Emerging enzymatic strategies for the synthesis of α-tertiary amino acids

Abstract

α-Tertiary amino acids (ATAAs), featuring a fully substituted α-carbon bearing two carbon substituents, are valuable building blocks in natural products, pharmaceuticals, and agrochemicals, yet their stereoselective synthesis remains a longstanding challenge in conventional chemical synthesis. Recent advances in enzymatic catalysis have provided efficient and sustainable solutions for ATAA construction. This review summarizes the advances in enzymatic strategies for ATAA synthesis reported over the past five years, with particular emphasis on pyridoxal 5′-phosphate (PLP)-dependent enzymes and non-heme iron enzymes. We highlight newly uncovered catalytic paradigms, including asymmetric radical C–C bond formation, Mannich-type reactions, [3 + 2] annulation, oxidative decarboxylation, and nitrogen migration. Photoenzymatic approaches that enable reactivity beyond canonical enzymatic pathways are also discussed. Finally, we consider the remaining challenges in expanding the chemical space and achieving predictable stereocontrol, and emphasize the opportunities arising from enzyme discovery, directed evolution, and photoenzymatic catalysis for ATAA synthesis.

α-tertiary amino acids (ATAA) are valuable building blocks yet their stereoselective synthesis remains a longstanding challenge. Here, the authors summarize recent advances in enzymatic catalysis that have provided efficient and sustainable solutions for ATAA construction.

Introduction

α-Tertiary amino acids (ATAAs) are non-proteinogenic amino acids characterized by a fully substituted α-carbon bearing two carbon substituents¹. This sterically congested motif is prevalent in natural products, pharmaceuticals, and other bioactive molecules, where it can profoundly influence conformational preference, proteolytic stability, and molecular recognition^2–5. Therefore, ATAAs are widely valued as privileged building blocks in medicinal chemistry, peptide design, and asymmetric synthesis¹.

Despite their importance, the stereoselective synthesis of ATAAs remains a longstanding challenge. The construction of a quaternary α-carbon center adjacent to both amino and carboxyl groups is intrinsically demanding because of the severe steric congestion and the need to control multiple reactive functionalities within a compact framework^1,6–9. Although modern chemical methods have significantly enhanced access to ATAAs, general solutions that combine broad substrate scope, predictable stereocontrol, and operational simplicity remain difficult to achieve^10–15.

Enzymatic catalysis offers a fundamentally different approach to this problem¹⁶. Instead of relying on permanently installed activating groups or highly reactive synthetic equivalents, enzymes generate and control reactive intermediates transiently within a defined active site^17–19. Among them, pyridoxal 5′-phosphate (PLP)-dependent enzymes are especially powerful because the PLP cofactor enables diverse transformations at the α-position of amino acids through aldimine, quinonoid, enamine, and iminium intermediates^20–22. In parallel, non-heme iron enzymes provide complementary strategies in which oxidative decarboxylation, radical relay, or heteroatom migration can reveal or install ATAA frameworks through pathways that differ fundamentally from classical synthetic disconnections^23–25.

In this review, we summarize recent progress in enzymatic strategies for ATAA synthesis, with particular emphasis on PLP-dependent enzymes and non-heme iron enzymes. To place this focused coverage in context, readers seeking a broader foundation in earlier biosynthetic routes to nonproteinogenic α-amino acids are referred to the comprehensive review by Ryan and coworkers¹⁶. Previously, the enzymatic synthesis of ATAAs has largely relied on threonine aldolases, which catalyze aldol reactions between glycine as the donor and aldehydes as acceptors to afford β-hydroxy-α-amino acids²⁶. Herein, we focus on the recent advances reported over the past five years, highlighting how enzymatic systems generate reactive intermediates, establish quaternary α-carbon centers, and expand the accessible chemical space of ATAAs. We also discuss the growing role of photoenzymatic catalysis and outline key opportunities and remaining challenges in enzyme discovery, engineering, and synthetic applications.

Reaction classes and catalytic principles

To complement the enzyme-family-based organization below, Table 1 provides a reaction-class overview of enzymatic strategies for ATAA formation. The reviewed systems can be grouped into Mannich-type C–C bond formation, formal [3 + 2] annulations and tandem γ-substitution/cyclization, strained-ring-forming intramolecular substitutions, hydroxymethyl transfer reaction, photoenzymatic radical α-functionalization, oxidative decarboxylation reaction, and radical C–N bond formation/nitrogen migration. This classification highlights how different enzyme families solve the same synthetic challenge, namely the construction of congested quaternary α-carbon centers, through distinct cofactors and reactive intermediates. Mechanistically, this classification distinguishes whether the key bond-forming species is a PLP-stabilized Cα quinonoid/carbanion, a β- or γ-elimination-derived vinylglycine or ketimine intermediate, an intramolecular substitution precursor, a hydroxymethyl transfer adduct, an enzyme-bound radical, or an Fe/O₂-derived radical intermediate.

Table 1.

Reaction classes and representative enzymatic strategies for ATAA formation

Reaction class	Representative enzymes	Key intermediate or step	Stereocontrol	Synthetic practicality
Mannich-type C–C bond formation	LolT27	PLP-bound Cα carbanion/quinonoid attacks an iminium electrophile	Y125 and W279 in the active-site pocket orient the iminium substrate and control anti/syn outcomes	Strong engineering precedent; LolTV4 kcat = 220 min−1 and Km = 33 mM for the model intermolecular reaction
PLP-dependent formal annulation and tandem γ-substitution/cyclization	SbzP/PseP36, FusB42, VtsC/KasA42, reprogrammed SphA28	SAM- or VAM-derived vinylglycine quinonoid/ketimine species undergo tandem C-C bond formation	Cofactor geometry and substrate positioning govern facial selectivity	SphAV4 has been evolved to >30-fold higher TTN, >10-fold higher kcat (24 s−1), and up to 96% yield with >99% ee
Strained-ring intramolecular substitution	ACCS50,51, GnmY53, Orf3054, PazB56	External aldimine or quinonoid intermediates undergo intramolecular displacement or cyclization	Substrate prefunctionalization and SN2-like displacement impose predictable stereochemical inversion or retention	Pairing of halogenases with PLP-dependent cyclases suggests ring-size engineering potential
Hydroxymethyl transfer	FmoH60, PbzB61	PLP-bound glycine quinonoid intermediate undergoing mTHF-dependent hydroxymethyl transfer	T68 and H78 help define the acceptor-binding pocket and stabilize a catalytically competent substrate-binding mode	General biocatalytic utility remains limited
Photoenzymatic radical α-functionalization	Engineered threonine aldolases64–66	PLP-quinonoid intermediates intercept externally generated carbon radicals	Radical capture occurs in a chiral active site; ternary complexes or local radical generation enhance selectivity	Preparative yields of up to 94% and excellent enantioselectivity (>99:1 e.r.) have been demonstrated in optimized systems.
Oxidative decarboxylation reaction	TqaM67	Metal/O2-mediated C2-H abstraction, peroxide addition, and decarboxylation	TqaM strictly recognizes the stereochemistry at the C2 position of the substrate	PaTqaM-Δ23 broadens bulky and aromatic substrate scope; for (R)-3-phenyllactic acid, kcat/Km = 2.0 × 10−3 min−1 μM−1
Radical C–N bond formation/nitrogen migration	Repurposed ACCO73	Nitrogen-centered radical generation, 1,5-HAT, and stereoselective radical rebound	Simultaneous Fe binding of radical and carboxylate partners constrains rebound geometry	Evolved ACCONim variants reached up to 2,260 TTN and enabled gram-scale synthesis of α-tri- and α-tetrasubstituted amino acids

