Metallopeptide Catalysts and Artificial Metalloenzymes Containing Unnatural Amino Acids

Abstract

Metallopeptide catalysts and artificial metalloenzymes built from peptide scaffolds and catalytically active metal centers possess a number of exciting properties that could be exploited for selective catalysis. Control over metal catalyst secondary coordination spheres, compatibility with library based methods for optimization and evolution, and biocompatibility stand out in this regard. A wide range of unnatural amino acids have been incorporated into peptide and protein scaffolds using several distinct methods, and the resulting unnatural amino acid containing scaffolds can be used to create novel hybrid metal-peptide catalysts. Promising levels of selectivity have been demonstrated for several hybrid catalysts, and these provide a strong impetus and important lessons for the design of and optimization of hybrid catalysts.

Introduction: Hybrid Metal-Peptide Catalysts and Expanding Hybrid Catalyst Reactivity Using Unnatural Amino Acids

Catalytic processes form an integral part of modern methods for producing commodity chemicals, fine chemicals, pharmaceuticals, and a wide range of other materials. The potential to exploit catalyst reactivity for applications, such as interrogating or augmenting the function of living organisms, outside the confines of flasks and specialized reactors has driven the development of catalysts ranging from metal complexes to engineered enzymes that function in complex media.¹ In all of these cases, catalyst selectivity, whether to produce a single enantiomer of a pharmaceutical under highly controlled conditions or to react with a target substrate in cellular milieu, is essential. Many natural enzymes catalyze reactions on their native substrates with near perfect efficiency (although this is far from universal),² and the adaptability of enzymes has been exploited to evolve enzymes with impressive levels of activity and selectivity toward unnatural substrates and reactions³. Essential to directed evolution efforts, however, is some initial activity to optimize, and chemists have devised a wide range of powerful transformations that find no analogues in nature. Furthermore, many potentially interesting enzymes are difficult to express, particularly in yields sufficient for high throughput analysis of reactions in the parallel formats required for directed evolution.

In hopes of combining the selectivity and adaptability of enzymes with the reactivity of metal catalysts, researchers have explored different methods to incorporate non-natural metal cofactors into peptides, proteins, and enzymes.⁴ Many approaches to form metallopeptide catalysts and artificial metalloenzymes (ArMs) have thus been developed using peptides and proteins comprised of the 20 natural amino acids. These systems can be roughly categorized as involving metal coordination, covalent attachment of substituted catalysts, and noncovalent anchoring of substituted catalysts. Unnatural amino acids (UAAs) are now routinely incorporated into peptides and proteins to enable functions beyond those accessible using natural amino acids.⁵ A growing family of UAAs have specifically been used for metallopeptide catalyst and ArM formation (Fig. 1). The syntheses of these UAAs will not be discussed in detail, but it should be noted that these are often complicated by the metal binding capabilities of the amino acid moiety itself and the need to minimize the distance between the amino acid C_α and the metal center (upon installation).⁶ This situates metal centers proximal to the stereochemical information at C_α, decreases movement of the metal center within the scaffold, minimally disrupts protein folds, and, important for some methods of incorporation, maximizes homology to native amino acids.

Figure 1.

(a) General scheme for construction of hybrid metal-peptide catalysts via coordination or bioconjugation of metal catalysts using a UAA. (b) Structures of UAAs used in the synthesis of metallopeptide catalysts and ArMs. Racemic mixtures were used for most UAAs incorporated via codon suppression since only the L isomer is accepted.

To date, UAAs have been used to enable new modes of metal coordination or covalent attachment (Fig. 1a), which has led to construction of many new ArMs and metallopeptides not possible using natural amino acids. Direct incorporation of organometallic UAAs into proteins⁷ and peptides⁶ has also been demonstrated, although only in the latter case were catalytically active metal centers incorporated. The unique physical properties of UAAs have also been used to investigate or alter the activity of native metalloenzymes. Together, these studies have revealed several benefits of incorporating UAAs into hybrid catalysts. These include the ability to introduce at a single site within a peptide scaffold: catalytically active residues, high affinity or chelating non-natural ligands capable of forming catalytically active metal centers upon metal binding, residues with unique physical properties (redox, pK_a, etc.), and bioorthogonal handles for installation of metal catalysts.

