Regular Articles
Mechanistic Insight into the TBHP-Mediated Decarboxylative Condensation of α-Ketoacids: Reaction Development and Application to Oligopeptide Synthesis
2023 Volume 71 Issue 5 Pages 354-359
Details
2023 Volume 71 Issue 5 Pages 354-359
With the aim of achieving the convergent elongation of peptide chains, an amide bond formation reaction that enables a peptide fragment coupling has long been pursued. The decarboxylative amidation recently reported by our group is a potential solution to this problem. In this article, a mechanistic analysis of the t-butyl hydroperoxide (TBHP) mediated-decarboxylative amidation of α-ketoacids that results in a significant advance in convergent peptide synthesis is described. Despite the observation of epimerization with low bulk substrates in preliminary studies, a systematic examination and understanding of the reaction mechanism enabled the development of a modified epimerization-free reaction whereby peptide fragment couplings using peptide α-ketoacids were successfully achieved.
Peptides are biopolymers produced from the dehydration of amino acids and play a wide variety of roles in biological organisms. Recently, peptides have gained an increased importance in pharmaceutical sciences due to the rapidly growing interest in the discovery of drugs based on middle molecules (molecules with a size between 500 Da and 20 kDa). Therefore, the development of chemical syntheses of peptides and their modified analogues, molecules that are harder to prepare via biological methods, is of great importance and thus, considerable efforts have been focused on achieving efficient methodologies for the formation of peptide bonds.1–4)
A common approach is via dehydrative condensation reactions between carboxylic acids and amines with the use of stoichiometric amounts of a condensation agent1) (Chart 1a). Recently, new amidation reactions that can be used for the fragment condensation of peptide chains consisting of multiple amino acid residues have been actively pursued in order to improve reaction efficiencies via convergent synthetic pathways.2,3) The most desirable of these methods enables the use of amines as a nucleophile thus making a wide range of starting materials accessible.5–10) We have previously described a novel t-butyl hydroperoxide (TBHP)-mediated decarboxylative amidation reaction that utilizes α-ketoacids and amines.11) This transformation is characterized by mild reaction conditions, high yields, a broad functional group tolerance, and excellent scalability. Preliminary mechanistic studies indicate that the reaction proceeds through an imine intermediate produced by the dehydration of a α-ketoacid and an amine. More importantly, it does not proceed via the formation of an active ester, which is often converted to an azlactone that results in the epimerization at the α-position (Chart 1b). We envisioned that this feature of our decarboxylative amidation would potentially make it suitable for the fragment condensation of peptide chains. However, the use of the reaction in peptide synthesis has not been explored well and thus, we describe herein a comprehensive investigation on the application of the decarboxylative amidation to peptide synthesis, including a full mechanistic analysis based on several mechanistic experiments and theoretical studies.
The results of the initial examination of the coupling between the peptide α-ketoacids and the amines are summarized in Chart 2. Recently, we have reported on the formation of amide bonds using peptide α-ketoacids that bear a leucine residue on the C-terminus. The corresponding peptides were obtained in good yields and with good diastereospecificity12) (Chart 2a). Additionally, an N-methyl amino acid (AA) residue, a residue that is recognized as being challenging to introduce due to its low reactivity, was successfully coupled without any epimerization even via a fragment condensation method (Chart 2b, 1). In the work reported here, we expanded the transformation to include α,α-disubstituted AA residues and found that the reaction also provides tetrapeptides 2 in excellent yields. Despite these remarkable results relating to the peptide synthesis, we encountered a difficulty concerning the utility of the method; we observed epimerization when a less bulky peptide α-ketoacid was used for the decarboxylative amidation (Chart 2c, 3). We believed that this phenomenon arises from imine-enamine tautomerization, however, the precise details were not well understood and the general applicability of the peptide bond formation was also unclear. Therefore, we attempted to explain the exact mechanism of the TBHP-mediated decarboxylative amidation, including the epimerization pathway, to demonstrate the utility of the transformation in peptide synthesis and as a means to suppress the observed epimerization. These mechanistic investigations are described herein.
Isolated yields are shown. Diastereomeric ratio (Dr) values were determined using chiral supercritical fluid chromatography (SFC) analysis. a) α-Ketoacid (1.2 equivalent (equiv.)), amine (1.0 equiv.), and TBHP (1.5 equiv.) in N,N-dimethylformamide (DMF) at room temperature.