These reaction classes also differ substantially in their current synthetic maturity. Some systems remain primarily biosynthetic or mechanistic discoveries with limited kinetic characterization, whereas others already show clear engineering tractability and preparative potential. For example, engineered LolT variants exhibit large improvements in turnover for intermolecular Mannich chemistry, including a reported k_cat of 220 min⁻¹ for LolT^V4 in the model reaction²⁷. The reprogrammed SphA platform displays more than a 30-fold improvement in total turnover number (TTN) after directed evolution, a greater than 10-fold increase in k_cat, and products in up to 96% yield and >99% ee²⁸.

PLP-dependent enzymes

Common catalytic principles of PLP-dependent enzymes for ATAA formation

PLP-dependent ATAA-forming enzymes, despite catalyzing chemically diverse transformations, are unified by a common cofactor-based strategy for generating and controlling reactive amino-acid intermediates. In canonical PLP catalysis, the catalytic cycle is initiated by transaldimination from a lysine-bound internal aldimine to a substrate-derived external aldimine²⁹. Subsequent enzyme-controlled proton abstraction, most commonly at Cα by an active-site base, generates a PLP-bound carbanion/quinonoid intermediate, in which the negative charge is stabilized by delocalization through the conjugated imine–pyridinium system of PLP (Fig. 1)³⁰. Depending on the enzyme family and substrate context, this activated intermediate can then be diverted into C–C bond formation, β- or γ-elimination/addition, decarboxylation, ketimine or iminium formation, intramolecular substitution, or radical interception. In this framework, Cα deprotonation provides quinonoid/carbanion nucleophiles for LolT-type Mannich chemistry and photoenzymatic radical α-functionalization, whereas β- or γ-elimination from amino-acid- or SAM-derived aldimines generates vinylglycine-like quinonoids or ketimines for SbzP/PseP/FusB/VtsC/KasA-type annulation pathways. In other cases, decarboxylation, folate-mediated C1 transfer, or substrate prefunctionalization redirects the PLP-bound intermediate toward SphA-type annulation, hydroxymethylation, or ACCS/PazB-type intramolecular substitution.

Fig. 1.

Representative reaction pathways of PLP-dependent enzymes.

Within this shared PLP manifold, ATAA formation requires precise control over both intermediate reactivity and, where applicable, the configuration of the newly formed quaternary α-carbon center. Stereocontrol is typically achieved through facially selective deprotonation or protonation of PLP-bound intermediates, defined positioning of electrophilic or radical reaction partners, and conformational restriction of bond-forming transition states by hydrogen-bonding, electrostatic, hydrophobic, or π-stacking interactions. Thus, LolT-type Mannich cyclases, SbzP/PseP/FusB/VtsC/KasA/SphA-type annulation enzymes, ACCS/PazB-type strained-ring-forming enzymes, serine hydroxymethyltransferase homologs, and photoenzymatic threonine aldolases can be viewed as distinct mechanistic adaptations of a conserved PLP-enabled strategy for generating and stereochemically trapping reactive amino-acid equivalents.

LolT

Loline alkaloids are insecticidal fungal pyrrolizidines that feature a highly strained tricyclic architecture (Fig. 2A)³¹. Hai and co-workers recently identified the PLP-dependent Mannich cyclase LolT, which catalyzes an intramolecular Mannich reaction to convert 1 into 2³². Subsequently, the PLP-dependent decarboxylase LolD catalyzes the stereoretentive decarboxylation of 2 to produce 3, furnishing the 1-aminopyrrolizidine scaffold of loline (4). This study established LolT as the first PLP-dependent enzyme to be structurally and biochemically characterized as catalyzing a Mannich-type C–C bond-forming reaction in natural product biosynthesis.

Fig. 2. Stereoselective intramolecular Mannich reaction catalyzed by the PLP-dependent Mannich cyclase LolT for construction of the pyrrolizidine core scaffold in loline (4) alkaloid biosynthesis.

A Biosynthesis of loline (4). B Active-site architecture of LolT in complex with the covalently bound PLP cofactor, which is linked to K236 through a Schiff base (PDB ID: 8DL5). C Proposed mechanism of LolT.

Structural studies revealed that LolT is a homodimeric fold type I PLP-dependent enzyme, closely related to the aspartate aminotransferase-like fold type I (AAT-I) family, with its active site located at the dimer interface (Fig. 2B)³². PLP is linked to K236 as an internal aldimine via a Schiff base and is positioned by an interaction network that includes π-stacking with Y125, a salt bridge between the pyridine nitrogen and D210, hydrogen bonding to H213, and extensive phosphate contacts with T100, D233, and T304 from monomer B.

Mechanistically, LolT follows the canonical PLP logic of external aldimine formation and Cα deprotonation, but diverts the resulting quinonoid intermediate toward Mannich cyclization, rather than the more typical aldol addition, elimination, or substitution pathways (Fig. 2C)³². Upon external aldimine formation, K236 acts as the catalytic base to stereospecifically deprotonate the l-configured α-amino acid, generating a PLP-stabilized Cα carbanion. This nucleophilic intermediate then attacks a tethered or externally generated cyclic iminium electrophile to forge the C–C bond. The stereochemical control appears to be dictated primarily by the protein environment, rather than by intrinsic transition-state preferences. Density functional theory (DFT) calculations for the native intramolecular cyclization showed only a small energy difference between the anti and syn transition states, whereas docking and mutagenesis studies indicated that Y125 and W279 constrain the productive approach trajectory of the iminium. In the native intramolecular reaction, LolT preferentially furnishes the anti-configured quaternary α-amino acid intermediate, which is subsequently decarboxylated by LolD with retention of this configuration.