This review will cover examples of metallopeptide catalysts and ArMs in which new catalytic activity is imparted to a peptide or protein scaffold or existing activity of a metalloenzyme is improved by UAAs. The majority of these examples were reported within the last ten years. The discussion is organized based on the method used to incorporate the UAA, and reviews that describe these methods in detail are cited. To ensure concise coverage of this topic, this review will not discuss in detail: hybrid catalysts that do not contain UAAs⁴ or possess scaffolds with significant non-peptide composition, non-catalytic metalloproteins containing UAAs,⁸ or optimization of enzymes lacking metals using UAAs⁹. Excellent reviews that cover these areas with varying degrees of overlap with the current subject are available.

Selenocysteine and the Need for Multiple Methods of UAA Incorporation

Based on the first of the above criteria, selenocysteine (1) falls outside of the scope of this review (it is not a UAA)¹⁰, but several examples of its incorporation into peptides and proteins will be discussed due to their relevance to current hybrid catalyst design efforts. Some general considerations are consolidated here. As its name implies, 1 is the Se analogue of cysteine, and while many properties of these amino acids are similar, 1 possesses a lower X-H pK_a (X=S, Se), greater nucleophilicity, and a lower reduction potential. 1 is co-translationally incorporated into proteins via suppression of a UGA stop codon, and, interestingly, this process involves conversion of a unique seryl-tRNA with a complementary UCA anticodon to the corresponding selenocysteyl-tRNA. The reactivity of 1 conveyed to natural selenoproteins and enzymes and synthetic selenopeptide catalysts and ArMs. Synthetic systems in which 1 is used as a ligand for metal catalysts or as a nucleophilic or redox catalyst itself have been designed (Fig. 1a).¹¹ Furthermore, 1 has been incorporated into peptide scaffolds using every method that has been reported to date for incorporating UAAs into hybrid catalysts: chemical synthesis, chemical mutagenesis, chemical ligation, auxotrophs, and codon suppression. The flexibility with which this amino acid can be incorporated into peptides has facilitated the development of a range of peptide-based selenium catalysts. While this will not be possible for all UAAs, it provides a clear illustration of the potential to impart new catalytic activity to a peptide scaffold using a single amino acid substitution and motivation for the development of multiple methodologies for incorporating any given UAA.

Peptide Synthesis

UAAs can of course be incorporated into peptides via chemical synthesis, and peptides up to approximately 50 residues in length can be readily prepared using solid phase peptide synthesis. These structures can adopt a range of secondary structural features (α-helices, β-turns, etc.) that can be exploited for selective catalysis. Assuming the side chains of the UAAs are compatible with the reagents typically used for solid phase synthesis, any desired UAA can be used. This requirement is met by a wide range of UAAs with metal binding ligands, which can be subsequently metallated following peptide synthesis (Fig. 1a).

Gilbertson provided several early examples of metallopeptide catalyst synthesis using phosphine-containing UAAs 3–5.¹² α-Helical peptides containing derivatives of 3 at proximal positions (e.g. i and i+1, 3, and 4) were shown to catalyze enantioselective hydrogenation of 2-acetamidoacrylate upon metallation with Rh(I) (Fig. 2a). While low enantioselectivity was initially observed, combinatorial optimization of the peptide sequence on solid support provided significant increases in selectivity (up to 38% ee).¹³ Different turn motifs, including Pro-D-Yyy, where D-Yyy is a D-amino acid, were also used to construct peptides containing the sequence Xxx-Pro-D-Yyy-Zzz, where Xxx and Zzz are also derivatives of 3. These peptides were metallated using [PdCl(η³-C₃H₅)]₂ to generate bisphosphine metallopeptide catalysts for asymmetric allylic alkylation (Fig. 2b). Optimizing peptide structure using a 96 member peptide library on solid support and varying phosphine substitution led to the identification of a peptide catalyst that provided up to 95% ee for the reaction of 3-acetoxycyclopentene with dimethylmalonate.¹⁴

Figure 2.

