Introduction

Amino acids serve as fundamental building blocks in nature, constituting core structural motifs of proteins and numerous other biologically important molecules (Fig. 1a)1,2,3,4,5. Beyond their natural functions, amino acids and their derivatives—particularly non-proteinogenic variants—have gained increasing importance in pharmaceutical development, material science, and asymmetric catalysis6. The recent surge in peptide-based therapeutics, accelerated by advances in screening technologies like mRNA display, has intensified the demand for structurally diverse amino acids with precise stereochemical control7. This technology enables the incorporation of non-natural amino acids into vast peptide libraries, significantly expanding the structural and functional diversity of potential drug candidates8. Consequently, developing efficient, scalable, and economical methods for synthesizing enantiomerically pure amino acids from readily available precursors has become increasingly urgent for advancing peptide drug discovery and development9. Traditional approaches to synthesizing chiral α-amino acids rely predominantly on polar/heterolytic processes10. These include asymmetric hydrogenations of dehydroamino acids11 or α-imino esters12,13, electrophilic aminations of enolates14, alkylations of glycine derivatives via phase-transfer catalysis15 and through Pd/Cu synergistic catalytic systems16,17, and nucleophilic additions to α-imino esters18. More recent advances include asymmetric Ugi-type multicomponent reactions, which are highly valuable but involve odorous and toxic isocyanides19. Other developments involve stereocontrolled 1,3-nitrogen migration via metal-catalyzed nitrene insertion that relies on activated nitrogen sources20,21.

Fig. 1: Bioactive chiral amino acid derivatives and synthetic approaches.
Fig. 1: Bioactive chiral amino acid derivatives and synthetic approaches.
Full size image

a Selected bioactive compounds and drugs containing α-amino acids. b Nickel-catalyzed enantioconvergent cross-coupling of racemic α-haloglycine derivatives with alkylzinc reagents. c Cobalt-catalyzed asymmetric reductive alkylation of dehydroglycine derivatives with alkyl halides. d Racemic synthesis of α-amino carbonyls with dihydropyridines. e Diastereoselective addition of carbamoyl anions to chiral imines. f Catalytic asymmetric synthesis of chiral α-amino amides from formamides, aldehydes and amines. TBSOTf tert-butyldimethylsilyltrifluoromethanesulphonate, LDA lithium diisopropylamide, CPA chiral phosphoric acid, HAT hydrogen atom transfer.

In recent years, radical-based methods22,23 have emerged as promising alternatives for asymmetric amino acid synthesis. Notably, nickel-catalyzed enantioconvergent cross-coupling of racemic α-haloglycine derivatives with alkylzinc reagents (Fig. 1b), and cobalt-catalyzed asymmetric reductive alkylation of dehydroglycine derivatives with alkyl halides (Fig. 1c) have demonstrated the potential of radical chemistry to overcome limitations of traditional approaches24,25,26, albeit that pre-prepared substrates with specific activation groups are still required. Among various radical species, carbamoyl radicals are particularly attractive for amino acid synthesis due to their nucleophilic character and potential for direct amide bond formation27. However, prevalent strategies for harnessing carbamoyl radicals typically rely on 4-substituted-1,4-dihydropyridines (DHPs) as precursors28,29,30, requiring multi-step synthesis and offering limited substrate scope, albeit that the generation of carbamoyl radicals from formamides via hydrogen atom transfer (HAT) is known for racemic Minisci-type carbamoylations31,32 and hydrocarbamoylations33. The most recent approach employs visible light and Lewis acid activation of DHPs to generate carbamoyl radicals for three-component reactions with in situ-formed imines or iminiums34, yet asymmetric variants remain undeveloped (Fig. 1d). Alternative strategies using simple formamides as carbamoyl anion precursors require strong bases (e.g., LDA) and chiral auxiliaries35, limiting practical applications (Fig. 1e). Thus, a general, catalytic asymmetric strategy to harness carbamoyl radicals from simple formamides for the synthesis of amino acid derivatives remains an unmet challenge.

