems in part from resolution of chiral-at-metal complexes,30 so we anticipated that their ability to control the selectivity of 4 and other large and/or rigid complexes would be limited. Hemoproteins, on the other hand, can completely encapsulate

Figure 2. (A) Dirhodium cofactor structures. (B) ArM formation via strain-promoted azide−alkyne cycloaddition. (C, D) Representative MS data for and model of tHisFAz50-4. (E) Reactions catalyzed by tHisF ArMs. Adapted with permission from ref 27. Copyright 2014 Wiley-VCH. C

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research metal cofactors (Figure 3F),34 but relatively flat complexes must be used to fit within the native heme binding pocket.29 Given the limitations of established ArM scaffolds and our conviction that the benefits of SPAAC would outweigh the design constraints that it placed on the structure of 4 (a bulky BCN group and a flexible linker), we initiated a search for new scaffolds that could accommodate 4. Success in this venture would also illustrate that protein scaffolds and genetic optimization, rather than cofactor structure and chemical synthesis, could dictate the course of ArM evolution. Qualitative evaluation of protein crystal structures led to the identification of several candidate proteins with deep binding pockets that could completely encapsulate 4 and manipulate its reactivity via second-sphere interactions.19 While successful bioconjugation of 4 to AzF variants of some of these systems was observed, selective catalysis was not. A parallel effort to construct ArMs from serine hydrolase scaffolds and phosphonate-substituted cofactors met with similar failure. In the latter case, however, we recognized that data on serine hydrolase bioconjugation to phosphonate-substituted fluorophores could be gleaned from activity-based protein profiling efforts.35 Reported phylogenetic analysis of human serine hydrolases that act on small-molecule substrates facilitated the evaluation of representative structures amenable to selective bioconjugation.36 This approach piqued our interest in prolyl oligopeptidase (POP) family enzymes.37 The peptidase domain of these enzymes is similar to canonical serine proteases like trypsin, but they are capped by a fused βpropeller domain, creating an interdomain cavity that appeared to be large enough to contain 4 (Figure 4A).38 Pyrococcus

furiousus (Pfu) POP39 was particularly appealing because of its high stability. While no crystal structure of this enzyme existed, homology models had been reported,39 and we reasoned that these would provide sufficient guidance for targeted mutagenesis efforts.



TARGETED MUTAGENESIS ENABLES SELECTIVE CATALYSIS Several reports in the literature suggested that substrates enter the POP active site via an aperture at the top of the propeller domain,40 so four bulky residues at the mouth of this aperture were mutated to alanine to facilitate cofactor entry (Figure 4A). Because the catalytically active serine residue (S477) of Pfu POP must project into the active site of the enzyme, this residue was mutated to AzF. High-efficiency bioconjugation of the resulting enzyme with 4 to give POP-ZA4-4 was observed. Gratifyingly, the resulting ArM provided low but reproducible enantioselectivity (Figure 5, entry 1) for cyclopropanation of styrene with

Figure 5. Optimization of the reaction conditions and active-site mutations in HFF.

phenyldiazoacetate 5 to give 6.41 Subsequent studies have shown that the A4 mutations have little effect on the bioconjugation efficiency or ArM catalysis, but these mutations were present in subsequent mutants investigated. A histidine residue was introduced at different positions within the active site to bind the Rh center, establish two-point cofactor binding, and create a single active site distal to the histidine (Figure 4A).41 Variant L328H provided substantially improved selectivity (Figure 5, entry 7), and introducing G99F and G594F led to the variant HFF, which catalyzed styrene cyclopropanation with 92% ee (Figure 4 and Figure 5, entry 8). Importantly, high selectivity was dependent on conducting the reaction in the presence of >0.5 M NaBr (Figure 5, entries 1−3) or NaCl (unpublished). These conditions were explored on the basis of studies of native POP peptidase activity, which follows a sigmoidal response to increasing salt concentration.39 The ability of salt to improve two completely different activities within POP led us to suspect that its effects were related to secondary-sphere effects within the scaffold (vide infra). Control reactions catalyzed by (AcO)Esp(RhOAc)2 (8) indicated that this complex provided a burst of activity during which alcohol 7 was produced as the major product along with the desired cyclopropane 6 and unidentified side products (Figure 5, entry 9). HFF gave slower conversion of 5 relative to 8 but provided a quantitative mass balance, 6-fold improved selectivity for 6 over 7, and improved reaction rates and yields (Figure 5, entry 8). This finding suggested that the ArM scaffold

Figure 4. (A) Strategy used for POP ArM formation and optimization. (B) Mutations in HFF mapped onto a POP homology model containing covalently linked 4. Adapted with permission from ref 41. Copyright 2015 Springer Nature. D

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 6. Overview of the procedure used for ArM evolution via random mutagenesis. Adapted with permission from ref 44. Copyright 2018 Springer Nature.