Metallopeptide catalyzed (a/b) hydrogenation and allylic alkylation using 3, (c) Diels-Alder reaction using 7, (d) hydrosilation using 8, and (e) transfer hydrogenation and imidate rearrangement using 9 and 10.

Gibney similarly reported the synthesis of 4-pyridylalanine (6) containing α-helical peptides that assembled into a 4-helix bundle and bound heme via pyridine coordination.¹⁵ While no catalysis was demonstrated using the resulting metallopeptides, both 3-pyridylalanine (7) and 6 were more recently used by Roelfes to synthesize metallopeptide catalysts. In this work, bovine pancreatic polypeptide (bPP), a 36-mer that adopts a helix-turn-helix tertiary structure and an antiparallalel homodimer quaternary structure was used as a scaffold.¹⁶ The natural peptide was truncated to 31 residues, and histidine, 6, or 7 were incorporated into the peptide 7-position using solid phase synthesis to generate bPP variants. Spectrophotometric analysis indicated that monomeric bPP-Cu metallopeptides formed in the presence of Cu(II). These metallopeptides catalyzed both the Diels-Alder reaction of azachalcones and cyclopentadiene with good levels of enantioselectivity (up to 83% ee with 73% conversion at 15 mol% Cu(II)/33 mol% peptide, Fig. 2c) and the Michael addition of dimethylmalonate to azachalcones (up to 86% ee with 85% conversion, also at 15 mol% Cu(II)/33 mol% peptide). Notably, only peptides containing the 3-PyrAla residue provided significant levels of enantioselectivity, and a 3.5 fold rate acceleration was observed in the presence of the best catalyst. Here also, scaffold mutations were used to improve metallopeptide catalyst selectivity.

Albrecht described the synthesis of short peptides containing a C-terminal N_δN,_ε -dimethylated histidinium UAA (8) that could be metallated using Rh(I) to form metallopeptide catalysts for ketone hydrosilation (Fig. 2d).¹⁷ A methionine residue was incorporated at the i+3 position of the peptide to chelate to the Rh center, but only minor changes in activity and selectivity were observed relative to a Rh-SMe₂ ligated derivative of 8 itself. Later work by this group further explored the impact of peptide sequence on metallopeptide catalyzed hydrosilation and established that both Cp*Ir and Rh-phosphine complexes could be generated from peptides containing 8.¹⁸

A key limitation of metal binding amino acids is that many metal catalysts cannot be formed by simply stirring together metals and ligands. If the conditions required for formation of a desired catalyst are not compatible with a peptide scaffold, metal binding UAAs cannot be used to form the corresponding metallopeptide catalysts or ArMs. To bypass this limitation, we developed a versatile approach to synthesize UAAs with catalytically active organometallic side chains (e.g. 9 and 10).⁶ We demonstrated that these amino acids could be incorporated into short peptides, including known β-turn scaffolds (Fig. 2e). This work was inspired by a number of reports of organometallic amino acids that have been used to prepare metallopeptides, but for which no catalysis was demonstrated.^19–21 In general, the large number of examples in which UAAs are used to create metal binding,^22–25 but not catalytically active peptides clearly speak for the significant challenges that face the design of hybrid catalysts.

Chemical Mutagenesis

The largest scaffold of the metallopeptide catalysts noted above is 31 residues. This is much shorter than most enzymes and lacks much of the functional potential of native protein folds. Early efforts to install UAAs and other functionality into native proteins involved “chemical mutation” of those proteins. Hilvert used this approach to create the first artificial selenoenzyme, selenosubtilisin.²⁶ Briefly, this involved activation of the active site serine (Ser221) with phenylmethanesulfonyl fluoride followed by displacement of the resulting sulfonate with hydrogen selenide (Fig. 3a). While selenosubtilisin lost much of the hydrolysis activity of the native enzyme, it gained peroxidase activity analogous to that of known selenoenzyme, glutathione peroxidase, and was able to reduce alkyl peroxides to the corresponding alcohols.²⁷ This reaction proceeds via oxidation of the active site selenol by the alkyl peroxide to a selenoxide (or further oxygenated species), which is sequentially reduced by two equivalents of glutathione to regenerate the active selenol (Fig. 3b).