Herein, we report a dual catalytic system that integrates a HAT-mediated carbamoyl radical generation with a stereocontrolled ionic and H-bonding activated asymmetric addition between carbamoyl radical and an in situ formed imine. This strategy enables rapid and modular assembly of α-amino amides from readily available aldehydes, amines, and formamides (Fig. 1f). We sort out the cumulative challenges that compatibility among the individual species in the reaction mixture (catalysts, substrates, intermediates), side reactions, catalyst deactivation, realization of broad substrate scope. The developed protocol affords a wide range of α-amino amide derivatives (>80 examples) with up to >99% yield and >99% enantioselectivity.

Results

Reaction optimization

We initiated our investigation with a reaction of N-tert-butylformamide (1a), cyclohexanecarbaldehyde (2a), and 3,4,5-trimethoxyaniline (3a, TMP-NH₂) to synthesize chiral α-amino amide 4a. Tetrabutylammonium decatungstate (TBADT), a well-established hydrogen atom transfer (HAT) catalyst36,37, was selected to generate a carbamoyl radical from formamide 1a under irradiation with a 10 W 390 nm LED at room temperature. This nucleophilic radical was anticipated to undergo addition to the in situ formed imine, affording product 4a. Initial experiments conducted without a Brønsted acid co-catalyst38,39 yielded the desired product in 81% yield, albeit in racemic form (Table 1, entry 1). Although the introduction of chiral phosphoric acid 5a only induced 6% ee (entry 2), employing the more sterically demanding TRIP catalyst (5b) significantly improved enantioselectivity to 37% ee, suggesting that the chiral Brønsted acids could control the stereochemical outcome by activating the imine substrate (entry 3). The enhanced enantioselectivity likely results from more effective facial discrimination of the imine intermediate due to the increased steric bulk around the active site.

Table 1 Optimization of the Reactiona
Full size table

We then explored the effect of various counterions by evaluating the corresponding phosphate salts (entries 4–7). The sodium salt catalyst exhibited the most promising results in terms of both yield and enantioselectivity. We hypothesize that the sodium cation provides an optimal balance of Lewis acidity for imine activation while maintaining appropriate solubility and catalyst conformation40,41. Modification of the 3,3’-position on the BINOL backbone from TRIP to 2,4,6-tricyclopentylphenyl substituents (Na[5c]) afforded quantitative yield with slightly enhanced enantioselectivity (entry 8 vs. entry 4), supporting our strategy of tuning the chiral pocket’s steric environment. Further optimization of reaction parameters revealed that the solvent composition and ratio (MeCN:DCM) substantially influenced both yield and enantioselectivity (entries 9–10). This solvent effect likely stems from changes in the imine formation equilibrium and the organization of the transition state assembly. Lowering the reaction temperature to −30 °C considerably improved enantioselectivity to 68% (entry 11). Increasing the light source power to 40 W further enhanced the ee to 74% (entry 12). When the reaction was conducted at −78 °C, the enantioselectivity increased dramatically to 93% ee, albeit with diminished yield (58%, entry 13). Extending the reaction time to 48 h at −78 °C proved optimal, furnishing product 4a in 85% yield with excellent enantioselectivity (96% ee, entry 14). Chiral Brønsted acid 5c instead of its sodium salt provided the product with diminished yield and similar enantioselectivity (67%, 95% ee; entry 15 vs. entry 14). Notably, reducing the sodium chiral phosphate catalyst loading to 10 mol% (entry 16) resulted in only slightly diminished yield and enantioselectivity, indicating that lower catalyst loadings remain viable for practical applications.