POP in 96-well plates.44 Following bioconjugation, azidesubstituted resin was used to scavenge excess cofactor that would otherwise catalyze nonselective background reactions. Only three rounds of directed evolution were required to obtain an ArM that catalyzes styrene cyclopropanation with 92% ee (Figure 7A). The evolved ArM, 3-VRVH, also catalyzes

imparted specificity to the dirhodium catalyst for organic substrates over bulk water, solving the problem faced by rhodium(II) tetraacetate24 or peptide23 complexes. In addition, examining different L328X mutations indicated that other metal-binding residues also provided improved selectivity and comparable/improved yields and 6/7 ratios relative to 328L/F (Figure 5, entries 3−7). While we rationalized the improved enantioselectivity of these variants on the basis of the formation of a single-site catalyst with an optimized reaction site (Figure 4A), it was less clear why the yield and the 6/7 ratio should improve with metal-binding L328X mutations. This was particularly notable given the known decrease in activity of dirhodium complexes upon Lewis base coordination and observation of the decreased yields upon introduction of metal-coordinating groups in peptide-based dirhodium catalysts.42 Here again, it seemed that the POP scaffold was modulating the dirhodium activity in a manner not consistent with primary-coordination-sphere effects.



SYSTEMATIC ARM EVOLUTION VIA RANDOM MUTAGENESIS Despite the success of our initial POP ArM design effort, extensive trial and error was required to identify the active-site mutations that ultimately led to the variant HFF. We suspected that this resulted in part from differences between the homology models that guided mutagenesis and the actual Pfu POP structure. This suspicion motivated efforts to crystallize Pfu POP (vide infra), but we also recognized that a crystal structure of the ArM might not provide sufficient information for targeted mutagenesis since ArM cofactors often possess low occupancy and high disorder.33 More fundamentally, we and others had already shown that active-site mutations proximal to ArM cofactors can improve their activity and selectivity.43 We wondered whether mutations beyond the second coordination sphere could be important for ArM activity, just as they are for natural enzymes. ArM optimization via directed evolution (Figure 6) underpinned our commitment to using SPAAC for cofactor bioconjugation into a thermostable scaffold, even if it necessitated a flexible linker in 4 and the use of a scaffold without a crystal structure. In practice, the stability of Pf u POP allowed the use of high-mutation-rate error-prone PCR and thus analysis of several mutations in each variant. Enzyme variants were expressed with high titers in 24-well plates, and the high efficiency of SPAAC allowed for rapid bioconjugation of 4 to

Figure 7. (A) ArM lineage selectivity and yields. (B) Reaction progress of cyclopropanation catalyzed by different dirhodium catalysts. Adapted with permission from ref 44. Copyright 2018 Springer Nature.

cyclopropanation of several different substituted styrenes and phenyldiazoacetate esters with high enantioselectivity (86−94% ee). 3-VRVH contained a total of 12 mutations, provided higher rates than HFF, and yielded more cyclopropane than EspOAc(RhOAc)2 at the earliest time point sampled, demonstrating that scaffold-accelerated catalysis19 was occurring (Figure 7B). Reverting each mutation in 3-VRVH to the wild-type residue revealed that only three mutations (S301G, G99S, and Y326H) were responsible for the improved selectivity of 3-VRVH. While E

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Figure 8. (A) Side and (B) top view of Pfu POP covalently modified with 4. 3-VRVH mutations from evolution rounds 0−3 are shown in gray, red, orange, and blue, respectively. (C) Side view of 0-ZA4 showing the locations of residue 477 (green), residue 413 (blue), and residues selected for combinatorial codon mutagenesis (gray). Adapted with permission from ref 44. Copyright 2018 Springer Nature.

maximize the benefits of random mutations and rapidly improve ArM selectivity.

