Introduction

Peptides are one type of the most important compounds with high biological activities, which are widely present in many natural products, pharmaceutically relevant molecules, and biological systems1,2,3. Especially in recent years, there appears a growing interest in therapeutic peptides1,2,3 and more and more peptide drugs have been developed (Fig. 1a)4,5,6. Development of alternative new methods for the synthesis of peptides is always highly desirable and potentially useful7,8.

Fig. 1: Peptides and transamination.
figure 1

a Peptide drugs. b Biological transamination. c Catalyst design. d 1,3-Proton shift. e Transamination of α-keto amides.

Full size image

Enzymatic transamination is an important process to produce chiral amines such as amino acids in biological systems (Fig. 1b)9,10, which is promoted by transaminases (Fig. 1c) with pyridoxal/pyridoxamine 5′-phosphates as the coenzyme9,10,11,12,13. Mimicking the biological process14, i.e., asymmetric biomimetic transamination, affords a highly intriguing method to synthesize NH2-free amines from readily available carbonyl compounds15,16,17. The chemistry has attracted much attention since the 1970s18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36. The studies mainly include stoichiometric chiral pyridoxamine-promoted asymmetric transamination of α-keto acids18,19,20,21, pyridoxal/pyridoxamine-catalyzed asymmetric transamination of α-keto acids22,23,24,25,26, and chiral base/Lewis acid-catalyzed asymmetric transamination of α-keto esters and activated ketones27,28,29,30,31,32,33,34,35,36. Asymmetric transamination of α-keto amides can potentially provide an appealing new strategy to produce peptides. However, to the best of our knowledge, asymmetric transamination of α-keto amides to peptides are barely reported37,38, although the reverse process, transamination of peptides at the N-termini to α-keto amides, has been widely developed and already have been successfully applied to protein modification38,39,40,41,42,43,44,45. In contrast to the transamination of α-keto acids, asymmetric transamination of α-keto amides to peptides remains a challenge for enzymatic catalysis likely due to non-naturally occurring process and it is also a challenge for chemical catalysis probably because the complicated structure of α-keto amides requires more active catalysts to promote transamination.

Previous studies have suggested that asymmetric 1,3-proton shift between the ketimine intermediate and the aldimine is likely a key step for biological transamination (Fig. 1d)11,12,13,46,47. In order to accelerate this step, evolution has elegantly optimized transaminases by incorporating a Lys residue at an appropriate position48,49,50. The ε-NH2 group of the Lys residue serves as an intramolecular base to deprotonate the benzylic C-H of the ketimine (Fig. 1d)48,49,50. In addition, the pKa values of the pyridine N in the coenzyme pyridoxal 5′-phosphate (PLP) and pyridoxamine 5′-phosphate (PMP) are around 8.551,52,53, thus the pyridine N is predominantly protonated in biological systems (near pH 7)53,54,55,56. The strong electron-withdrawing property of the protonated pyridine ring helps to increase the acidity of the benzylic C-H bond (Fig. 1d)54,55,56,57,58. The two effects work together to promote the transformation from the ketimine to the aldimine via 1,3-proton shift, magically accelerating transamination process. Inspired by the controlled protonation of the pyridine N of PLP in biological systems, Rapoport has developed N-methylpyridinium-4-carboxaldehyde benzenesulfonate (Rapoport’s salt) as an effective transamination reagent for conversion of amines to carbonyl compounds59. Francis have found that Rapoport’s salt displays significantly improved efficiency in the transamination of proteins as compared to pyridinium-4-carboxaldehyde, converting the N-termini into the corresponding carbonyl groups44,45. Recently, we have proved that introducing an amine side arm to a chiral pyridoxamine can remarkably increase its activity and enantioselectivity for asymmetric transamination of α-keto acids26 and also have observed that quaternization of the pyridine N of a chiral pyridoxal leads to dramatical improvement of catalytic activity for asymmetric biomimetic Mannich reaction60 and aldol reaction61 of glycinate. On the basis of the structural characteristics of transaminases46,47,48,49,50,53,54,55,56 as well as the previously reported studies26,44,45,59,60,61,62, we has designed N-quaternized biaryl axially chiral pyridoxamines 1 bearing an amine side arm, mimicking transaminases in multiple aspects for catalytic asymmetric transamination of α-keto amides to peptides (Fig. 1c). The quaternization always keeps the pyridine ring with strong electron-withdrawing ability to improve the benzylic C-H acidity of the ketimine intermediate during transamination, no matter under acidic or basic conditions60,61. Like the Lys residue does in biological transamination, the amine side arm can serve as an intramolecular base to facilitate 1,3-proton shift.