Figure 3.

(a) Formation of selenosubtilisin via chemical mutagenesis. (b) Mechanism of selenosubtilisin-catalyzed peroxide reduction. (c) Representative reaction scope (¹confirguration not determined; ²opposite sense of induction observed for A and P).

Subsequent researchers found that seleosubtilisin could deracemize alkyl peroxides to generate enantioenriched alcohols (Fig. 3c).²⁸ Notably, selenosubtilisin possessed a substrate scope for peroxidase activity analogous to that of subtilisin for hydrolase activity, and this could be used to predict the outcome of peroxide reduction reactions of different substrates.²⁹ This result highlights a key advantage to using known folds as ArM scaffolds: native function (e.g. substrate binding) can be maintained and exploited to control reactivity imparted by UAAs. This work also highlights a key benefit of using highly stable scaffolds, such as subtilisin, for hybrid catalyst formation: they can tolerate a wide range of chemistries on multigram, preparative scales. More recently, the active site serine of subtilisin was converted to a tellurocysteine (2), a bonafide, catalytically active UAA, albeit a metalloid rather than a transition metal.³⁰ The resulting artificial telluroenzyme possessed activity similar to that of Se variant. While chemical mutagenesis is therefore a convenient means to introduce functionality into proteins, it requires a uniquely reactive residue within the protein for the requisite chemical transformations. The use of this technique to introduce UAAs with structures analogous to the 20 natural amino acids, rather than to accomplish what was referred to in the introduction as “covalent attachment of substituted catalysts”,⁴ has been limited to only a handful of UAAs (e.g. 1 and 2), but other metal binding UAAs should be accessible via this approach.

Amino Acid Auxotrophs

Culturing strains auxotrophic for a particular amino acid in the presence of a structural analogue of that amino acid (e.g. methionine and selenomethionine, 11) has long been used to incorporate UAAs into proteins in a process termed isomorphous replacement.⁵ Several examples of artificial enzymes created by incorporating 1 into proteins expressed in cysteine auxotrophic strains possess peroxidase activity analogous to that outlined above.³¹ More recently, 2 was incorporated into a glutathione transferase to create a glutathione peroxidase.³² While auxotrophic systems thus eliminate the need for chemical modification of protein scaffolds, they can only be used to incorporate UAAs sufficiently similar to natural amino acids for in vivo incorporation, and genetic construction of the auxotrophic strains themselves is required. Furthermore, global incorporation of the UAA throughout the scaffold (and host organism) occurs, which complicates the design of selective hybrid catalysts (although it is quite useful for other applications).

Chemical Ligation

While chemical mutagenesis and amino acid auxotrophs have been used to create artificial enzymes via incorporation of 1 and 2 into protein scaffolds, incorporation of other UAAs to enable formation of other ArMs using these methods has not been reported, presumably due to the many limitations noted above. Various chemical ligation strategies, including native chemical ligation (NCL) and expressed protein ligation (EPL), have been developed to enable the synthesis of a wide range of proteins and enzymes.³³ These methods bypass problems that may arise with in vivo expression of scaffold proteins and greatly expand the range of scaffolds into which UAAs can theoretically be incorporated with no inherent limitation on UAA structure. While these methods have been used to incorporate 1, 11, and a wide range of true UAAs,³⁴ remarkably, none of the latter were intended for metal binding or catalysis.