the variant containing only these mutations (1-GSH) provided the same enantioselectivity as 3-VRVH, it also gave significantly reduced yields (Figure 7A), mirroring the compensatory effects of mutations identified during directed evolution of natural enzymes.45 Toward the end of our directed evolution campaign, we solved the crystal structure of Pfu POP, which provided further insights into the mutations identified.46 Out of the 12 mutations in 3-VRVH, only G99S and Y326H are in the POP active site (Figure 8A,B). These mutations are similar to two of those in HFF (G99F and L328H),41 but the former conferred superior activity. This finding shows that even when reasonable targeting strategies are used, selecting specific sites and mutations at these sites is challenging, particularly if homology models are used. The residues and orientations of amino acids lining the Pfu POP active site differed substantially between the crystal structure and homology model. More importantly, the fact that the remaining mutations, including one responsible for improving the selectivity, were outside of the active site illustrates the importance of second-sphere mutations for improving ArM activity and of efficient methods to identify such mutations. Both of these findings are commonly accepted and practiced in efforts to evolve natural enzymes,45 but they had never been exploited for ArM engineering,43 suggesting that far greater improvements in the performance of other ArMs could be achieved using this approach. The Pfu POP crystal structure was used to target several additional sites for AzF mutation and cofactor linkage to generate ArMs with altered selectivity (Figure 8C). Gratifyingly, an ArM constructed from POP variant F413Z provided modest selectivity (30% ee) for the opposite enantiomer than was obtained using 3-VRVH. Structures for cofactor-bound ArMs could not be obtained, so it was unclear which sites might be reasonable to target to further improve POP-F413Z-4. A hybrid mutagenesis approach in which random mutations were targeted to active-site positions was therefore pursued. Specifically, combinatorial codon mutagenesis47 was used to install an average of two degenerate (NDT) codons in each member of a library targeted to the 25 residues projecting into the active site of POP-413Z. Screening only 92 variants from this library using a modified version of our original protocol (Figure 6) in which library members were immobilized on Ni-NTA resin led to the identification of 1-RFY (POP-413Z Q98R/G99F/ P239Y, Figure 8C), which provided 80% ee for the desired cyclopropanation enantiomer (Figure 7A). This effort showed how even minimal structural information can be leveraged to



BEYOND CYCLOPROPANATION In addition to improved cyclopropanation activity, variants in our ArM lineage provided improved yields and/or selectivities for formal carbene insertion into Si−H, S−H, and N−H bonds (Figure 9).44 Of these, the N−H insertion reaction is particularly

Figure 9. Additional X−H insertion reactions catalyzed by dirhodium ArMs.

notable because of the challenge of catalyzing this reaction in an enantioselective fashion and potential implications of its mechanism (Figure 9, entry 3).22 The enantiodetermining step of this reaction is believed to follow dissociation of a reactive ylide from the dirhodium catalyst.48 Enantioselective ArM-catalyzed N−H insertion therefore suggests that selective catalysis within the POP active site may be possible even in the absence of metal coordination, which would greatly broaden the scope of reactions that could be pursued. Further evolution of dirhodium ArMs for enantioselective N−H insertion could itself prove fruitful in light of the fact that to date hemoproteincatalyzed N−H insertion reactions have been limited to ethyl diazoacetate,49,50 which forms achiral insertion products, or to cyclic diazo-containing compounds.51 The ArM lineage also provided new enzymes that could be examined as catalysts for C−H insertion, which had proven elusive because of the low rate of this reaction relative to formal water O−H insertion. Preliminary results show that 3-VRVH catalyzes carbene insertion into the highly activated C−H bonds F

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research of 1,4-dihydrobenzene (Figure 9, entries 4 and 5), and efforts are underway to evolve enzymes that provide high yield and selectivity for this and more challenging C−H insertion reactions. Davies has shown that dirhodium complexes catalyze carbene insertion into unactivated primary, secondary, and tertiary C−H bonds,52 suggesting that evolution could ultimately enable all of these within dirhodium ArMs. Unlike impressive examples of carbene insertion into C−H bonds catalyzed by iridium-substituted53 and evolved54 cytochromes P450, however, this effort will test our ability to engineer proteins to modulate the reactivity of a completely foreign cofactor. In this regard, it is notable that efforts to engineer streptavidin-based ArMs containing a cofactor with a dirhodium fragment identical to that in 4 provided no enantioselectivity for carbene insertion into olefins or C−H bonds.55 This finding highlights the benefits of complete cofactor encapsulation within POP, and it will be interesting to learn whether recent efforts to engineer streptavidin variants with more-enclosed active sites56 could also achieve this goal.