Beyond its biosynthetic role, LolT proved to be a versatile biocatalyst for the stereoselective synthesis of 1-azabicyclic α-amino acids (Fig. 3A)³². Systematic variation of the cyclic iminium precursor showed that LolT can generate pyrrolizidine, indolizidine, quinolizidine, and related 1-azabicyclo[m.n.0]alkane (‘izidine’) frameworks, while maintaining high enantioselectivity and a consistent anti relationship between the vicinal tertiary and quaternary stereocenters. The catalytic efficiency decreased as the ring closure became more demanding, in agreement with structural analyses indicating that the expansion of the cyclization transition state is constrained by the size of the active-site pocket. Coupling LolT with LolD further enabled efficient chemoenzymatic access to 1-aminopyrrolizidine alkaloids and related amino-izidines, including the scalable synthesis of intermediates used in the total syntheses of alkaloids such as (+)-laburnamine (5) and (+)-absouline (6)³³.

Fig. 3. Biocatalytic synthesis of ATAAs catalyzed by the PLP-dependent Mannich cyclase LolT.

A Synthesis of 1-azabicyclic α-amino acids. B Two-component Mannich cyclization reactions. C Synthesis of α,β-diamino acids.

The same mechanistic platform was subsequently extended to two-component Mannich cyclization reactions (Fig. 3B)³². LolT was shown to generate PLP-stabilized carbanions from free diamino acids such as l-2,4-diaminobutanoic acid (Dab), l-Orn, and related donors, as evidenced by the rapid, stereoretentive H/D exchange at Cα in D₂O. These carbanionic intermediates could be trapped by cyclic imines formed in situ from the side-chain amino group and exogenous aldehydes, thereby enabling the direct synthesis of pyrrolidine- and piperidine-containing quaternary α-amino acids. In these intermolecular reactions, LolT predominantly furnished products arising from highly selective C–C bond formation, with the anti-diastereomer generally obtained under kinetic control, whereas the competing syn-isomer could form nonenzymatically under conditions that promoted epimerization.

Directed evolution additionally reprogrammed LolT into an intermolecular Mannich cyclase for the direct synthesis of α,β-diamino acids from free α-amino acids and enolizable cyclic imines (Fig. 3C)²⁷. In the engineered variants, mutations surrounding the active-site pocket, particularly near Y125, W279, and P427, improved the activity toward non-native amino acid donors while preserving high stereocontrol. The optimized variant, coupled with imine-generating enzymes, further expanded the acceptor scope to furnish a range of unprotected α,β-diamino acids. Directed evolution provided LolT^V4 with a reported k_cat of 220 min⁻¹ and K_m of 33 mM for the model intermolecular Mannich reaction. A mechanistic analysis suggested that K236 remains as the catalytic base responsible for amino-acid deprotonation, whereas Y125 and W279 cooperatively position the cyclic imine and govern the diastereoselectivity.

Collectively, LolT defines a distinct PLP-dependent platform for stereoselective C–C bond formation at the α-position of amino acids. In its biosynthetic role, it constructs the pyrrolizidine core of loline alkaloids through intramolecular Mannich cyclization followed by LolD-catalyzed decarboxylation. In engineered systems, the same catalytic logic can be redirected toward the synthesis of 1-azabicyclic α-amino acids, two-component Mannich cyclization products, and α,β-diamino acids. LolT therefore illustrates how a naturally evolved PLP-dependent Mannich cyclase can function as both a biosynthetic enzyme and a programmable biocatalytic scaffold for ATAA synthesis.

SbzP/PseP

The actinomycete-derived alkaloid altemicidin (7) features a structurally distinctive 6-azatetrahydroindane scaffold (Fig. 4A)^34,35. Abe and co-workers recently identified the PLP-dependent enzyme SbzP, together with its stable homolog PseP, as responsible for the stepwise [3 + 2] annulation between the pyridinium ring of β-nicotinamide adenine dinucleotide (β-NAD) and a β,γ-unsaturated quinonoid intermediate derived from S-adenosylmethionine (SAM) (Fig. 4A)^36,37. This transformation represents an unusual example of a PLP-dependent enzyme that directly exploits β-NAD as a carbon electrophile to generate the azaindane dinucleotide 8 in natural product biosynthesis.

Fig. 4. [3 + 2] Annulation between the pyridinium ring of β-NAD and a β,γ-unsaturated quinonoid intermediate derived from SAM, catalyzed by the PLP-dependent enzyme SbzP/PseP.

A Reactions catalyzed by SbzP/PseP. B Active-site architecture of PseP in complex with β-NAD and the covalently bound PLP cofactor, which is linked to K629 (PDB ID: 8K3H). C Proposed mechanism of SbzP/PseP.

Cryo-EM analysis of the β-NAD-bound PseP complex revealed a homodimeric type I PLP enzyme with the active site located at the dimer interface, while PseP contains an unusual N-terminal region that is important for the activity (Fig. 4B)³⁶. PLP forms the canonical internal aldimine with K629. The aromatic residues F413 and Y418 from monomer A, together with F457, R466 and Y699 from monomer B, form a π-stacking and hydrogen-bonding network that precisely positions the pyridinium ring of β-NAD at the dimer interface. The adenine ring of β-NAD is anchored through aromatic and electrostatic interactions, including π-stacking with F457 and a π–cation contact with R466, complemented by hydrogen bonding involving Y699. The O-2 of the diphosphate forms a hydrogen bond with R466. The nicotinamide riboside moiety extends into an interfacial pocket formed by α-helix 5 of monomer A and loop 6 of monomer B. Within this cavity, the nicotinamide ring forms a π-stacking interaction against F413, the ribose hydrophobically contacts Y418, and the 3′-hydroxyl group hydrogen bonds with D462. This binding mode is fundamentally distinct from the classical Rossmann-fold NAD recognition found in oxidoreductases^38,39. Importantly, α-helix 5, which harbors F413 and Y418, is part of the distinctive N-terminal extension of PseP, underscoring its importance in β-NAD recognition³⁶. Structure-guided mutagenesis, combined with thermal shift and surface plasmon resonance (SPR) analyses, demonstrated that F413, Y418, F457, R466 and Y699 are important for the reaction³⁶.