Substrate scope

Having established optimal reaction conditions, we proceeded to explore the substrate scope of this transformation. We began by investigating various formamide derivatives while maintaining cyclohexanecarbaldehyde (2a) and 3,4,5-trimethoxyaniline (3a) as standard coupling partners (Fig. 2). The optimized conditions proved effective across a diverse range of formamides. Primary formamide was also tolerated, affording product 4b in 95% yield, albeit with lower enantioselectivity (73% ee) compared to 4a, suggesting that increased steric bulk on the formamide might be beneficial for enantiocontrol. Under this guidance, we found N-methyl (4c), N-ethyl (4d), N-isopropyl (4e) and other formamides (4f, 4g) were well-tolerated, delivering the desired products in moderate to excellent yields (53-91%) and higher enantioselectivities (83-93% ee). Notably, the reaction accommodated formamides bearing cyclic motifs of varying ring sizes. The cyclopropyl derivative 4h (96% yield, 65% ee) and cyclohexyl derivative 4i (98% yield, 93% ee) were obtained in excellent yields, though with some variation in enantioselectivity. Sterically demanding formamides (4j, 4k) afforded the corresponding products in lower yields but maintained good stereoselectivity. N-Aryl formamide could also undergo the reaction, albeit with lower yield and enantioselectivity (4l). We next examined the scope with respect to the aniline component. Aniline (4m) and para-substituted anilines bearing various functional groups including F (4n), Cl (4o), Br (4p), Me (4q), OMe (4r), and OPMP (4 s) were all compatible with the reaction conditions, furnishing products in generally high yields (68-98%) and excellent enantioselectivities (82-95% ee). Meta-substituted anilines (4t-4y) displayed similar reactivity, with yields ranging from 63% to 78% and consistently high enantioselectivities (88–94% ee). Ortho-substituted anilines were not tolerated, probably due to steric effect. The formamides derived from secondary amines (e.g., DMF) could not undergo the reaction (Please see Supplementary Information for details, Page S54). The methodology also accommodated more elaborate substitution patterns, as exemplified by products 4z-4ad (67–93% yield, 80–94% ee).

Fig. 2: Substrate scope of formamides and anilines.
Fig. 2: Substrate scope of formamides and anilines.
Full size image

Reaction conditions: The reaction of 1 (0.05 mmol), 2a (0.1 mmol) and 3 (0.125 mmol) was carried out in MeCN (0.25 mL) and DCM (0.5 mL) in the presence of TBADT (9 mol%), (R)-Na[5c] (20 mol%) under 40 W 390 nm LED light at −78 °C for 48 h. PMP 4-methoxyphenyl, Ph phenyl.

Having established the broad compatibility of formamides and anilines, we next explored the scope with respect to the aldehyde components (Fig. 3). Cyclic aliphatic aldehydes performed exceptionally well, with cyclopentyl (6a) and 4,4-difluoro-cyclohexyl (6b) derivatives providing high yields (88-99%) and excellent enantioselectivities (93–95% ee). Aldehydes derived from tetrahydropyran (6c) and N-Boc piperidine (6d) maintained high stereoselectivity (94–98% ee). Other aliphatic aldehydes (6e-6l), including simple unbranched and variously branched structures, consistently delivered the desired products with high enantioselectivities (78–92% ee). Notably, the protocol accommodated diverse functional groups including alkenes (6 m), alkynes (6n), halogens (6o, 6p), and protected functionalities (6q, 6r), providing orthogonal functionalities for subsequent molecular assembly, all delivering products with good stereoselectivity (86–94% ee). Aliphatic aldehydes with aromatic substituents, 6s-6u, also delivered high enantioselectivities (85–90% ee). Aromatic aldehydes could also undergo the reaction, albeit with much lower yield and enantioselectivity, e.g., benzaldehyde (Ar-TM, 46%, 37% ee). α-Substituted racemic aldehydes (6v-6x) afforded β-branched noncanonical amino acid derivatives, valuable molecules in modern drug development efforts42, with up to 8:1 diastereoselectivities and moderate to high enantioselectivities (71–85% ee). These products exhibited syn-selectivity, which aligns well with the Felkin-Anh model43 for α-substituted imines44,45, as shown in TS-1. The methodology could also be applied to more complex substrates. For example, oleic aldehyde (6y), aldehyde derived from erucic acid (6z), citronellal (6aa), and cyclamen aldehyde (6ab) were successfully transformed with good stereoselectivity (78–92% ee/de). The methodology’s compatibility with bioactive molecule-derived components was demonstrated through the successful incorporation of orlistat (6ac) and dopamine (6ad) moieties, underlining potential applications in medicinal chemistry and late-stage functionalization.