POP AS A SUPRAMOLECULAR CATALYST The structure of Pfu POP and molecular dynamics (MD) simulations based on this structure provided explanations for a number questions that had emerged during the course of our evolution efforts.46 Among the most basic of these was why POP bioconjugation proceeds so efficiently within a structure that previously reported kinetic57 and homology39 models suggested was closed. The POP crystal structure showed that the enzyme actually has an interdomain angle between those seen in structures of POPs that crystallized in either a closed or open conformation (Figure 10A). MD simulations of the apo structure showed that it can sample fully closed and open states (Figure 10B), the latter of which would allow access of 4 to the active site (Figure 10D). While the crystal structure has an active-site volume of 9561 Å3, the volumes of the open and closed structures from MD simulations were 13 643 and 5379 Å3, respectively, showing that significant reduction of space occurs as a result of POP conformational dynamics. Similar opening and closing motions were observed in MD simulations of the ArM (Figure 10C), suggesting that the ability of the scaffold to access a closed conformation with reduced active-site volume could give rise to the selectivity observed in our early ArM designs, which did not possess a Rh-binding histidine residue (Figure 10D). Another major question was why metal coordinating residues led to improved ArM yields and specificity for carbene insertion into olefins over the water O−H bond (see the column labeled 6/7 in Figure 5). This can be rationalized by noting that coordination of residues in the β-propeller domain to a dirhodium complex anchored to the peptidase domain would create an internal cross-link that would favor the closed form of the ArM (Figure 10D). MD simulations of the apo structure also indicated that chloride binding within the POP active site (observed in the crystal structure) led to significant conformational changes in loops at the interdomain interface and a smaller interdomain angle, consistent with the beneficial effects of high halide concentration on ArM selectivity. Thus, both cofactor and halide binding appear to result in a more ordered, hydrophobic active site with significantly reduced volume to improve secondary-sphere interactions and reaction selectivity (Figure 10E). These findings have led us to view POP not simply as either an enzyme or an ArM but as a supramolecular host in which either of these activities and others can occur. Ongoing

Figure 10. (A) Side and top views of Pfu POP. (B, C) Interdomain angle (θ) from MD simulations of apo Pf u POP (B) and POP-Z477-4 (C). The green bars indicate intermediate angles between the open and closed states. (D, E) Bioconjugation to the open conformation of Pf u POP and selective catalysis within the closed conformation.

studies are exploring the generality of this system and the impact of conformational dynamics on its function.



CONCLUSION AND OUTLOOK The studies outlined above document our efforts to engineer dirhodium ArMs for selective catalysis. Seven years of design, optimization, and evolution was required to obtain an ArM that catalyzes selective C−H functionalization. This system is now being used as a starting point to evolve ArMs that catalyze a variety of challenging reactions, including site-selective nondirected C−H functionalization of unactivated substrates. The ArMs evolved to date display improved activity as a result of mutations throughout the POP scaffold. Mutations within or near the ArM active site have provided most of the selectivity improvements observed, while those far from the active site, beyond what could be considered even the second coordination sphere of the dirhodium cofactor, have been essential for improving the activity. Ongoing efforts are also focused on understanding the effects of scaffold mutations on dirhodium ArM structure, catalysis, and dynamics. We have shown that both primary- and secondarysphere effects (e.g., scaffold residue coordination to the dirhodium center and creation of a hydrophobic active-site cavity, respectively) appear to influence dirhodium reactivity G