Even examples in which 1 and 11 are used for metal binding or catalysis are relatively rare. For example, NCL was used to incorporate 1 into glutaredoxin to generate an artificial enzyme with peroxidase activity, again, analogous to the several examples outlined above (Fig. 3b).³⁵ The Lu group has extensively studied the effects of substituting with UAAs residues that coordinate the active site Cu(II) of azurin.³⁶ EPL was used to replace Cys112, which forms a strongly covalent equatorial S-Cu bond in azurin, with 1.^37,38 This marked the first instance in which 1 was incorporated into the active site of a metalloprotein. The absorption spectrum of azurin-1-Cu was dominated by a transition assigned to the Se-Cu(II) charge transfer band, which was red-shifted nearly 50 nm from that of the S-Cu(II) absorbance in native azurin. Interestingly, the reduction potentials of azurin and azurin-1-Cu were quite similar (328 and 316 mV s NHE, respectively), and the authors suggested that compensating geometric changes in the active site were responsible for moderating the effects that would otherwise be expected for sulfur to selenium substitution. Subsequent studies focused on replacing Met121, which coordinates to an axial site on the active site Cu(II), with a variety of UAAs, including 11, using EPL.^39,40 A range of detailed studies ultimately revealed that hydrophobic effects dominate changes to azurin reduction potential observed upon variation of the residue at position 121, and this work was recently reviewed.³⁶ Despite the potential to introduce via chemical ligation essentially any type and number of metal-coordinating functionalities into protein scaffolds for applications in metalloenzyme design, these methods seem under-exploited for this capacity to date.

Codon Suppression Methods

To expand the range of amino acids that can be genetically encoded into proteins,⁵ extensive effort has been devoted to engineering various components of protein biosynthetic pathways in different organisms. The full scope of these efforts, including modification of elongation and release factors, tRNAs, aminoacyl-tRNA synthetases (aaRSs), ribosomes, and the genomes of expression hosts,⁴¹ cannot be summarized here. Essential to the use of this approach for UAA incorporation, however, is engineering an orthogonal tRNA/aaRS pair that selectively recognizes (suppresses) a nonsense codon by incorporating the desired UAA without interfering with any endogenous tRNA/aaRS pairs. This has been most frequently accomplished in E. coli using an engineered Methanococcus jannaschii amber stop codon suppressor tyrosyl-tRNA/tyrosyl-tRNA synthetase (TyrRS) pair (Fig. 4a).⁴² To enable incorporation of UAAs for ArM formation, multiple rounds of positive and negative selection are used to evolve the amino acid substrate scope of the TyrRS to include the desired metal binding UAA. An amber stop codon must be introduced into the scaffold gene, and the scaffold is co-expressed with the orthogonal tRNA/aaRS pair in the presence of the desired UAA.

Figure 4.

(a) General scheme for incorporating UAAs into scaffold proteins using amber codon suppression. (b) Cartoon representation of a CAP-12 ArM nuclease. (c) Optimization of LmrR-12 ArMs for enantioselective Friedel-Crafts acylation of indoles.

In 2007, Schultz first demonstrated the use of this method to genetically encode a metal binding UAA, bipyridylalanine (12), into T4 lysozyme in E. coli.⁴³ It is worth noting that the selection scheme developed by the Schultz group to identify aaRSs for a wide range of UAAs was initially unsuccessful in the case of 12. Instead, a substrate walking approach was used, which involved first engineering an aaRS with activity on biphenylalanine, and the using this enzyme as a parent to engineer a variant with activity on 12. Given the significant structural differences between many metal-binding ligands, most organometallic complexes, and the side chains of the 20 natural amino acids, substrate walking will likely be a common requirement in efforts to engineer aaRSs compatible with UAAs intended for ArM formation. The ability of a T4 lysozyme-12 mutant to bind Cu(II) was confirmed by mass spectrometry and UV-Vis spectroscopy. Cu(II)-bipyridine complexes react with O₂ in the presence of sulfide reducing agents to generate ROS (e.g. hydroxyl radical) that randomly cleave nucleic acids and proteins. Several groups have developed hybrid catalysts exploiting this reactivity for selective nuclease and protease activity.⁴ Indeed, Schultz found that incorporating 12 into the E. coli catabolite activator protein (CAP) at a site proximal to the CAP-DNA interface led to the formation, after metallation with Cu(II), of a site specific ArM nuclease (Fig. 4b).⁴⁴

