New Additions to the Arsenal of Biocatalysts for Noncanonical Amino Acid Synthesis

Abstract

Noncanonical amino acids (ncAAs) merge the conformational behavior and native interactions of proteinogenic amino acids with nonnative chemical motifs and have proven invaluable in developing modern therapeutics. This blending of native and nonnative characteristics has resulted in essential drugs like nirmatrelvir, which comprises three ncAAs and is used to treat COVID-19. Enzymes are appearing prominently in recent syntheses of ncAAs, where they demonstrate impressive control over the stereocenters and functional groups found therein. Here we review recent efforts to expand the biocatalyst arsenal for synthesizing ncAAs with natural enzymes. We also discuss how new-to-nature enzymes can contribute to this effort by catalyzing reactions inspired by the vast repertoire of chemical catalysis and acting on substrates that would otherwise not be used in synthesizing ncAAs. Abiotic enzyme-catalyzed reactions exploit the selectivity afforded by a macromolecular catalyst to access molecules not available to natural enzymes and perhaps not even chemical catalysis.

Graphical Abstract

Introduction:

Noncanonical amino acids (ncAAs) are prominent features in small molecules and peptide therapeutics, also known as peptidomimetics, used to treat many diseases [1,2]. The success of ncAAs in drug discovery is related partly to their resemblance to proteinogenic amino acids, typically (but not always) comprising an amine, a carboxylate, a sidechain, and stereogenic centers (Figure 1A). In turn, small molecules and peptidomimetics that contain ncAAs interact with biological targets akin to native peptides or proteins, adopting similar conformations and engaging in standard non-covalent interactions (e.g., hydrogen-bonding, electrostatic interactions, and pi-stacking, among others). Yet, ncAAs are not limited to the functional groups found in the twenty genetically encoded amino acids and instead employ myriad chemical motifs. This diversity of possible functional groups enables ncAAs to engage in new contacts with, and even covalently modify, biological targets, affecting their function and thereby creating opportunities for drug development.

Figure 1.

Enzymes have proven formidable tools for the synthesis of ncAAs found in many drugs. (A) Selected examples of proteinogenic and noncanonical amino acids. (B) Selected examples of approved and investigational drugs containing ncAAs [3,4,6]. (C) The process-scale synthesis of the antidiabetic sitagliptin uses an engineered transaminase that operates with near-perfect selectivity [4].

Examples of ncAAs in modern therapeutics include nirmatrelvir (Figure 1B), part of the lifesaving therapy Paxlovid for treating COVID-19 [3]. This peptide composed entirely of ncAAs contains a nitrile warhead that covalently binds the catalytic cysteine in main protease (M^pro), an essential enzyme in the viral cycle of SARS-CoV-2. Sitagliptin, an antidiabetic with a central β-ncAA is another notable example [4]: it inhibits the enzyme dipeptidyl peptidase 4 (DDP-4), which breaks down gastrointestinal peptidyl hormones that modulate blood glucose levels. Lastly, compound 44 is an investigational bicyclic peptide strung nearly entirely of ncAAs developed to inhibit the protein-protein interaction (PPI) of proprotein convertase subtilisin-like/kexin type 9 (PCSK9) and low-density lipoprotein receptor (LDLR). This mechanism of action is used to treat cardiovascular diseases with costly protein therapeutics [5,6]. More recently, new modalities such as cyclic peptides comprising ncAAs have effectively targeted atypical biological targets like PPIs [7–10]; they combine the outstanding potency and selectivity seen with biologics, where they can bind large, often featureless, protein surfaces with the attractive properties of small molecule drugs, like high stability and membrane permeability. Compound 44 has ‘antibody-like’ activity, made possible using ncAAs to improve potency, bioavailability, and specificity [11].