Mechanistically, the reaction initiates with transaldimination between PLP and SAM to generate an external aldimine, followed by Cα deprotonation to form a quinonoid intermediate (Fig. 4C)³⁶. Subsequent Cβ deprotonation promotes the γ-elimination of 5′-methylthioadenosine (MTA), affording a conjugated β,γ-unsaturated quinonoid species. This intermediate undergoes nucleophilic Cγ addition to the C4 position of the β-NAD pyridinium ring from the Re-face, forming the first C–C bond. Enzyme-facilitated isomerization, involving lysine-assisted reprotonation at Cβ from the Si-face, restores the electrophilicity at Cα and enables a second intramolecular C–C bond-forming event via the Si-face attack of the dihydropyridine C5 onto Cα, thereby completing the assembly of the azaindane scaffold. Final deprotonation and transaldimination release the cyclized product 8. A kinetic analysis supported a ping-pong Bi–Bi mechanism, consistent with the sequentially ordered SAM and β-NAD processing. Molecular dynamics simulations further corroborated a catalytically competent arrangement in which the Cγ of the quinonoid intermediate remains within reactive distance of the C4 atom of β-NAD, while DFT calculations disfavored a concerted pericyclic pathway and instead supported a stepwise mechanism involving two discrete C–C bond-forming events.

Collectively, SbzP/PseP illustrates how PLP-dependent enzymes can promote nucleophilic additions at both the Cγ and Cα positions while repurposing β-NAD as a biosynthetic building block. This work broadens the known catalytic repertoire of PLP-dependent enzymes and creates new opportunities by engineering PLP-dependent enzymes to access β-NAD-derived and other pyridinium cofactor-derived scaffolds for drug discovery and development.

FusB, VtsC, KasA

Cycloleucine (9), a cyclic ATAA bearing a quaternary α-center, is a notable motif in both bioactive small molecules and natural products (Fig. 5A)^40,41. Beyond cycloleucine itself, related cyclopentane-constrained amino-acid frameworks occur in structurally diverse metabolites such as fusarilin A (10), stephadiamine (11), altemicidin (7), and tagetitoxin (12), underscoring the broader biosynthetic relevance of this scaffold family⁴². Recent biosynthetic studies by Tang and co-workers have revealed that cyclopentane motifs can be generated by a distinct class of PLP-dependent enzymes that couple latent SAM-derived electrophiles with transiently formed imine or enamine nucleophiles, thereby extending classical γ-substitution chemistry into tandem annulation processes involving carbon–carbon bond formation⁴².

Fig. 5. Cycloleucine (9) formation catalyzed by PLP-dependent enzymes in natural product biosynthesis.

A Cycloleucine-containing natural products. B Reactions catalyzed by FusA and FusB. C Proposed mechanism of FusB.

In fusarilin A (10) biosynthesis, the substituted cycloleucine core is assembled through the combined actions of FusA and FusB (Fig. 5B)⁴². FusA, an Fe(II)/α-ketoglutarate (αKG)-dependent enzyme, converts l-phenylalanine (13) into 2-phenylethylimine (14), which is proposed to equilibrate with the corresponding enamine and serve as the nucleophilic partner for the subsequent PLP-dependent transformation. The PLP-dependent enzyme FusB engages SAM via the canonical external aldimine and processes it into a reactive γ-substitution intermediate, consistent with PLP-dependent elimination chemistry. An in vitro analysis demonstrated that FusB converts SAM into l-vinylglycine, supporting the involvement of a SAM-derived electrophilic intermediate in this reaction. The resulting SAM-derived intermediate is then intercepted by the 2-phenylethylimine nucleophile to form the first carbon-carbon bond, followed by a second bond-forming step that closes the cyclopentane ring. Overall, FusB is proposed to catalyze a formal [3 + 2] annulation through tandem γ-substitution and Cα-based intramolecular aldol-type cyclization, generating the substituted cycloleucine precursor.

The proposed mechanism for FusB is supported by isotope-labeling experiments indicating proton abstraction at both Cα and Cβ of the SAM-derived fragment, consistent with a canonical γ-elimination-addition pathway involving a vinylglycine ketimine-like intermediate (Fig. 5C)⁴². The precise trajectory of the initial bond-forming step remains unresolved and may proceed through either stepwise elimination-addition or a closely related γ-substitution pathway. In either case, FusB does not terminate after the γ-substitution, but instead channels the intermediate into a second cyclization step to form the cycloleucine scaffold.

A related but mechanistically more elaborate platform is represented by VtsC and KasA, which convert l-serine (16) and SAM into 2-aminocycloleucine (17) (Fig. 6A)⁴². These enzymes are didomain PLP proteins composed of an N-terminal serine dehydratase-like domain (VtsC/KasA-β) and a C-terminal FusB-like cycloleucine synthase domain (VtsC/KasA-γ). Serine dehydratase originally catalyzes the β-dehydration of l-serine (16) to generate dehydroalanine (18). As for VtsC and KasA, the VtsC/KasA-β domain catalyzes a similar reaction, whereas the C-terminal domain uses SAM as the electrophilic partner in the annulation sequence. The product of the VtsC/KasA pathway was identified as 2-aminocycloleucine (17), revealing that the same general biosynthetic logic can be adapted to produce a distinct cycloleucine derivative. Mechanistically, the early steps of the VtsC/KasA-γ domain are proposed to resemble those of FusB, generating a SAM-bound enamine or equivalent γ-substitution intermediate (Fig. 6B)⁴². The reaction with dehydroalanine then forms the first coupled adduct, which can tautomerize through ketimine and quinonoid states to a cyclization-competent intermediate. From this species, the PLP-stabilized Cα carbanion attacks the imine in a 5-exo-trig fashion, giving the aldimine product through an intramolecular Mannich-type cyclization, rather than the aldol-type closure proposed for FusB. The resulting aldimine then undergoes decarboxylation and reprotonation to release 17 and restore the internal aldimine state of the enzyme. In this regard, VtsC and KasA expand the FusB paradigm by integrating γ-substitution, Cα-based Mannich cyclization, and α-decarboxylation into a single catalytic sequence. Isotope labeling experiments further clarified the origin of the carbon framework in the VtsC/KasA products. Whereas the FusB pathway retains the α-carbon of SAM as the quaternary center in the cycloleucine product, VtsC and KasA first annulate SAM with dehydroalanine and then selectively remove the SAM-derived carboxylate during decarboxylation. As a consequence, the quaternary carbon in 2-aminocycloleucine (17) originates from the α-carbon of l-serine, rather than from SAM. This contrast highlights how related PLP-dependent annulation platforms can generate distinct substitution patterns and carbon-lineage outcomes by varying both the identity of the nucleophilic partner and the timing of decarboxylation relative to ring formation.

Fig. 6. 2-Aminocycloleucine (17) formation catalyzed by the PLP-dependent enzymes VtsC/KasA.