Fig. 3: Substrate scope of aldehydes, drugs and natural products related substrates.
Fig. 3: Substrate scope of aldehydes, drugs and natural products related substrates.
Full size image

Reaction conditions: The reaction of 1a (0.05 mmol), 2 (0.1 mmol) and 3a (0.125 mmol) was carried out in MeCN (0.25 mL) and DCM (0.5 mL) in the presence of TBADT (9 mol%), (R)-Na[5c] (20 mol%) under 40 W 390 nm LED light at −78 °C for 48 h. aReaction conditions: The reaction of 1 (0.25 mmol), 2 (0.1 mmol) and 3b (0.125 mmol) was carried out in MeCN (0.25 mL) and DCM (0.5 mL) in the presence of TBADT (4.5 mol%), (R)-Na[5c] (20 mol%) under 40 W 390 nm LED light at −78 °C for 48 h. b(S)-Na[5b] was used. Ar1 4-bromo-3,5-dimethylphenyl, Ar2 4-(tert-butyl)phenyl.

To demonstrate the broader utility of our methodology, we extended its application to more structurally complex and biologically important targets, including α,β-diamino amides46,47, glycosylated amino acids48, and peptides (Fig. 4). First, the reaction was applied to chiral N-Boc α-amino aldehydes for synthesizing α,β-diamino amides. Under standard conditions, the aldehyde derived from L-phenylalanine afforded the corresponding syn-α,β-diamino amide syn-7a with excellent diastereoselectivity, 38% yield and 98% ee. Interestingly, conducting the reaction at −30 °C without the chiral sodium phosphate catalyst still provided syn-7a with 34% yield and 93% ee. This high intrinsic diastereoselectivity likely originates from substrate control49, and the anti-Felkin-Anh selectivity could be attributed to TS-2, featuring an intramolecular hydrogen bond between the Boc-N-H and the imine. Analogous behavior was observed in the aldol-type reaction of N-Boc α-amino aldehydes50. The temperature-dependent erosion of the ee is attributed to partial racemization of the chiral aldehyde (Supplementary Table S2). High diastereoselectivity (>20:1 dr) and excellent enantioselectivity (>90% ee) were consistently observed for other N-Boc α-amino aldehydes, yielding products 7b and 7c. The protocol also proved effective in carbohydrate chemistry, converting galactopyranose-derived aldehydes into complex glycosylated amino acid derivatives (7e-7g) and glycopeptide 7h with high diastereoselectivity. Consistent with the α-amino aldehyde results, significant substrate-induced diastereocontrol was observed, allowing for high selectivity even without the chiral catalyst. The absolute and relative configuration of 7g was unambiguously determined by single-crystal X-ray analysis. Furthermore, the methodology was successfully applied to peptide synthesis—a critical area in chemical biology and medicinal chemistry51—by employing N-formyl amino esters or peptides. This approach generated a series of dipeptides (7i-7m) in good to excellent yields (65–90%) with high levels of stereoselectivity (89-90% ee for achiral formylglycinate; 87–88% de for chiral N-formyl amino esters). The method proved compatible with diverse amino acid side chains, as exemplified by glycine (7i, 7j), valine (7k), leucine (7l) and tert-leucine (7m). Notably, for these substrates, the diastereoselectivity appeared to be primarily governed by the chiral catalyst rather than the substrate’s stereocenter (7a-7h), likely due to the greater distance between the existing chiral center and the reacting prochiral imine (Please see Supplementary Information for details, Page S52). This strategy was readily extended to tripeptide synthesis, furnishing derivatives 7n-7r with high diastereoselectivities (74–90% de) using N-formyl dipeptides as precursors.