Given the triumphs of peptidomimetics in drug discovery, ncAAs will continue to be used to develop next-generation therapeutics. In keeping with this trend, investigations on the efficient synthesis of these privileged structures, especially in methodologies that sample vast, complex chemical space, are warranted. It is well appreciated that complexity, as measured by the presence of sp³ and chiral centers, is a good predictor of the success of molecules in the clinic [12,13]. In this regard, biocatalysis has much to offer, as it has delivered complex pharmaceuticals with unmatched levels of selectivity [14]. The use of biocatalysts in the pharmaceutical industry is growing rapidly due to numerous advancements, which include recombinant DNA technologies, next-generation sequencing, bioinformatics, high-throughput screening strategies, and directed evolution [15–19]. The use of iterative cycles of mutagenesis and screening to accumulate performance-enhancing mutations, known as directed evolution, has transformed how biocatalysts are perceived. Whereas reaction conditions were once fixed by what the enzyme required, directed evolution now permits scientists to tune enzymes for a much wider range of temperatures, pH, solvents, and substrate concentrations, greatly improving process performance. But the exquisite stereoselectivity enzymes impart is where they have really distinguished themselves from conventional chemocatalysts. The active sites of enzymes can be molded using directed evolution to accept a desired substrate and orchestrate reaction selectivity, leading to high control over emerging stereocenters. This molecular precision also facilitates chemoselectivity: enzymes effortlessly manage Lewis basic functional groups like those found in ncAAs that would typically poison small-molecule catalysts, resulting in processes devoid of protecting group manipulations and thus shortening synthetic routes. Moreover, biocatalysts conform to ten out of the twelve green chemistry principles, including the use of renewable materials, innocuous media, and safer procedures, making biocatalytic processes attractive from an economic and environmental perspective [20,21].

The process-scale synthesis of sitagliptin superbly captures the benefits of biocatalysis (Figure 1C). Directed evolution of a transaminase with no activity toward the prositagliptin ketone delivered a variant with 27 mutations capable of establishing the β-ncAA in sitagliptin with excellent selectivity (>99% ee, 92% assay yield) [4]. This biocatalytic approach supplanted a precious metal-catalyzed process, increased productivity by 53%, and reduced waste by 19%. The manufacturing of sitagliptin using biocatalysis is not unique; there is a rich history of using enzymes to synthesize active pharmaceutical ingredients (API), especially ncAAs. For instance, in nirmatrelvir, the l-tert-leucine and the bicyclic[3.1.0]proline, which are found in numerous APIs, can be obtained using a leucine dehydrogenase and an amine oxidase, respectively [14,22]. Moreover, in compound 44, the requisite 5-fluoro-tryptophan can be prepared using the β-subunit of tryptophan synthase (TrpB) [23].

The continued development of enzymes to synthesize ncAAs will contribute to developing new medicines. The biological world is rife with enzymes that can promote diverse chemistries, giving us a treasure trove from where to begin searching and developing biocatalysts to synthesize ncAAs. This brief review highlights recent efforts to expand the arsenal of enzymes that generate ncAAs and introduce newcomers to biocatalysis to the power of these remarkable catalytic machines. Our purview is limited to enzymatic reactions that produce ncAAs and have been published in the last two years. Moreover, we will not review biocatalytic cascades [24] or late-stage functionalization [25] of amino acids. Instead, we examine the potential of enzymes that promote abiological reactions to access complex ncAAs. We believe these enzymes will play a prominent role in the future because they act on unexplored substrate classes and enable entry to uncharted chemical space in ncAAs.

Reductive Amination with Imine Reductases

An effective strategy in growing the substrate scope of a biocatalytic reaction is to compile libraries of homologous enzymes for evaluation with substrates of interest. This was recently demonstrated by Turner and coworkers, who reported the identification and assembly of 384 sequence-diverse NAD(P)H-dependent imine reductases (IREDs) from metagenomic libraries, capable of accepting sterically demanding ketones and amines [26]. Enzymes from this library were used to synthesize N-substituted β-amino esters, prevalent β-ncAAs in pharmaceuticals like sitagliptin, from amines and β-keto esters (Figure 2A). Use of IREDs in amine synthesis solves ongoing challenges such as overalkylation and stereoselectivity. The α-position of β-keto esters is susceptible to racemization in aqueous conditions due to keto-enol tautomerization, which the authors used to create a dynamic kinetic resolution process. In an outstanding example, complementary IREDs granted access to three out of four possible isomers from the coupling of a cyclic β-keto ester and an amine, with excellent yields and, in one case, low diastereoselectivity. Reactions that furnish multiple diastereomers are often desired in the early stages of drug discovery as it facilitates stereochemical structure-activity relationship studies. Another report by the same group demonstrated that IREDs could convert α-keto esters to N-substituted α-amino esters, which are known to improve bioavailability [27].