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research

Small Molecule Catalysts. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 20678−20685. (5) Christensen, C. M. The Innovator’s Dilemma, 5th ed.; Harvard Business Review Press: Boston, 2016; pp 53−56. (6) Sletten, E. M.; Bertozzi, C. R. Bioorthogonal Chemistry: Fishing for Selectivity in a Sea of Functionality. Angew. Chem., Int. Ed. 2009, 48, 6974−6998. (7) Rebelein, J. G.; Ward, T. R. In Vivo Catalyzed New-to-Nature Reactions. Curr. Opin. Biotechnol. 2018, 53, 106−114. (8) Köhler, V.; Turner, N. J. Cascades with Multiple Engineered and/ or Heterologously Expressed Enzymes in Vivo or as Whole Cell Reactions. Chem. Commun. 2015, 51, 450−464. (9) Eliot, A. C.; Kirsch, J. F. Avoiding the Road Less Traveled: How the Topology of Enzyme−Substrate Complexes Can Dictate Product Selection. Acc. Chem. Res. 2003, 36, 757−765. (10) Garcia-Viloca, M.; Gao, J.; Karplus, M.; Truhlar, D. G. How Enzymes Work: Analysis by Modern Rate Theory and Computer Simulations. Science 2004, 303, 186−195. (11) Bhabha, G.; Biel, J. T.; Fraser, J. S. Keep on Moving: Discovering and Perturbing the Conformational Dynamics of Enzymes. Acc. Chem. Res. 2015, 48, 423−430. (12) Hemschemeier, A.; Happe, T. The Plasticity of Redox Cofactors: From Metalloenzymes to Redox-Active DNA. Nat. Rev. Chem. 2018, 2, 231−243. (13) Ragsdale, S. W. Metals and Their Scaffolds to Promote Difficult Enzymatic Reactions. Chem. Rev. 2006, 106, 3317−3337. (14) Noble, M. A.; Miles, C. S.; Chapman, S. K.; Lysek, D. A.; MacKay, A. C.; Reid, G. A.; Hanzlik, R. P.; Munro, A. W. Roles of Key Active-Site Residues in Flavocytochrome P450 BM3. Biochem. J. 1999, 339, 371−379. (15) Rittle, J.; Green, M. T. Cytochrome P450 Compound I: Capture, Characterization, and C−H Bond Activation Kinetics. Science 2010, 330, 933−937. (16) Green, M. T.; Dawson, J. H.; Gray, H. B. Oxoiron (IV) in Chloroperoxidase Compound II Is Basic: Implications for P450 Chemistry. Science 2004, 304, 1653−1656. (17) Barona-Castaño, J.; Carmona-Vargas, C.; Brocksom, T.; de Oliveira, K. Porphyrins as Catalysts in Scalable Organic Reactions. Molecules 2016, 21, 310−327. (18) Upp, D. M.; Lewis, J. C. Selective C−H Bond Functionalization Using Repurposed or Artificial Metalloenzymes. Curr. Opin. Chem. Biol. 2017, 37, 48−55. (19) Schwizer, F.; Okamoto, Y.; Heinisch, T.; Gu, Y.; Pellizzoni, M. M.; Lebrun, V.; Reuter, R.; Kohler, V.; Lewis, J. C.; Ward, T. R. Artificial Metalloenzymes: Reaction Scope and Optimization Strategies. Chem. Rev. 2018, 118, 142−231. (20) Yang, J.; Gabriele, B.; Belvedere, S.; Huang, Y.; Breslow, R. Catalytic Oxidations of Steroid Substrates by Artificial Cytochrome P450 Enzymes. J. Org. Chem. 2002, 67, 5057−5067. (21) Davies, H. M. L.; Manning, J. R. Catalytic C−H Functionalization by Metal Carbenoid and Nitrenoid Insertion. Nature 2008, 451, 417−424. (22) Timmons, D. J.; Doyle, M. P. Chiral Dirhodium(II) Catalysts and Their Applications. In Multiple Bonds between Metal Atoms; Cotton, F. A., Murillo, C. A., Walton, R. A., Eds.; Springer: Boston, 2005; pp 591−632. (23) Ball, Z. T. Designing Enzyme-Like Catalysts: a Rhodium(II) Metallopeptide Case Study. Acc. Chem. Res. 2013, 46, 560−570. (24) Antos, J. M.; Francis, M. B. Selective Tryptophan Modification with Rhodium Carbenoids in Aqueous Solution. J. Am. Chem. Soc. 2004, 126, 10256−10257. (25) Popp, B. V.; Ball, Z. T. Proximity-Driven Metallopeptide Catalysis: Remarkable Side-Chain Scope Enables Modification of the Fos bZip Domain. Chem. Sci. 2011, 2, 690−695. (26) Espino, C.; Fiori, K.; Kim, M.; Du Bois, J. Expanding the Scope of C−H Amination Through Catalyst Design. J. Am. Chem. Soc. 2004, 126, 15378−15379. (27) Yang, H.; Srivastava, P.; Zhang, C.; Lewis, J. C. A General Method for Artificial Metalloenzyme Formation Through Strain-