Fig. 4: Scope of a-amino aldehydes, galactopyranose based and peptides related substrates.
Fig. 4: Scope of a-amino aldehydes, galactopyranose based and peptides related substrates.
Full size image

aReaction conditions: The reaction of 1 (0.25 mmol), 2 (0.1 mmol) and 3 (0.125 mmol) was carried out in MeCN (0.5 mL), DCM (1.0 mL) in the presence of TBADT (4.5 mol%), (R)-Na[5c] (20 mol%) under 40 W 390 nm LED light at −78 °C for 96 h. b7a was synthesized without CPA-Na at −30 °C. cWithout CPA-Na. dThe de was determined by 1H NMR while other products were determined by 1H NMR and HPLC. eReaction conditions: The reaction of 1 (0.25 mmol), 2a (0.1 mmol) and 3b (0.125 mmol) was carried out in MeCN (0.5 mL), DCM (1.0 mL) in the presence of TBADT (4.5 mol%), (R)-Na[5c] (20 mol%) under 40 W 390 nm LED light at −78 °C for 96 h. Ar1 4-bromo-3,5-dimethylphenyl, Boc tert-butyloxycarbonyl, Bn benzyl.

The practical utility and scalability of our protocol were demonstrated by conducting the synthesis of α-amino amides 4a and 4z on a 1.0 mmol scale, achieving good yields (70 (64)% and 68%, respectively) and high enantioselectivities (94 (84)% and 93% ee) (Fig. 5a). Albeit with slightly lower yield and ee, a significant improvement in reaction efficiency was realized in flow (e.g., 1 mmol scale of 4a prepared in around 3.8 h in flow vs. 72 h in batch). Importantly, these amide products could be efficiently converted into the valuable free α-amino acids. For instance, 4a and 4z underwent a two-step sequence involving oxidative removal of the arylamine moiety using TCCA/HCl52, followed by tert-butylamide hydrolysis with 6 M HCl, to afford the corresponding free amino acids 8 (95% yield, 91% ee from 4a; 90% yield from 4z) with minimal erosion of stereochemical integrity. Furthermore, the method readily facilitates the incorporation of isotopic labels (Fig. 5b). Using ¹⁵N-labeled aniline (9) yielded the ¹⁵N-containing amino amide 10 with excellent conversion and high enantioselectivity (88% ee). Access to α-deuterated derivatives was demonstrated using D₁-3-phenylpropanal, providing D₁-6s in 67% yield and 89% ee. Showcasing compatibility with in situ generated substrates, D₁₁-cyclohexanecarbaldehyde—prepared via TBADT-photocatalyzed carbonylation53,54 of D₁₂-cyclohexane under CO atmosphere and light irradiation—was directly utilized in the three-component reaction to furnish the heavily deuterated product D₁₁-4a (16%, >99% ee). These results underscore the methodology’s robustness, scalability, and adaptability for preparing structurally diverse and isotopically labeled amino acid derivatives.

Fig. 5: The practical utility and mechanism studies of the reaction.
Fig. 5: The practical utility and mechanism studies of the reaction.
Full size image

a 1.0 Mmol scale reactions. b The incorporation of isotopic labels. c Control experiments. d Light on-off experiments. e Linear effect study. f Proposed mechanism. TCCA trichloroisocyanuric acid, HRMS High-Resolution Mass Spectrometry, Cy cyclohexyl, TBADT tetrabutylammonium decatungstate, TMP-NH2 3,4,5-trimethoxyaniline, PCET proton-coupled electron transfer.