First, we investigated the reaction pathway of the decarboxylative amidation reaction (Chart 3). We had previously excluded a radical-type mechanism, e.g., photo-mediated decarboxylative amidation,13,14) via radical scavenger experiments11) (Chart 3a). Therefore, we conducted various labelling experiments using 18O-labelled compounds in order to understand where the oxygen atom of the amide group comes from (Chart 3b). α-Ketoacids 4-18O and 7-18O, in which either the carbonyl or carboxy oxygen is labelled with an 18O atom, were utilized in the decarboxylative amidation with phenylethylamine 5. However, we did not observe the incorporation of any 18O atoms into either amide 6 or 8 in either of the reactions, whilst the addition of H218O also failed to provide 18O-incorporated 6. We were unable to attempt a reaction using 18O-labelled TBHP as it is difficult to prepare. Despite this, these results indicate that an oxygen atom from TBHP must be installed into the amide.
Isolated yields are shown. The 18O incorporation ratio was determined using electrospray ionization mass spectrometry (ESI-MS).
Considering the above results, there are three possible reaction paths A–C (Chart 4). Path A, the generation of a hydroxylamine followed by α-ketoacid-hydroxylamine (KAHA) ligation,15,16) can be excluded on the basis of the control experiment shown in Chart 5a; the treatment of amine 9 with TBHP did not provide the corresponding hydroxylamine 10.11) In both the alternative paths, B and C, the mixture of the amine and α-ketoacid affords an imine intermediate that is followed by the addition of TBHP to the electrophilic imino group. The final stepwise decarboxylation step differentiates path B from the concerted decarboxylation step of path C. The oxaziridine intermediate in path B was also discussed in relation to KAHA ligation,17,18) however, 11 did not react with TBHP regardless of whether a stoichiometric amount of a carboxylic acid was present (Chart 5b). This indicates that the reaction would not proceed through an oxaziridine intermediate.
DFT calculations were conducted using Gaussian 16 software at the SMD(DMF)/B3LYP-D3(BJ)/6-311G + (3df,3pd)//B3LYP-D3(BJ)/6-311G(d,p) level of theory.
To support the experimental mechanistic analysis of the decarboxylation step, we conducted density-functional theory (DFT) calculations19) on both the stepwise (path B) and the direct decarboxylation (path C) mechanisms. As a result of this exploration some possible transition states (TSs) for both pathways were obtained (Chart 5c). Regardless of whether proton shuttling was accounted for, a low Gibbs free energy value (approx. 17 kcal/mol) was obtained for the path C TS1, whilst a significantly higher value was found for the stepwise mechanism TS2 (path B). Collectively, the experimental and theoretical analyses suggest that the reaction proceeds through path C consisting of α-imino acid formation, TBHP addition, and a concerted decarboxylation.20)
Next, we focused on the relationship between the steric bulkiness of the AA side chains and the ease with which epimerization occurs. The preliminary results shown in Chart 2 meant that we expected the less bulky substituents to be somewhat problematic, however, the details behind this steric effect remained unclear. Therefore, we prepared three α-ketoacids 13–15 derived from alanine (Ala), norvaline (Nva), and leucine (Leu) and four AA esters 16–19 derived from Ala, Nva, valine (Val), and proline (Pro). In order to systematically examine the steric effects of both the α-ketoacid and amine, we conducted amidation reactions for all possible combinations of each component. The results of the 12 reactions are summarized in Chart 6 and suggest three significant tendencies for the diastereospecificity of the transformation. Firstly, when a less bulky α-ketoacid was used more epimerization was observed. Secondly, when a bulkier amine was used more epimerization was observed, and thirdly, the use of proline did not cause any epimerization regardless of the bulk of the α-ketoacid.
Isolated yields are shown. Dr values were determined using chiral SFC analysis.