This methodology has since been used to genetically encode 12 into other proteins for ArMs formation. For example, Baker and Stoddard used 12 in several metalloproteins designed using the Rosetta software package developed in the Baker lab.⁴⁵ While no catalytic activity was demonstrated in these metalloproteins, several important design lessons, which will likely prove useful for future ArM engineering efforts using metal binding UAAs, were identified. For example, initial efforts to metallate scaffolds containing 12 with Fe(II) led to structures in which the loop containing 12 was exposed to the solvent rather than packed into the designed active site. This was proposed to result from the well known tendency of bipyridines, such as 12, to form tris-chelate complexes with divalent metals. Coordination of two additional molecules of 12 to the desired scaffold-12-Fe(II) complex would lead to such a tris-chelate complex, which could not be sterically accommodated within the designed active site. This problem was eliminated by designing a metal binding site with additional natural amino acids that constrained the UAA to bind metals within the desired ArM active site rather than allowing metal binding to excess 12 to alter the scaffold fold. While effective, this approach requires precise initial placement of both the UAA and natural amino acid residues within the active site--a challenge that metal binding UAAs were intended to mitigate.⁴³ The use of highly stable scaffolds that can fully contain metal binding UAAs could eliminate this problem by precluding the formation of unwanted ligation to begin with. Further active site design could then be focused on installing suitable functional groups for synthetic transformations rather than ensuring tight metal binding.

Very recently, Roefles showed that incorporating 12 into the hydrophobic dimer interface of LmrR (Lactoccocal multidrug resistance regulator) led to the formation, after metallation with Cu(II), of ArMs that catalyze a Friedel-Crafts acylation.⁴⁶ As in their previous work using Cu(II)-bipyridine catalysts covalently linked to cysteine residues in LmrR,⁴⁷ both the site at which 12 was incorporated and residues proximal to the putative active site Cu(II) catalyst could be used to optimize the enantioselectivity of the acylation reaction (Fig. 4c). Ultimately, up to 83% ee could be obtained for acylation of 2-methylindole through these efforts.

While 12 has thus received the most attention from researchers due to the availability of a robust tRNA/aaRS orthogonal pair for this UAA, Wang has made a concerted effort to expand the range of metal-binding UAAs that can be genetically encoded into proteins.⁸ Thiomethyl- and imidazole-substituted tyrosine derivatives 13 and 14 were synthesized and genetically encoded into the tyrosine 33 position of a myoglobin variant containing a designed copper binding site to create ArM models for probing the function of tyrosine-cysteine⁴⁸ and tyrosine-histidine⁴⁹ crosslinks in native heme enzymes. Encoding 13 led to creation of a cytochrome c nitrite reductase mimic that reduced hydroxylamine to ammonia with four fold higher activity than the scaffold lacking the UAA. In a similar fashion, encoding 14 led to a heme copper oxidase mimic that catalyzes O₂ reduction to water with significantly enhanced efficiency relative to analogous tyrosine variants. These improvements were attributed to the subtle differences between the pK_a and reduction potential of the phenol moiety in 13 and 14 relative to that in tyrosine, in analogy to similar explanations proposed for the effects of tyrosine-cysteine and tyrosine-histidine crosslinking in cytochrome c nitrite reductase and heme copper oxidase. Both Schultz and Wang have genetically encoded several additional metal binding UAAs (15–17) into proteins to generate non-catalytic metalloproteins, although it could be possible to use these for ArM formation by binding appropriate metal ions.^50–52 Finally, Schultz also demonstrated direct incorporation of a ferrocenyl UAA (18) into a protein,⁷ and while this was notable in that it is the only example in which a organometallic UAA has been directly incorporated into a protein, it is not, of course, catalytically active.