Here we show that asymmetric biomimetic transamination of α-keto amides 2 can be achieved by using chiral pyridoxamines 1 as the catalyst, to produce various peptides 3 with excellent enantiopurities (Fig. 1e).

Results

Catalyst synthesis

The synthesis of chiral pyridoxamines 1 started with reductive amination of compound 526 to introduce the amine side chain. Protecting the two amine groups with di-tert-butyl dicarbonate gave intermediates 6 (Fig. 2). Treatment of 6 with methyl iodide and subsequent deprotection with hydrochloric acid afforded N-methyl pyridoxamines 1a-e in good yields.

Fig. 2: Synthesis chiral pyridoxamines 1.
figure 2

The detailed synthetic procedures have been presented in Supplementary Information (SI). THF = tetrahydrofuran, DCM = dichloromethane.

Full size image

Condition optimization

With diphenylglycine (4) as the amine source25,26,63, catalyst chiral pyridoxamine 1b was first tested for the transamination of glycinyl α-keto amide 2a (Fig. 3, entry 1). The originally-formed NH2-free transamination product was treated with di-tert-butyl dicarbonate to avoid the cyclization to piperazinedione during the isolation64, to give the corresponding N-Boc-protected dipeptide 3a in 20% yield with 76% ee. Additives have significant impacts on the reaction in terms of enantioselectivity and activity. Increased yield and enantioselectivity were obtained for transaminations performed in MeOH/H2O or TFEA/H2O with HOAc/KOAc or HOAc/Na2HPO4 as the additives (Fig. 3, entries 2 and 10 vs 3–9). Chiral pyridoxamine 1b exhibited the best performance among the catalysts 1a-e examined (Fig. 3, entries 10–14).

Fig. 3: Investigation of reaction parameters.
figure 3

TFEA = 2,2,2-trifluoroethanol. aReaction conditions: 2a (0.10 mmol), 4 (0.11 mmol), 1 (0.0050 mmol), HOAc (0.40 mmol), base (0.20 mmol) in solvent (0.48 mL) and H2O (0.12 mL) at 20 °C for 48 h unless otherwise stated. The reaction mixtures were then treated with di-tert-butyl dicarbonate (0.30 mmol) at rt for 3 h. bIsolated yields based on α-keto amide 2a. cThe ee values were determined by HPLC analysis. dReaction time was 72 h.

Full size image

Substrate scope

Under the optimal conditions, various glycinyl α-keto amides containing alkyl (for 3b-e), aromatic (for 3a and 3f-i), or heteroatomic alkyl (for 3j-k) groups were all smoothly transaminated to give the corresponding N-Boc-protected glycinyl dipeptides 3a-k in 70-94% yields with up to 98% ee (Fig. 4). Chiral glycinyl α-keto amide (for 3 l) displayed excellent diastereoselectivity (98:2 dr). Transamination of α-keto phenylbutanamides of chiral amino acid esters produced various N-Boc-protected dipeptides 3m-y in 56–93% yields with up to 99:1 diastereoselectivity. Peptidyl α-keto amides were also effective for the asymmetric transamination, to form tripeptides 3z-ab and tetrapeptides 3ac-ad in 60–87% yields with excellent diastereoselectivities under very mild conditions. Various functional groups such as C-C double bond (for 3c, 3 l and 3 y), NH2-sensitive bromide (for 3j), silyl group (for 3k), OH group of Tyr (for 3 s), NH group of Trp (for 3t), amide CONH2 of Asn (for 3 v), Boc-protected Lys residue (for 3w), Boc-protected guanidine (for 3x), and basic NH2 group of Lys (for 3ab) were all well tolerated by the transamination likely due to the mild reaction conditions.