within POP. It will be interesting to establish whether further modulation of organometallic reactivity within POP can be achieved via ligand substitution involving natural or unnatural amino acids or by imposing geometric or electronic constraints on primary-coordination-sphere ligands or the metal centers themselves.12 Of course, BCN cofactors based on any number of transition metal catalysts can also be incorporated into POP, and we anticipate that a wide range of selective reactions involving C−H oxygenation58 and visible-light photocatalysis59 will ultimately be possible. Likewise, while Pfu POP is an effective ArM scaffold, other POP family enzymes could provide better starting points for ArM engineering efforts. Our ability to begin exploring these exciting new directions at the intersection of enzyme, transition metal, and supramolecular catalysis has been uniquely enabled by the fundamental design efforts outlined above. We envision that these approaches will ultimately enable the evolution of ArMs with the extreme catalyst control needed to enable new families of catalytic bioorthogonal reactions.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Jared C. Lewis: 0000-0003-2800-8330 Notes

The author declares no competing financial interest. Biography Jared C. Lewis was born and raised in Effingham, IL. He obtained his B.S. in chemistry from the University of Illinois (2002, Prof. Eric Oldfield), earned his Ph.D. in chemistry from the University of California, Berkeley (2007, Profs. Jonathan Ellman and Robert Bergman), and conducted postdoctoral studies at Caltech (2010, Prof. Frances Arnold). He started his independent career at the University of Chicago in 2011 and moved to Indiana University as an Associate Professor in 2018.



ACKNOWLEDGMENTS I am deeply indebted to the students and postdoctoral researchers who conducted the research summarized in this Account. I am also grateful to the David and Lucile Packard Foundation and the Searle Scholars Program for funding this research at an early phase and to the NSF (CHE-1351991 and the Center for Chemical Innovation Center for Selective C−H Functionalization, CHE-1700982), the U.S. Army Research Laboratory, and the U.S. Army Research Office (W911NF-14-10334, W911NF-18-1-0200, and 66796-LS-RIP) for continued funding. Finally, I thank my colleagues in the Department of Chemistry at Indiana University for their support.



REFERENCES

(1) Mahatthananchai, J.; Dumas, A. M.; Bode, J. W. Catalytic Selective Synthesis. Angew. Chem., Int. Ed. 2012, 51, 10954−10990. (2) Völker, T.; Dempwolff, F.; Graumann, P. L.; Meggers, E. Progress Towards Bioorthogonal Catalysis with Organometallic Compounds. Angew. Chem., Int. Ed. 2014, 53, 10536−10540. (3) Metrano, A. J.; Miller, S. J. Peptide-Based Catalysts Reach the Outer Sphere Through Remote Desymmetrization and Atroposelectivity. Acc. Chem. Res. 2019, 52, 199−215. (4) Knowles, R. R.; Jacobsen, E. N. Attractive Noncovalent Interactions in Asymmetric Catalysis: Links Between Enzymes and H