A Reactions catalyzed by VtsC/KasA. B Proposed mechanism of the VtsC/KasA-γ domain.

Collectively, FusB, VtsC, and KasA define a distinct branch of PLP-dependent carbon–carbon bond-forming enzymes that construct cycloleucine-type ATAAs from simple amino-acid-derived precursors. In biosynthetic terms, these enzymes show how SAM can be repurposed as a latent carbon electrophile and paired with transient imine or enamine nucleophiles to access cyclopentane amino-acid scaffolds through tandem annulation chemistry. More broadly, the occurrence of cycloleucine-related motifs in metabolites suggests that analogous logic may underlie a wider range of cycloleucine-derived natural product scaffolds than is currently recognized.

SphA

A conceptually distinct extension of vinylglycine quinonoid (VGQ, 19) chemistry was realized through the mechanistic reprogramming of SphA, a PLP-dependent enzyme that natively catalyzes a decarboxylative Claisen condensation in sphingofungin biosynthesis⁴³. In this re-engineered system, SphA was repurposed to mediate a decarboxylative formal [3 + 2] annulation between vinylaminomalonate (VAM, 20) and electron-deficient alkenes, thereby providing direct access to densely functionalized carbocyclic ATAAs bearing up to three contiguous stereogenic centers (Fig. 7A)²⁸. Unlike SbzP/PseP, in which VGQ is generated through the γ-elimination of SAM³⁶, the present system accesses the same reactive intermediate through the α-decarboxylation of VAM, an achiral precursor whose irreversible decarboxylation provides a strong thermodynamic driving force for VGQ formation. Screening of activated alkenes identified productive annulation with strongly electron-withdrawing partners, particularly trifluoromethyl enones and nitroalkenes, and iterative saturation mutagenesis of SphA furnished a quadruple variant SphA^V4, which exhibited a more than 30-fold increase in TTN and a greater than 10-fold increase in k_cat (24 s⁻¹) relative to the wild-type enzyme, affording annulation products in yields of up to 96% with excellent enantioselectivity. These results establish SphA as an abiotic PLP platform for decarboxylative annulation, rather than a natural VGQ-forming cyclase.

Fig. 7. Formal [3 + 2] annulations catalyzed by the repurposed PLP-dependent enzyme SphA for the construction of cyclopentane-based ATAAs.

A Reactions catalyzed by SphA. B Active-site architecture of SphA in complex with the VGQ intermediate (PDB ID: 9DRI). C Proposed mechanism of repurposed SphA for decarboxylative [3 + 2] annulation.

Structural analysis clarified how this non-native chemistry is accommodated within the SphA scaffold. Upon the addition of VAM, SphA generated a long-lived species with a characteristic absorption band at approximately 530 nm, and crystallographic analysis of the SphA complex directly captured the corresponding PLP-bound VGQ intermediate, confirming that VAM had undergone enzymatic decarboxylation in the active site (Fig. 7B)²⁸. The VGQ-bound structure revealed an s-cis conformation stabilized in an active-site environment resembling that of other 8-amino-7-oxononanoate synthase (AONS) family enzymes. A hydrogen-bonding network involving H156, N135, S158, and D241 supports VGQ formation, while the conserved H156 residue also participates in a water-mediated interaction with N303 from the adjacent monomer.

Mechanistically, the reaction is proposed to proceed in a stepwise manner, rather than through a concerted cycloaddition (Fig. 7C)²⁸. DFT calculations supported the initial vinylogous Michael-type addition of VGQ to the activated alkene to form a Cγ-alkylated enolate, followed by PLP-mediated isomerization to generate a Cα-electrophilic iminium species. Intramolecular nucleophilic addition of the enolate to this iminium then furnishes the cyclopentane ring. Computational and mutagenesis analyses further suggested that the enzyme discriminates between the endo and exo pathways through a combination of intrinsic transition-state energetics and differential active-site stabilization. For trifluoromethyl enones, the favored endo transition state is stabilized by hydrogen-bonding interactions with H156 and N303, whereas for nitroalkenes, both pathways can be accommodated more similarly, accounting for the diminished diastereoselectivity observed with that substrate class. In the absence of an electrophilic alkene partner, the VGQ intermediate instead undergoes stereoselective Cα-protonation to give predominantly l-vinylglycine, consistent with the enzymatic control of both VGQ formation and its subsequent fate. Overall, this work demonstrates that the annulating propensity of VGQ is not restricted to naturally evolved SbzP/PseP-like systems, but can also be realized in a mechanistically reprogrammed PLP scaffold through decarboxylative access to the same rare intermediate.

ACCS, GnmY, Orf30, PazB

Strained small-ring amino acids, including cyclopropane- and cyclobutane-containing motifs, constitute valuable ATAA building blocks in natural product biosynthesis^44–46. The incorporation of these units can impose conformational constraints and introduce steric and electronic properties that are difficult to achieve with linear amino-acid scaffolds^47–49. In several pathways, PLP-dependent enzymes play a central role by converting amino-acid or SAM-derived substrates into reactive external-aldimine or quinonoid intermediates that undergo intramolecular C–C bond formation, thereby enabling ring closure under mild conditions.

A canonical illustration of this logic is the formation of 1-aminocyclopropane-1-carboxylate (ACC, 23) from SAM by ACC synthase (ACCS) (Fig. 8A)^50,51. In plants, ACC serves as the direct precursor to ethylene, with its conversion catalyzed by the non-heme iron enzyme 1-aminocyclopropane-1-carboxylic acid oxidase (ACCO)⁵². In guangnanmycin A (25) biosynthesis, a bacterial ACCS homolog, GnmY, was shown to generate free ACC as a dedicated building block⁵³. In vitro reconstitution demonstrated that recombinant GnmY converts SAM into ACC. Together, these results establish GnmY as a bacterial ACCS that supplies a strained cyclopropane amino-acid precursor for downstream pathway steps.

Fig. 8. Formation of cyclopropane- and cyclobutane-containing moieties catalyzed by PLP-dependent enzymes in natural product biosynthesis.

A ACC (23) formation catalyzed by ACCS/GnmY. B Formation of (1S,2S)-MeACC (26) catalyzed by Orf30. C Formation of pazamine (28) and 32 catalyzed by PazB. D Proposed mechanism of PazB.