Fig. 4: Asymmetric biomimetic transamination of α-keto amides.
figure 4

TBDPS = tert-butyldiphenylsilyl. aReaction conditions: 2 (0.10 mmol), 4 (0.11 mmol), 1b (0.0050 mmol), HOAc (0.40 mmol), Na2HPO4 or KOAc (0.20 mmol) in CF3CH2OH or MeOH (0.48 mL) and H2O (0.12 mL) for 48 or 72 h unless otherwise stated (See SI). For 3m-y, the reactions were carried out in a double scale. Isolated yields were based on α-keto amides 2. The ee and dr values were determined by HPLC analysis. bThe crude reaction mixture was treated with di-tert-butyl dicarbonate (0.30 mmol) at room temperature for 3 h. cThe reaction was carried out in a double scale with 7 (0.040 mmol, 20 mol %) at 50 °C for 72 h.

Full size image

In order to investigate the impacts of catalyst and substrates on diastereomeric induction, several representative α-keto amides (for 3m-n, 3r, 3v-y, and 3aa-ab) were examined respectively using (S)-1b (5 mol%), (R)-1b (5 mol%), and achiral pyridoxamine 7 (20 mol%) as the catalyst. The corresponding peptides were formed with S configurations of the newly generated chiral centers from catalyst (R)-1b and R configurations from (S)-1b. The chiral pyridoxamine catalyst dominated the stereoselectivity, while the chiral groups on the amino acid residues of α-keto amides throwed little influence on the diastereomeric induction probably due to being far away from the reaction centers as well as the flexibility of the skeletons of the α-keto amides. No matter which configuration of the catalyst 1b was applied, excellent diastereoselectivities were always obtained, even for α-keto amides (for 3n and 3aa) with a nearby bulky chiral amino acid residue and for those that displayed obvious substrate-induction on diastereoselectivity in 7-catalyzed non-asymmetric transamination (5:95 dr for 3r and 19:81 dr for 3x). For α-keto amide 2 y bearing two nearby chiral centers, a pair of diastereomers (R,R,S)-3y and (R,S,S)-3y were respectively obtained in good yields with high enantiopurities by using (S)-1b and (R)-1b as the catalyst. The absolute configurations of the newly generated chiral centers of peptides 3 were assigned by analog, based on the X-ray analysis of 3d, 3 m, and 3r (also see Supplementary Figs. 1–3 in SI).

Synthetic applications

Divergent extending an additional amino acid unit from a central peptide is of great interest for peptide drug screening and bioactivity studies. The synthesis would be difficult when the extended unit is a commercially unavailable unnatural amino acid. The transamination process provides an efficient strategy for the amino acid extending. For example, starting from the benzyl ester of protease inhibitor Ubenimex (8)65, condensation with α-keto acids and subsequent asymmetric transamination afforded a variety of enantiopure peptides 3ae-ai with one more amino acid residue extended (Fig. 5a). The unprotected OH group remained untouched during the condensation and transamination.

Fig. 5: Synthetic applications.
figure 5

a Divergent synthesis of peptides. NMM = N-methylmorpholine. b Synthesis of peptides via successive transamination. EDCl = 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride. HOBt = 1-hydroxybenzotriazole.

Full size image

Based on the “condensation-transamination” process, a new strategy for the synthesis of peptides also can be developed. As illustrated in Fig. 5b, the methyl ester of DPP-IV inhibitor Diprotin A (10)66 underwent condensation with α-keto acid 9a and subsequent asymmetric transamination, forming tetrapeptide 3ad with excellent diastereoselectivity. Repeating the reaction sequence two more times afforded hexapeptide 3ak with high enantiopurity. The chirality of the extended amino acid residues was established along with the transamination process. The protocol is especially attractive for the synthesis of peptides made of unnatural amino acids, since it doesn’t need great efforts to make NH2-protected chiral unnatural amino acids before the amide bond formation.