Robust methods for aaRS engineering^41,42 and the remarkably broad scope of engineered aaRS variants⁵³ has enabled incorporation of a wide range of UAAs,⁵ including those noted above for ArM formation. The repertoire of UAAs available for ArM formation will no doubt increase as our ability to manipulate protein biosynthesis improves.⁴¹ Furthermore, the combination of stop codon suppression methods with cell free protein synthesis methods⁵⁴ has the potential to greatly expand the range of UAAs that can be considered for ArM formation by removing potential incompatibilities between UAAs and living cells. Despite these and other potential advances, it will almost certainly not be possible to genetically encode UAAs bearing a wide range of important metal-binding ligands, let alone metal complexes, for ArM formation.

Several UAAs, however, have been developed to install bioorthogonal functional groups for bioconjugation reactions.⁵⁵ While this approach has been used extensively for incorporating a wide range of probes, including metal complexes, into biomolecules, we recently showed that it could also be used to incorporate catalytically active metal complexes into protein scaffolds. Our approach involves genetically encoding an azidophenylalanine residue (19) into a scaffold protein of interest and synthesizing a derivative of the desired catalyst with a bicyclo[6,1,0]nonyne substituent.⁵⁶ Rapid strain promoted azide-alkyne cycloaddition between these two components provides the desired ArM in high yield. This method preserves the broad scaffold/cofactor scope of ArM formation via covalent scaffold modification while eliminating selectivity problems associated with bioconjugation reactions to native amino acid residues. We demonstrated that this method could be used to install Mn-terpyridine, Cu-terpyridine, and dirhodium tetracarboxylate catalysts into a model scaffold. We have since identified scaffolds that can be engineered to impart high levels of selectivity to dirhodium tetracarboxylate catalysts, enabling styrene cyclopropanation with >90% ee (unpublished results).

Conclusion and Outlook

As summarized above, several methods can be used to introduce UAAs into peptide scaffolds and thus prepare metallopeptide catalysts and ArMs. Each of these methods has advantages and disadvantages that must be evaluated based on the nature of the hybrid catalyst target. As in any synthetic endeavor, having many options available with broad substrate scope (i.e. different UAAs and scaffolds) to access such complex targets (metallopetide catalysts and ArMs) will greatly facilitate progress in the field. From a substrate perspective, fundamental methodology development should focus on UAAs with clear relevance to catalysis, rather than metal binding alone, and on scaffolds that can provide robust active sites to contain metal catalysts. The unique synthetic challenges of synthesizing relevant UAAs and assembling UAAs and scaffolds to form functional catalysts cannot be marginalized nor obfuscated with related efforts intended for applications other than catalysis. Improving the compatibility of peptide synthesis and chemical ligation methods with metal containing UAAs, engineering aaRSs to accept UAAs substituted with diverse ligands, metal complexes, and other catalytic moieties, and developing methods for bioconjugation of metal complexes to UAAs in complex media are all important methodology goals.

From the target perspective, metallopeptide catalysts and ArMs with properties superior not only to analogues lacking UAAs, but to any other catalyst system, must be pursued. Achieving this goal will involve improving both methodology for incorporating UAAs into peptide scaffolds and methods to design and optimize metallopeptide catalysts and ArMs. Computational protein design methods and molecular dynamics simulations can assist in these efforts, and improving the ease by which non-native metal cofactors can be included in these calculations would greatly facilitate use of these techniques for hybrid catalyst design. Perhaps the greatest impediment to developing hybrid catalysts employing UAAs (and ArMs in general), however, is the lack of library methods for iterative scaffold optimization.⁴ Small peptide libraries and scaffold point mutations have been used to improve hybrid catalyst selectivity, but these encouraging efforts are far narrower in scope than current methods for selecting peptides for other applications or evolving natural enzymes.³ As similar optimization of hybrid catalysts becomes possible, so too might be realized the potential of these species to not only effect selective transformations, but to enable these transformations in living systems or to enable new transformations not possible with small molecule catalysts or enzymes.

Highlights.

• Metal-peptide hybrids have unique potential for catalysis.

• Unnatural amino acids (UAAs) can be used to create novel hybrid catalysts.

• Several methods have been developed to incorporate UAAs into peptide scaffolds.

• Promising levels of selectivity have been demonstrated for several hybrid catalysts.

• Methods to design and optimize hybrid catalysts are needed for practical applications.