Figure 2.

Recent examples of biocatalytic synthesis of ncAAs and a snapshot of their proposed mechanisms. (A) Reductive amination of amines and β-keto esters catalyzed by IREDs to synthesize N-substituted β-amino esters [26]. This reaction uses glucose dehydrogenase (GDH) and glucose to regenerate NADPH from NADP⁺. (B) Kilogram-scale hydroamination of fumaric acid with n-butyl amine catalyzed by an ammonia-lyase [28]. (C) Dynamic kinetic resolution of racemic β-alkyl aromatic pyruvic acids allows access to β-branched aromatic α-amino acids [31]. (D) Deuterium incorporation at the α-carbon using a PLP-dependent enzyme [34]. (E) The decarboxylative aldol reaction of l-aspartic acid and aldehydes afford γ-hydroxy amino acids [36]. (F) TrpB catalyzes amino-acrylate formation from serine and can accept enol-derived nucleophiles for conjugate addition [37].

Hydroamination with Ammonia-Lyase

Aspartate ammonia-lyase is an enzyme involved in primary metabolism, where it must be highly specific for its native function of reversibly deaminating aspartate to fumarate and ammonia. Distinct from IREDs, this enzyme does not require a cofactor and operates through acid-base reactions catalyzed by its protein scaffold. The use of this enzyme in the reverse direction is of interest as the products would constitute the formal hydroamination of olefins from two abundant precursors. Unfortunately, the high specificity of these enzymes has made it difficult to create more general catalysts. The combination of computational tools and directed evolution methodologies has proven effective in accelerating protein engineering campaigns. Wu and coworkers used a structure-based computational enzyme design to engineer an ammonia-lyase from Bacillus sp. YM551, AspB, to accept acrylate electrophiles and amine nucleophiles (Figure 2B) [28,29]. Reshaping the enzyme’s active site permitted access to aliphatic, polar, and aromatic α, β-unsaturated carboxylic acids. An AspB variant containing four mutations catalyzed the kilogram-scale synthesis of N-butyl-aspartic acid using whole cells, underscoring the scalability of biocatalytic reactions.

PLP-Dependent Enzymes can Catalyze Transamination, Deuteration, and C–C bond Formation

Pyridoxal phosphate (PLP)-dependent biocatalysts are used extensively to synthesize ncAAs and have been applied in the industrial synthesis of APIs [30]. Amino transferases are a class of PLP-dependent enzymes used to transfer amino groups selectively from an appropriate amine donor to an acceptor. Renata and coworkers identified a promiscuous thermophilic aromatic aminotransferase from Thermus thermophilus, TtArAT, that can catalyze the dynamic kinetic resolution of racemic β-alkyl aromatic pyruvic acids into β-branched aromatic α-amino acids with exquisite diastereo- and enantioselectivity (Figure 2C) [31]. Notably, the identified enzyme used l-glutamine (Gln) as the amine donor, whose deaminated byproduct, 2-oxoglutaramate, readily undergoes intramolecular cyclization to drive the transamination equilibrium in the desired direction. NcAAs containing multiple stereogenic centers are exciting products for drug discovery; they are conformationally more rigid, adopting a major, active conformer, and can engage in multiple interactions.