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research Promoted Azide−Alkyne Cycloaddition. ChemBioChem 2014, 15, 223−227. (28) Reetz, M. T.; Rentzsch, M.; Pletsch, A.; Taglieber, A.; Hollmann, F.; Mondiere, R. J. G.; Dickmann, N.; Hoecker, B.; Cerrone, S.; Haeger, M. C.; Sterner, R. A Robust Protein Host for Anchoring Chelating Ligands and Organocatalysts. ChemBioChem 2008, 9, 552−564. (29) Lewis, J. C.; Guardiola, K. E. Preparation of Artificial Metalloenzymes. In Artificial Metalloenzymes and MetalloDNAzymes in Catalysis: From Design to Applications; Diéguez, M., Bäckvall, J.-E., Pàmies, O.. Eds.; Wiley-VCH: Weinheim, Germany, 2018; Chapter 1, pp 1−40. (30) Lewis, J. C. Artificial Metalloenzymes and Metallopeptide Catalysts for Organic Synthesis. ACS Catal. 2013, 3, 2954−2975. (31) Ward, T. R. Artificial Metalloenzymes Based on the Biotin− Avidin Technology: Enantioselective Catalysis and Beyond. Acc. Chem. Res. 2011, 44, 47−57. (32) Dürrenberger, M.; Heinisch, T.; Wilson, Y. M.; Rossel, T.; Nogueira, E.; Knörr, L.; Mutschler, A.; Kersten, K.; Zimbron, M. J.; Pierron, J.; Schirmer, T.; Ward, T. R. Artificial Transfer Hydrogenases for the Enantioselective Reduction of Cyclic Imines. Angew. Chem., Int. Ed. 2011, 50, 3026−3029. (33) Creus, M.; Pordea, A.; Rossel, T.; Sardo, A.; Letondor, C.; Ivanova, A.; Le Trong, I.; Stenkamp, R. E.; Ward, T. R. X-Ray Structure and Designed Evolution of an Artificial Transfer Hydrogenase. Angew. Chem., Int. Ed. 2008, 47, 1400−1404. (34) Ueno, T.; Ohashi, M.; Kono, M.; Kondo, K.; Suzuki, A.; Yamane, T.; Watanabe, Y. Crystal Structures of Artificial Metalloproteins: Tight Binding of Fe III(Schiff-Base) by Mutation of Ala71 to Gly in ApoMyoglobin. Inorg. Chem. 2004, 43, 2852−2858. (35) Liu, Y.; Patricelli, M. P.; Cravatt, B. F. Activity-Based Protein Profiling: the Serine Hydrolases. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 14694−14699. (36) Simon, G. M.; Cravatt, B. F. Activity-Based Proteomics of Enzyme Superfamilies: Serine Hydrolases as a Case Study. J. Biol. Chem. 2010, 285, 11051−11055. (37) Polgár, L. The Prolyl Oligopeptidase Family. Cell. Mol. Life Sci. 2002, 59, 349−362. (38) Rea, D.; Fülöp, V. Structure-Function Properties of Prolyl Oligopeptidase Family Enzymes. Cell Biochem. Biophys. 2006, 44, 349− 365. (39) Harris, M. N.; Madura, J. D.; Ming, L.-J.; Harwood, V. J. Kinetic and Mechanistic Studies of Prolyl Oligopeptidase From the Hyperthermophile Pyrococcus Furiosus. J. Biol. Chem. 2001, 276, 19310− 19317. (40) Kaushik, S.; Etchebest, C.; Sowdhamini, R. Decoding the Structural Events in Substrate-Gating Mechanism of Eukaryotic Prolyl Oligopeptidase Using Normal Mode Analysis and Molecular Dynamics Simulations. Proteins: Struct., Funct., Genet. 2014, 82, 1428−1443. (41) Srivastava, P.; Yang, H.; Ellis-Guardiola, K.; Lewis, J. C. Engineering a Dirhodium Artificial Metalloenzyme for Selective Olefin Cyclopropanation. Nat. Commun. 2015, 6, 7789. (42) Sambasivan, R.; Zheng, W.; Burya, S. J.; Popp, B. V.; Turro, C.; Clementi, C.; Ball, Z. T. A Tripodal Peptide Ligand for Asymmetric Rh(II) Catalysis Highlights Unique Features of on-Bead Catalyst Development. Chem. Sci. 2014, 5, 1401−1407. (43) Hyster, T. K.; Ward, T. R. Genetic Optimization of Metalloenzymes: Enhancing Enzymes for Non-Natural Reactions. Angew. Chem., Int. Ed. 2016, 55, 7344−7357. (44) Yang, H.; Swartz, A. M.; Park, H.-J.; Srivastava, P.; EllisGuardiola, K.; Upp, D. M.; Lee, G.; Belsare, K.; Gu, Y.; Zhang, C.; Moellering, R. E.; Lewis, J. C. Evolving Artificial Metalloenzymes via Random Mutagenesis. Nat. Chem. 2018, 10, 318−324. (45) Romero, P. A.; Arnold, F. H. Exploring Protein Fitness Landscapes by Directed Evolution. Nat. Rev. Mol. Cell Biol. 2009, 10, 866−876. (46) Ellis-Guardiola, K.; Rui, H.; Beckner, R.; Srivastava, P.; Sukumar, N.; Roux, B.; Lewis, J. C. Pyrococcus furiosus Prolyl Oligopeptidase: A Dynamic Supramolecular Host for Peptidase and Dirhodium Catalysis. 2018, DOI: 10.26434/chemrxiv.7053812.v1. ChemRxiv: The Preprint