In nature, α-tertiary cyclopropane amino acids can also arise through the upstream tailoring of SAM before the PLP-dependent cyclization step. In the biosynthesis of 1-amino-2-methylcyclopropane-1-carboxylate (MeACC, 26), a two-enzyme module couples cobalamin-dependent radical SAM C-methylation with PLP-dependent ring construction (Fig. 8B)^54,55. The radical SAM methylase Orf29 first methylates SAM to generate 4″R-Me-SAM (27), which is then selectively processed by the PLP-dependent enzyme Orf30 to afford (1S,2S)-26 with concomitant MTA release. This pathway illustrates a modular strategy in which a radical SAM tailoring step generates a PLP-reactive intermediate, enabling stereocontrolled access to α-tertiary cyclopropane building blocks beyond the canonical ACC.

A conceptually broader biosynthetic strategy emerges when a dedicated leaving group is installed prior to PLP catalysis, thereby enabling intramolecular substitution chemistry that can be adapted to different ring sizes. In pazamine (28) biosynthesis, the αKG-dependent halogenase PazA catalyzes the regio- and stereoselective C4 chlorination of l-lysine (29) to generate 4R-Cl-l-lysine (30), which is subsequently cyclized by the PLP-dependent enzyme PazB (Fig. 8C)⁵⁶. Current mechanistic proposals suggest that PazB forms an external aldimine with the chlorinated amino acid, generates a quinonoid intermediate upon Cα deprotonation, and then promotes an intramolecular S_N2-like displacement at C4 with stereo-inversion to furnish the cyclopropane product⁵⁶. Computational analyses further support a stepwise mechanism in which the quinonoid formation precedes a rate-determining cyclization event, and indicate that active-site water molecules may facilitate chloride departure by stabilizing the developing leaving group and promoting productive near-attack conformations (Fig. 8D)⁵⁷. Notably, this cryptic halogenation/PLP-catalyzed cyclization logic is not limited to cyclopropane formation. By pairing PazB with the alternative lysine halogenase HalB, which installs chlorine at C5 to produce 5R-Cl-l-lysine (31), the same platform was extended to the biosynthesis of the cyclobutane-containing amino acid 32 in vivo, highlighting the potential for ring-size diversification through leaving-group engineering⁵⁶. A related strategy operates in coronamic acid biosynthesis, where the αKG-dependent halogenase CmaB chlorinates the carrier protein-tethered l-allo-isoleucine, and the downstream metalloenzyme CmaC converts the resulting γ-chloro intermediate into the cyclopropane product through the intramolecular displacement of chloride⁵⁸. These studies illustrate how halogenation can be coupled to PLP- or metal-dependent cyclization chemistry to expand the biosynthetic repertoire of strained carbocyclic amino acids.

Collectively, the ACCS/GnmY, Orf29/Orf30, and PazA/PazB systems exemplify how PLP enzymes, either operating directly on SAM or acting downstream of tailoring steps, enable the efficient biosynthesis of strained α-tertiary amino-acid building blocks through intramolecular C–C bond formation.

FmoH, PbzB

PLP-dependent serine hydroxymethyltransferases classically catalyze the reversible interconversion of glycine and l-serine (16), using 5,10-methylenetetrahydrofolate (mTHF) as the one-carbon carrier⁵⁹. In natural product biosynthesis, homologs from this enzyme family have been proposed to accommodate alternative amino-acid substrates, enabling, for instance, the formation of α-methyl-l-serine (33) from d-alanine (34) in JBIR-34 (35) biosynthesis⁶⁰ and the formation of 2-hydroxymethylserine (36) from l-serine (16) in benzobactin B (37) biosynthesis⁶¹.

In JBIR-34 (35) biosynthesis, FmoH, annotated as a glycine/serine hydroxymethyltransferase homolog, was implicated in supplying 33 for downstream assembly (Fig. 9A)⁶⁰. Biochemical characterization indicated that tetrahydrofolate is required and that PLP contributed to the activity. Notably, recombinant FmoH predominantly catalyzed the reverse reaction in vitro, converting 33 to d-alanine 34.

Fig. 9. Serine hydroxymethylation reactions catalyzed by PLP-dependent serine hydroxymethyltransferases in natural product biosynthesis.

A FmoH in JBIR-34 (35) biosynthesis. B PbzB in benzobactin B (37) biosynthesis.

A conceptually related role was proposed for PbzB in benzoxazolinate-containing metabolite biosynthesis, where it was assigned to generate 36 as a dedicated nonribosomal peptide building block (Fig. 9B)⁶¹. To probe substrate accommodation, Bode and coworkers created the pocket-restricting T68Y/H78Y double variant designed to remodel the acceptor-binding environment. This variant abolished the detectable formation of 36 in vitro, implicating T68 and H78 in shaping a productive acceptor-binding geometry and maintaining a catalytically competent active-site configuration.

Although these serine hydroxymethyltransferase homologs are less developed as synthetic biocatalysts than LolT, SphA, or threonine aldolases, they broaden the mechanistic landscape of PLP-dependent ATAA formation by demonstrating that one-carbon transfer chemistry can also generate α-substituted amino-acid building blocks. Their current value lies mainly in revealing how modest remodeling of the serine hydroxymethyltransferase active site can accommodate noncanonical amino-acid substrates. This family could be further engineered into a more general platform for the synthesis of hydroxymethylated and alkylated ATAAs.

Threonine aldolase

Threonine aldolases (TAs), classically used for PLP-dependent aldol chemistry at the α-position of amino acids^62,63, have recently been repurposed as radical biocatalysts for the stereoselective synthesis of ATAAs under visible light irradiation^64–66. In these systems, the native PLP-dependent formation of a persistent quinonoid intermediate is retained, but its reactivity is redirected from two-electron aldol chemistry toward intermolecular radical C–C bond formation, thereby enabling the non-natural α-functionalization of unprotected amino acids.

In one implementation, Yang and coworkers used engineered threonine aldolases combined with an external photoredox catalyst and a stoichiometric oxidant to enable asymmetric sp³–sp³ oxidative coupling between organoboron reagents and amino acids (Fig. 10A)⁶⁴. In this platform, the enzyme forms the quinonoid intermediate from glycine or α-branched amino acids, while visible-light photoredox catalysis generates an external carbon-centered radical from the boron reagent. Radical addition to the α-carbon of the enzyme-bound quinonoid is followed by oxidation of the resulting nitrogen-centered radical to furnish a new external aldimine, which then undergoes transaldimination to release the coupled amino acid product. Directed evolution of Thermotoga maritima threonine aldolase afforded variants such as TmPLP^α₁ and TmPLP^α₂, enabling efficient access to α-trisubstituted and α-tetrasubstituted noncanonical amino acids, including products bearing contiguous stereocenters. Mechanistic studies supported benzyl radical intermediacy, rapid quinonoid formation, and an irreversible radical addition step that governs both regio- and enantioselectivity.