Proposed mechanism

To gain insight into the reaction mechanism, several control experiments were conducted (Fig. 5c–f). Subjecting cyclopropanecarbaldehyde (2b) to the standard conditions afforded the corresponding α-amino amide 11 in 65% yield, with no evidence of cyclopropane ring-opening. This suggests that the reaction likely proceeds via the addition of a relatively nucleophilic radical intermediate55, avoiding intermediates prone to inducing ring-opening. The involvement of radical species was further substantiated by adding the radical scavenger TEMPO (1.0 equiv) to the standard reaction involving cyclohexanecarbaldehyde 2a. This resulted in a significantly decreased yield of the desired product 6a (34%) and the detection by HRMS of the TEMPO adduct of the carbamoyl radical (12) as well as the adduct (13) derived from trapping the subsequent α-amino radical intermediate formed after C-C bond formation. Conversely, species 14, potentially arising from single-electron reduction of the imine intermediate, was not detected. Additionally, light on/off experiments demonstrated that product formation ceased in the dark and resumed upon re-irradiation, confirming the reaction’s dependence on light energy and suggesting a photocatalytic process rather than a self-sustaining radical chain mechanism (Fig. 5d). We conducted an experiment to investigate the relationship between the ee of Na[5b] and the ee of product 4a. A linear correlation was observed, suggesting that a single molecule of the Na[5b] catalyst is involved in the enantiodetermining step (Fig. 5e). Based on these experimental findings and relevant literature56,57,58,59,we propose a plausible catalytic cycle (Fig. 5f). Initially, visible light excitation of the photocatalyst TBADT (PC) generates its excited state (PC*). This excited state abstracts a hydrogen atom (HAT) from formamide 1 to produce the key carbamoyl radical intermediate (I) and the reduced form of photocatalyst (PC-H). Concurrently, aldehyde 2 and amine 3 reversibly condense to form imine II. The chiral sodium phosphate catalyst (Na[5c]) then coordinates to, activates and stabilize imine II, setting the stage for the enantioselective addition of the carbamoyl radical I. This forms a transient α-amino radical species which, upon subsequent oxidation (likely involving PC-H regenerating PC) and protonation, yields the final α-amino amide product 4 and finishes the catalytic cycle. We propose transition state TS-3 to describe the enantioselective addition of carbamoyl radical I. The chiral sodium phosphate catalyst plays a multifunctional role: (i) Na⁺ acts as a Lewis acid to activate the imine intermediate; (ii) Na⁺ stabilizes the transition state via cation–π interactions; and (iii) the P = O moiety serves as a hydrogen bond acceptor for the formamide N–H, which increases electron density on the nitrogen and thereby enhances the nucleophilicity of the carbamoyl radical. This proposed hydrogen bonding interaction is supported by the observation that formamides derived from secondary amines, which lack the N–H bond, failed to yield the desired products (see Supplementary Information, pages S54 and S69).

Discussion

In conclusion, we have established a highly enantioselective photoredox protocol employing a dual catalytic system of TBADT and a chiral sodium phosphate. This method enables the direct assembly of valuable chiral α-amino amides from simple, readily available aldehydes, amines, and formamides, circumventing the need for substrate pre-functionalization. The broad utility was showcased by synthesizing structurally diverse targets, including α,β-diamino acids, glycosylated derivatives, and peptides with excellent stereocontrol. Furthermore, facile conversion to free amino acids, amenability to isotopic labeling, and demonstrated scalability underscore the practicality of this approach, offering a powerful platform for accessing diverse chiral scaffolds relevant to peptide science and medicinal chemistry.

Methods

General procedure for the synthesis of chiral amino acid derivatives

A flame-dried 10 mL Schlenk tube was charged with aniline 3 (0.125 mmol, 2.5 equiv), TBADT (0.0045 mmol, 9 mol%), and (R)-Na[5c] (0.010 mmol, 20 mol%). The vessel was subjected to three cycles of evacuation and backfilling with N2. Then, MeCN, DCM, 1 (0.05 mmol, 1.0 equiv), and 2 (0.1 mmol, 2.0 equiv) were added. The resulting mixture was frozen in liquid nitrogen, and the headspace was evacuated. The resulting solution was stirred under LED (390 nm 40 W) at −78 °C for 48 h. After completion, the mixture was removed from irradiation and concentrated under reduced pressure. Purification by silica gel chromatography (petroleum ether/ethyl acetate) afforded the target products.