Deuterated drugs can improve half-lives, slow racemization of dynamic stereocenters, and enhance stability, leading to lower required dosing and, accordingly, fewer side effects [1,32,33]. PLP-dependent aminotransferases promote the deuteration of amino acids with some degree of substrate promiscuity, but their lack of site- and stereo-selectivity has impeded application in the synthesis of deuterated ncAAs. Narayan and coworkers demonstrated that SxtA AONS, a PLP-dependent enzyme that natively converts amino acids to α-amino ketones in the biosynthesis of saxitoxin, can incorporate deuterium from D₂O stereoselectively at the α-carbon of a large panel of l-α-amino esters and some amino acids (Figure 2D) [34]. In a complementary approach, Buller and coworkers used a two-enzyme system to label amino acids at the α-carbon and β-carbon positions [35]. The enzymes, an aminotransferase (DsaD) and a small partner protein (DsaE), were found from biosynthetic studies of l-alloisoleucine in Streptomyces scopuliridis, a ncAA found in several bacterial peptide natural products that form from l-isoleucine. DsaD, a PLP-dependent enzyme, can catalyze α-carbon deuteration through a mechanism analogous to the one shown by Narayan. In contrast, complexation of DsaD with DsaE promotes consecutive enamine formation and protonation with D₂O, exhaustively deuterating the α- and β-carbon positions. Resubmission of this deuterated product to DsaD and H₂O allows for exclusive β-carbon deuteration.

Whereas the biocatalytic toolbox is replete with enzymes that allow the seamless interconversion of functional groups, practical enzymes that form C–C bonds are not common. PLP-dependent enzymes can catalyze the formation of nucleophilic or electrophilic intermediates that can be used in C–C bond formation. Buller and coworkers demonstrated that UstD, a PLP-dependent enzyme involved in the biosynthesis of ustiloxin B, can form γ-hydroxy amino acids from aldehydes and l-aspartic acid through a decarboxylative aldol reaction (Figure 2E) [36]. UstD was evolved over three rounds of directed evolution and an additional round of computationally guided engineering to provide UstD^V2.0. This enzyme can promote decarboxylation of the sidechain of l-aspartate, forming a putative nucleophilic enamine intermediate. The enamine was shown to be capable of trapping structurally and electronically varied aldehydes giving the said products in high yield and selectivity. Although the focus of this report was limited to aldehydes, the authors also reported activity with a ketone substrate, demonstrating that the substrate scope is not limited to a single electrophile class.

In contrast to UstD, the β-subunit of tryptophan synthase (TrpB) catalyzes the formation of an electrophilic amino-acrylate intermediate from serine that can be intercepted with exogenous nucleophiles (Figure 2F) [23]. Watkins-Dulaney and coworkers demonstrated that TrpB can accept enols derived from ketones as nucleophiles and that directed evolution can improve this activity [37]. Moreover, the resultant products were found to undergo spontaneous cyclization to pyrroline derivatives that can serve as precursors to noncanonical prolines via reduction. The wealth of mechanistic paradigms accessible to PLP-dependent enzymes makes them exceptionally versatile in accessing diverse ncAAs from different starting materials. This wealth is only increasing: recently discovered PLP-dependent enzymes that catalyze the formation of alkynes [38] and [3+2] annulations [39] add new directions to explore in ncAA synthesis.

Exploring Biocatalytic New-to-Nature Reactions for ncAA Synthesis

The biocatalytic reactions described above show the readiness with which enzymes can accept new substrates and how directed evolution can improve their activity. However, these biological machineries are limited in the chemical space they sample, as they operate on a narrow set of functional groups. Chemocatalysts, and the diversity of transformations they perform, significantly exceed established biocatalytic methodologies in scope. Therefore, researchers have sought to introduce and engineer new mechanistic paradigms onto protein scaffolds to grow the classes of reactions and, consequently, the substrates accessible to biocatalysts. This expansion of reactivity has been particularly effective when exploiting cofactor versatility, often guided by drawing parallels between cofactors and their chemocatalyst equivalents [40,41]. Because metal-catalyzed reactions are at the core of modern synthetic chemistry, metalloenzymes — and Fe-heme proteins in particular — have been exploited to achieve expanded biocatalytic reactivity.