Fig. 10. Representative photoenzymatic strategies for the synthesis of ATAAs.

A Asymmetric sp³–sp³ oxidative coupling. B α-Alkylation of amino acids using Katritzky salts.

A complementary strategy was developed independently for the α-alkylation of amino acids, using Katritzky pyridinium salts as radical precursors (Fig. 10B)^65,66. In the system devised by the Hyster group, threonine aldolases were paired with Rose Bengal, and mechanistic studies indicated that productive radical generation is facilitated by a ternary interaction among the photocatalyst, the pyridinium salt, and the protein scaffold⁶⁵. This arrangement is proposed to localize the radical formation near the active site, allowing the enzyme-bound quinonoid to intercept the radical with high stereocontrol. Under these conditions, glycine and alanine could be converted directly into ATAAs as single major enantiomers, and the method tolerated a range of benzylic pyridinium salts, with more limited reactivity for non-benzylic substrates.

By contrast, the Yang group showed that TA-catalyzed α-alkylation with Katritzky salts can proceed without an added photoredox catalyst⁶⁶. In this case, the reaction was found to operate through an unconventional photoinduced radical initiation mode involving a PLP-derived aldimine formed from free PLP and the amino acid substrate. TEMPO trapping, EPR analysis, and luminescence studies indicated that this aldimine species is markedly more competent for radical generation under blue light irradiation than free PLP alone, after which the externally generated radical diffuses into the enzyme active site and adds to the quinonoid intermediate. This redox-neutral platform enabled the highly enantioselective synthesis of both α-trisubstituted and α-tetrasubstituted amino acids from glycine and α-branched amino acids, and further highlighted the unusual ability of engineered TAs to process both enantiomers of certain α-branched substrates in an enantioconvergent manner.

Collectively, these studies establish threonine aldolases as versatile PLP scaffolds for new-to-nature photobiocatalytic α-functionalization. Although the radical initiation modes differ, involving external photoredox/oxidant-mediated oxidative coupling with organoborons, Rose-Bengal-assisted alkylation with pyridinium salts, and PLP-aldimine-enabled photocatalyst-free radical initiation, all three approaches rely on the same central design principle: the interception of an enzyme-bound quinonoid intermediate by an externally generated carbon radical to forge the α-C–C bond with high stereocontrol. These advances substantially expand the synthetic scope of threonine aldolases beyond their native aldol chemistry and provide practical biocatalytic entry to ATAAs that are difficult to access by conventional asymmetric synthesis.

Non-heme iron enzymes

TqaM

A distinct non-heme iron-dependent strategy for ATAA biosynthesis is exemplified by TqaM, a non-heme iron enzyme that catalyzes the oxidative decarboxylation of α-hydroxy-β-amino acid precursors in the fungal biosynthesis of 2-aminoisobutyric acid (AIB, 38) and l-isovaline (39) (Fig. 11A)⁶⁷. Notably, the AIB (38) moiety is found in peptaibol antibiotics such as alamethicin, in which it promotes the helical conformation, and is also present in the alkaloid tryptoquialanine^68,69. In the pathways of 38 and 39, the αKG-dependent oxygenase TqaL first generates an aziridine intermediate, which is subsequently opened by the haloalkanoic acid dehalogenase-like enzyme TqaF to furnish a β-amino acid intermediate^70–72. TqaM then converts this intermediate into the corresponding ATAA through oxidative decarboxylation⁶⁷. In contrast to PLP-dependent annulation pathways or radical-mediated C–C bond-forming strategies, this route establishes the ATAA scaffold through carbon excision, rather than carbon–carbon bond construction.

Fig. 11. Oxidative decarboxylation by the non-heme iron oxygenase TqaM to produce ATAAs in natural product biosynthesis.

A The biosynthetic pathways of AIB (38) and l-isovaline (39). B Proposed mechanism of TqaM. C Active-site architecture of TqaM in complex with the substrate AHMBA (44) (PDB ID: 9M09). D Selected products of PaTqaM-Δ23.

Mechanistic studies showed that TqaM is highly sensitive to the substrate stereochemistry at C2, but more tolerant at C3⁶⁷. Only the C2R-configured native α-hydroxy-β-amino acid substrate was recognized, and deuterium-labeling experiments supported C2-H abstraction as a partially rate-limiting step. Stoichiometric analysis further indicated the consumption of one equivalent of O₂ per substrate, consistent with an oxidative decarboxylation pathway involving C2-H abstraction, carbonyl intermediate formation, peroxide addition, and decarboxylation to release the product (Fig. 11B)⁶⁷.

Further structural analysis provided a basis for this reactivity⁶⁷. TqaM adopts an α/β-sandwich fold related to class II fuculose-1-phosphate aldolases and functions as a homotetramer, with the active site located at the monomer interface (Fig. 11C). The metal center is coordinated by three His residues. In the substrate complex structure, 3-amino-2R-hydroxy-3-methylbutanoic acid (AHMBA, 44) binds through the bidentate coordination of the C1-carboxylate and C2-hydroxyl groups to the metal ion. The C1-carboxylate is stabilized by hydrogen-bonding interactions with H74, N123, and the backbone amide of G73, whereas the C2-hydroxyl group forms a hydrogen bond with Q166. The C3-amino group interacts with S72 and is additionally fixed through water-mediated hydrogen-bonding networks involving W271 and E275, as well as Q166 and D167. In this binding mode, the C2-hydrogen atom is oriented on the same side as the metal-bound water site, consistent with a mechanism initiated by hydrogen abstraction at C2⁶⁷. A mutational analysis further showed that the residues involved in substrate binding and organization are important for catalysis, as most alanine substitutions caused a marked loss of activity.

A further notable feature is the role of the N-terminal loop in the substrate selectivity⁶⁷. Truncation of this region had little effect on the native substrate, but markedly broadened the acceptance of non-native aromatic analogs. In particular, PaTqaM-Δ23 showed enhanced activity toward bulky and aromatic α-hydroxy acids. For (R)-3-phenyllactic acid, the catalytic efficiency (k_cat/K_m) was 2.0 × 10⁻³ min^–1 μM^–1, supporting that this loop acts as a lid-like element that modulates the substrate promiscuity. These features enabled the use of TqaM as a biocatalyst for non-proteinogenic amino-acid synthesis, including diverse ATAAs. The truncated PaTqaM-Δ23 variant accepted a range of 3-monosubstituted and 3,3-disubstituted β-amino acids to furnish the corresponding α-amino acids and ATAAs, including products bearing propyl, isopropyl, cyclohexyl, aryl, and heteroaryl substituents (Fig. 11D)⁶⁷. Collectively, TqaM defines a mechanistically distinct non-heme iron biocatalytic platform for ATAA biosynthesis, in which the oxidative decarboxylation of α-hydroxy-β-amino acids furnishes the corresponding ATAA product.