Server for Chemistry. https://chemrxiv.org/articles/Pyrococcus_ furiosus_Prolyl_Oligopeptidase_A_Dynamic_Supramolecular_ Host_for_Peptidase_and_Dirhodium_Catalysis/7053812/1 (accessed Dec 7, 2018). (47) Belsare, K. D.; Andorfer, M. C.; Cardenas, F. S.; Chael, J. R.; Park, H.-J.; Lewis, J. C. A Simple Combinatorial Codon Mutagenesis Method for Targeted Protein Engineering. ACS Synth. Biol. 2017, 6, 416−420. (48) Xu, B.; Zhu, S.-F.; Xie, X.-L.; Shen, J.-J.; Zhou, Q.-L. Asymmetric N-H Insertion Reaction Cooperatively Catalyzed by Rhodium and Chiral Spiro Phosphoric Acids. Angew. Chem., Int. Ed. 2011, 50, 11483− 11486. (49) Wang, Z. J.; Peck, N. E.; Renata, H.; Arnold, F. H. Cytochrome P450-Catalyzed Insertion of Carbenoids Into N−H Bonds. Chem. Sci. 2014, 5, 598−601. (50) Sreenilayam, G.; Fasan, R. Myoglobin-Catalyzed Intermolecular Carbene N−H Insertion with Arylamine Substrates. Chem. Commun. 2015, 51, 1532−1534. (51) Chen, K.; Zhang, S.-Q.; Brandenberg, O. F.; Hong, X.; Arnold, F. H. Alternate Heme Ligation Steers Activity and Selectivity in Engineered Cytochrome P450-Catalyzed Carbene-Transfer Reactions. J. Am. Chem. Soc. 2018, 140, 16402−16407. (52) Liao, K.; Pickel, T. C.; Boyarskikh, V.; Bacsa, J.; Musaev, D. G.; Davies, H. M. L. Site-Selective and Stereoselective Functionalization of Non-Activated Tertiary C−H Bonds. Nature 2017, 551, 609−613. (53) Dydio, P.; Key, H. M.; Nazarenko, A.; Rha, J. Y. E.; Seyedkazemi, V.; Clark, D. S.; Hartwig, J. F. An Artificial Metalloenzyme with the Kinetics of Native Enzymes. Science 2016, 354, 102−106. (54) Zhang, R. K.; Chen, K.; Huang, X.; Wohlschlager, L.; Renata, H.; Arnold, F. H. Enzymatic Assembly of Carbon−Carbon Bonds via IronCatalysed Sp3 C−H Functionalization. Nature 2019, 565, 67−72. (55) Zhao, J.; Bachmann, D. G.; Lenz, M.; Gillingham, D. G.; Ward, T. R. An Artificial Metalloenzyme for Carbene Transfer Based on a Biotinylated Dirhodium Anchored Within Streptavidin. Catal. Sci. Technol. 2018, 8, 2294−2298. (56) Pellizzoni, M. M.; Schwizer, F.; Wood, C. W.; Sabatino, V.; Cotelle, Y.; Matile, S.; Woolfson, D. N.; Ward, T. R. Chimeric Streptavidins as Host Proteins for Artificial Metalloenzymes. ACS Catal. 2018, 8, 1476−1484. (57) Juhás z, T.; Szeltner, Z.; Polgár , L. Truncated Prolyl Oligopeptidase From Pyrococcus Furiosus. Proteins: Struct., Funct., Genet. 2007, 69, 633−643. (58) Zhang, C.; Srivastava, P.; Ellis-Guardiola, K.; Lewis, J. C. Manganese Terpyridine Artificial Metalloenzymes for Benzylic Oxygenation and Olefin Epoxidation. Tetrahedron 2014, 70, 4245−4249. (59) Gu, Y.; Ellis-Guardiola, K.; Srivastava, P.; Lewis, J. C. Preparation, Characterization, and Oxygenase Activity of a Photocatalytic Artificial Enzyme. ChemBioChem 2015, 16, 1880−1883.

I

DOI: 10.1021/acs.accounts.8b00625 Acc. Chem. Res. XXXX, XXX, XXX−XXX