Since the discovery of the latent ability of heme proteins to participate in carbene transfer a decade ago, many fundamental reactions known in synthetic chemistry have now been successfully recapitulated enzymatically [42]. The rapid growth of this platform has shifted the focus of these new-to-nature reactions toward accessing products of medicinal value like ncAAs. Taking inspiration from dual catalytic strategies in chemical catalysis for enantioselective N–H insertion reactions, Liu and coworkers sought to identify an enzyme that could promote ammonium ylide formation and enantioselective protonation [43]. Libraries of P411 carbene transferases generated from cytochrome P450_BM3 from Bacillus megaterium were challenged with an amine and α-diazolactone as the carbene precursor (Figure 3A). This effort identified variant L7_FL as an enzyme capable of promoting highly enantioselective N–H insertion on alkyl and aryl, primary and secondary amines, providing α-amino-γ-butyrolactones that can be readily converted to ncAAs.

Figure 3.

Engineered Fe-heme proteins from different organisms and with disparate functions can promote carbene transfer. (A) Enantioselective N–H insertion of amines with α-diazolactone catalyzed by a P411 carbene transferase [43]. (B) Enantiodivergent synthesis of chiral trifluoroethylamines via N–H insertion with an acceptor-acceptor carbene catalyzed by Ht-Cc552 [44]. (C) [1,2]-Stevens rearrangement of aziridinium ylides from aziridines and ethyl diazoacetate to afford ring-expanded azetidines catalyzed by P411-AzetS [45].

New-to-nature heme enzymes have been demonstrated to be exceptionally promiscuous, readily accepting structurally diverse carbene precursors and carbene acceptors [46]. Recently, Fasan and coworkers engineered cytochrome c₅₅₂ from Hydrogenobacter themophilus (Ht-Cc552), to promote the N–H insertion of aryl amines with an acceptor-acceptor carbene donor, affording chiral trifluoroethylamines (Figure 3B) [44]. Strategic deployment of ncAAs bearing chiral trifluoroethylamines in peptides and peptidomimetic small molecules eliminates metabolic liabilities such as hydrolysis by proteases [47,48]. Intriguingly, when the identity of the ester group on the carbene precursor was varied Ht-Cc552 could promote enantiodivergent N–H insertion, offering a complementary approach where both enantiomers are obtained without additional engineering.

In a striking example of bringing new chemistry to life, Miller and coworkers discovered and engineered a variant of cytochrome P450_BM3 that catalyzes the [1,2]-Stevens rearrangement of aziridinium ylides from aziridines and ethyl diazoacetate, providing enantioenriched azetidines (Figure 3C) [45]. This reaction is unknown on this substrate and is the first report of highly enantioselective [1,2]-Stevens rearrangement. The obtained azetidines are bioisosteres of proline and have been applied in developing small-molecule therapeutics [1]. These new-to-nature reactions are selective biocatalytic alternatives to known multi-step chemical approaches.

Conclusion

The continued growth in the application of ncAAs in drug discovery necessitates efficient and scalable production as well as new methods that grant access to their vast chemical space. In many respects, biocatalysis is unmatched in accessing these important molecules: enzymes offer strict control over nascent stereocenters and are evolvable to meet user-defined metrics, providing a reliable platform for continued catalyst development. Future research in the biocatalytic synthesis of ncAAs should continue to focus on the evaluation and engineering of enzymatic activity with nonnative substrates and the discovery and development of new enzymes for use in ncAA synthesis, for example, through the study of the biosynthesis of secondary metabolites or homology searches in metagenomic libraries. New-to-nature reactions will also play a prominent role in this endeavor by acting on untapped pools of starting materials in biocatalysis. Although natural enzymes can act on nonnative substrates, they still require the necessary functional groups they natively use in catalysis. By teaching enzymes to perform abiological reactions on new substrate classes, we will expand the arsenal of biocatalysts and access new collections of ncAAs.

Highlights:

Noncanonical amino acids are prominent structural elements in pharmaceuticals.

Enzymes are formidable tools for synthesizing ncAAs.

Enzymes can be modified and optimized using directed evolution.

New-to-nature enzymes act on unexplored substrate classes in biocatalysis.