ACCO

A distinct radical strategy for the synthesis of ATAAs was realized by repurposing the plant enzyme ACCO for asymmetric 1,3-nitrogen migration⁷³. In this transformation, azanyl esters derived from carboxylic acids are converted into N-protected noncanonical amino acids through the intramolecular migration of the nitrogen substituent, providing direct access to α-trisubstituted and α-tetrasubstituted amino-acid products (Fig. 12A).

Fig. 12. Repurposing of the non-heme iron enzyme ACCO for asymmetric 1,3-nitrogen migration.

A Reactions catalyzed by the repurposed ACCO. B Proposed mechanism for asymmetric 1,3-nitrogen migration catalyzed by the repurposed ACCO. C Selected products of the repurposed ACCO.

Mechanistically, the reaction is proposed to proceed through a radical relay pathway, enabled by the open coordination environment of the non-heme Fe center (Fig. 12B)⁷³. Fe-mediated activation of the azanyl ester is proposed to initiate N–O bond cleavage, generating an iron–nitrenoid or nitrogen-centered radical intermediate together with a coordinated carboxylate. This intermediate undergoes intramolecular 1,5-hydrogen atom transfer to form a carboxylate α-radical. Subsequent radical rebound to the Fe-bound nitrogen group forges the C–N bond and furnishes the migration product, with product release regenerating the ferrous enzyme. This design is thought to benefit from the ability of the non-heme Fe center to bind both the nitrogen-centered radical and the substrate carboxylate, thereby facilitating the otherwise challenging intramolecular 1,5-HAT manifold. The stereochemical outcome is proposed to arise from the simultaneous coordination of the carboxylate and nitrogen-transfer fragment to the Fe center, which confines the geometry of the 1,5-HAT and subsequent radical rebound steps within a chiral active-site environment.

This platform was then developed through directed evolution to address progressively more demanding substrate classes. Initial screening identified ACCO as a promising scaffold, already giving the model migration products in modest yield but with high enantioselectivity. Subsequent evolution delivered ACCO_Nim variants for the enantioselective amination of prochiral secondary C(sp³)-H bonds, the enantioconvergent amination of tertiary C(sp³)-H bonds from racemic substrates, and ultimately the asymmetric construction of minimally differentiated methyl-ethyl stereocenters. The evolved enzymes showed marked improvements in activity and turnover relative to wild-type ACCO and enabled the synthesis of a broad range of α-trisubstituted and α-tetrasubstituted amino acids, including representative products on preparative scales (Fig. 12C)⁷³.

The ACCO study is particularly instructive from a synthetic-practicality standpoint because the native enzyme was converted into a family of ACCO_Nim variants through directed evolution, rather than being used only as a biosynthetic curiosity. This trajectory shows that non-heme iron scaffolds with open coordination sites can be tuned for non-native radical relays while preserving high enantiocontrol. At the same time, the requirement for azanyl-ester substrates and the substrate-dependent evolution of different variants indicate that further engineering will be needed before nitrogen migration becomes a broadly general ATAA synthesis platform.

Summary and outlook

Recent advances have established enzymatic catalysis as a powerful platform for the synthesis of ATAAs. As discussed in this review, PLP-dependent enzymes now support a remarkably broad range of ATAA-forming transformations, including Mannich-type C–C bond formation, γ-substitution followed by annulation, strained-ring construction, and photoenzymatic radical α-functionalization. In parallel, non-heme iron enzymes provide complementary solutions through oxidative decarboxylation and radical nitrogen migration. Together, these studies show that enzymatic ATAA synthesis now encompasses multiple distinct catalytic strategies for quaternary α-center formation.

At the same time, many currently known enzymes retain narrow substrate scopes, and the mechanistic bases of reactivity and stereocontrol are still incompletely understood in several recently discovered systems. Future progress will therefore depend on the combined development of enzyme discovery, structural and mechanistic analysis, and directed evolution^74,75. In particular, photoenzymatic catalysis offers an important opportunity to expand the accessible reaction space beyond native enzyme function, while retaining the stereochemical advantages of biocatalysis^76–78.

The adjacent field of enzymatic α-tertiary amine synthesis also offers important insights for ATAA biocatalysis. Engineered heme proteins have established carbene and nitrene transfer as highly evolvable platforms for stereoselective C–H functionalization⁷⁹. In carbene-transfer chemistry, engineered cytochrome c variants have enabled enantioselective N–H insertion of acceptor–acceptor carbenes to access chiral α-trifluoromethyl amino ester products in >99% yield, 95:5 e.r., and up to 752 total turnovers⁸⁰. Although these transformations do not directly furnish α-tertiary amino acids, they demonstrate that protein active sites can control metal–carbenoid intermediates in the synthesis of densely functionalized chiral amine motifs. More directly related to α-tertiary amine synthesis, an engineered P411 nitrene transferase was recently shown to aminate tertiary C–H bonds to afford chiral α-tertiary primary amines with up to 2300 total turnovers and >99% ee⁸¹. Although these products lack the carboxylate functionality required for ATAAs, the underlying design principles, including control of high-energy metal–nitrenoid or metal–carbenoid intermediates, active-site shielding from competing pathways, and directed evolution for sterically demanding C–H functionalization, are directly relevant to future enzymatic strategies for the late-stage construction or diversification of ATAA derivatives.

Looking ahead, a central goal for the field will be the transition from individual enzymatic case studies to more general and programmable platforms for ATAA synthesis. Continued discovery of novel biosynthetic enzymes, together with the engineering of known PLP-dependent and non-heme iron scaffolds, would further broaden the range of accessible ATAA architectures. As these efforts advance, enzymatic catalysis is likely to become an increasingly valuable complement to chemical synthesis for the preparation of structurally complex and stereochemically demanding ATAAs.

Author contributions

H.W.: writing—original draft. I.A.: writing—review & editing, supervision.

Peer review

Peer review information

Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work.

Funding

I.A. discloses support for the research of this work from a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (JSPS KAKENHI Grant Number JP26K21766). H.W. discloses support for the research of this work from the JSPS Research Fellowship for Young Scientists. The authors declare no additional relevant funding.

Data availability

No new data were generated or analyzed in this study. Data sharing is not applicable to this article, as all information discussed in this review is derived from previously published studies and publicly available sources.

Competing interests

The authors declare no competing interests.