Reaction mechanism

A plausible mechanism was proposed for the transamination (Fig. 6a)11,12,13,25,26. Pyridoxamine 1b condenses with α-keto amide 2 to form ketimine 11, which undergoes asymmetric 1,3-proton shift to aldimine 13 under the assistance of the amine side arm46,47,48,49,50. Hydrolysis of aldimine 13 releases peptide 3 and generates the pydridoxal, which is in situ converted into iminium 14 via intramolecular condensation. The iminium 14 then undergoes decarboxylative transamination with the amine source diphenylglycine (4) back to pyridoxamine catalyst 1b25,26,63, completing a catalytic cycle.

Fig. 6: Reaction mechanism.
figure 6

a Proposed reaction pathway. b Comparison of catalysts. c Proposed transition state for the asymmetric 1,3-proton shift.

Full size image

As expected, N-quaternization of the pyridine ring of the chiral pyridoxamines resulted in higher catalytic activity and better enantioselectivity for the asymmetric transamination (Fig. 6b, 1b vs 1g). The stronger electron-withdrawing property makes the benzylic C-H of ketimine 11 more acidic and also stabilizes the corresponding delocalized carbanion 12 better54,55,56,57,58,67,68,69, thus favoring the 1,3-proton shift and accelerating the transamination process. The control experiment confirmed the amazing effect of the amine side arm again (Fig. 6b, 1b vs 1f). Introducing an acetyl group onto the nitrogen to eliminate the basicity of the amine on the side arm led to marked decreases in activity and enantioselectivity. The amine side arm not only promotes the 1,3-proton shift by acting as an intramolecular base to deprotonate the benzylic C-H of ketimine 11 (Fig. 6a) but also helps to orient the α-keto amide by hydrogen bonding with the carbonyl oxygen of the amide group (Fig. 6c), resulting in improved activity and stereoselectivity48,49,50. Protonation of the delocalized carbanion 12 occurs at α−position of the amide group from the up side of the pyridine ring away from the amine side arm26,46,47, to form the newly generated chiral center with S configuration from catalyst (R)-1b.

Discussion

In summary, based on the concept of multiple imitation of transaminases, we have developed N-quaternized axially chiral pyridoxamines 1 containing an amine side arm. With pyridoxamine 1b as the catalyst, challenging substrates α-keto amides were successfully transaminated to peptides in good yields with excellent enantio- and diastereoselectivities. The catalyst dominated the diastereoselective control for the transamination of chiral α-keto amides. Thus, a pair of diastereomeric peptides could be respectively obtained with high enantiopurities by switching the configuration of the pyridoxamine catalyst. The strong electron-withdrawing property of the N-quaternized pyridine ring together with the cooperative catalysis of the amine side arm account for the increased catalytic activity and selectivity of the pyridoxamine 1b in the transamination. The reaction can provide an efficient strategy for divergent and successive extension of peptides via condensation-transamination reaction sequence, which is especially attractive for the synthesis of peptides made of unnatural amino acids.

Methods

General procedure for the asymmetric biomimetic transamination Reaction (Fig. 4)

A mixture of α-keto amide 2 (0.10 mmol), chiral pyridoxamine 1b (0.0050 mmol), 2,2-diphenylglycine (4) (0.11 mmol), HOAc (0.40 mmol), Na2HPO4 or KOAc (0.20 mmol), CF3CH2OH or MeOH (0.48 mL), and H2O (0.12 mL) was stirred at 16–25 °C for the specified time. For glycinyl α-keto amides (for 3a-l) and α-keto phenylbutanamides of amino acid esters (for 3m-y), the crude reaction mixtures were treated with di-tert-butyl dicarbonate (0.3 mmol) at room temperature for 3 h after the transamination was completed, then concentrated via rotary evaporator to remove most of the solvent and isolated by column chromatography on silica gel with a mixed solvent ethyl acetate and petroleum ether as the eluant to give the products N-Boc-protected dipeptides 3a-y. For Peptidyl α-keto amides (for 3z-ad), the reaction mixtures were submitted to concentration via rotary evaporator to remove most of the solvent and then isolated by column chromatography on silica gel with a mixed solvent of dichloromethane, methanol and ammonia solution in ethanol (2.9 M) as the eluant to give the transamination products tripeptides 3z-aa and tetrapeptides 3ab-ad without NH2-protection. The ee and dr values of 3a-ac were determined by HPLC analysis.