Having revealed the steric effects for the decarboxylative amidation, we next conducted experiments tracking the racemization rate of the α-ketoacids (Chart 7). Given that the proposed epimerization mechanism involves an imine-enamine tautomerization, racemization should proceed even in the absence of TBHP and thus, we assumed that its rate would decrease when sterically bulky substrates were used because the enamine form would be destabilized. We began by mixing α-ketoacids 13–15 with alanine benzyl ester 16 in N,N-dimethylformamide (DMF) at room temperature, and, after the desired amount of time had elapsed, a hydrogen peroxide solution, which immediately decomposes the α-ketoacids to the corresponding carboxylic acid,21) was added to quench the reaction (Chart 7a, eq 1). The determination of the enantiomeric excess (ee) of the carboxylic acid at each time point would indirectly provide the racemization rate of the α-ketoacids, and, as initially expected, the use of bulkier α-ketoacids consistently resulted in a slower rate of racemization. Subsequently, we conducted similar experiments using a series of amines (Et3N, 16, 18, and 19) with a Leu-derived α-ketoacid 15 (Chart 7b, eq 2). In contrast to the results seen with the α-ketoacids, the bulkier amine 18 provided a higher racemization rate and the racemization in the presence of proline is significantly slower. Importantly, similar experiments that tracked the diastereomeric excess (de) of the desired amide after the treatment with TBHP demonstrated a comparable relationship between the diastereospecificity and the steric bulk of the α-ketoacids and the amines. Therefore, we can exclude the possibility that the observed trends were simply brought about by the rate of imine formation (Chart 7, eqs 3 and 4). Finally, triethylamine which cannot form an imine with the carbonyl compound did not cause any racemization despite being a stronger base than amines 16–19. This strongly supports our theory that the racemization involves an imine-enamine tautomerization.
Ee and de values were determined using chiral SFC analysis.
Based on these two systematic examinations of the epimerization, a plausible mechanism for the epimerization is shown in Chart 8. The degree of epimerization is most likely determined by the relative rates of the TBHP addition-decarboxylation pathway and the imine-enamine tautomerization. The steric bulk of the α-ketoacid side chain would reduce the rate of the epimerization by destabilizing the enamine tautomer but, on the other hand, the steric bulk of the amine side chain would cause a relative increase in the rate of epimerization by suppressing the competing TBHP addition and decarboxylation pathway. When a secondary amine such as proline is used, the TBHP addition to the cationic iminium intermediate would be extremely fast whilst the enamine tautomer would be relatively unstable due to strong intramolecular steric repulsion.
These detailed mechanistic considerations encouraged us to explore modified reaction conditions to enable an increase in the relative rate of the desired TBHP addition-decarboxylation path. We chose the amidation reaction that affords dipeptide 22 as our model decarboxylative amidation for the re-optimization of the reaction conditions because it would be particularly challenging with regards the unwanted epimerization reaction (Table 1). We started by adding LiCl to the original reaction conditions (entry 1), as we expected it to act as a Lewis acid that would activate the imine intermediates. When LiCl was added the diastereomeric ratio of 22 increased slightly (entry 2). On the other hand, the addition of LiOAc significantly dropped both the yield and diastereospecificity (entry 3). Considering these contrasting results, we assumed that the presence of the carboxylate might accelerate the epimerization and thus, we undertook a modified procedure in which the solution of α-ketoacid 15 is slowly added to the reaction mixture. The slow addition protocol was remarkably effective in the presence of LiCl and the desired dipeptide 22 was obtained in a 98 : 2 diastereomeric ratio (entry 5), while no difference was observed in the absence of the LiCl between original and modified procedures (entry 4). Interestingly, further screening of different additives under the slow addition protocol suggested, contrary to our original expectations, that the presence of a halide anion is important to suppress the epimerization; the addition of LiOTf resulted in a moderate diastereomeric ratio, whereas, the use of the tetrabutylammonium halides (TBAC and TBAB) yielded suppression effects similar to those observed with LiCl (entries 6–8). Further support for this observation was obtained by using amine hydrochloride salt 18·HCl as a halide anion source in conjunction with Et3N for neutralization (entry 9). Sufficient suppression of the epimerization occurred and resulted in a diastereomeric ratio of 95 : 5. Finally, the use of a proton sponge as a base and the fine tuning of the slow addition rate of α-ketoacid 13 provided the desired product 18 in a good yield as almost the sole diastereomer (entries 10 and 11).
 | ||||
|---|---|---|---|---|
| Entry | Additive | Method of addition | Yield (%)a) | Drb) |
| 1 | None | One portion | 82 | 85 : 15 |
| 2 | LiCl | One portion | 69 | 89 : 11 |
| 3 | LiOAc | One portion | 33 | 61 : 39 |
| 4 | None | Dropwise for 30 min | 79 | 86 : 14 |
| 5 | LiCl | Dropwise for 30 min | 58 | 98 : 2 |
| 6 | LiOTf | Dropwise for 30 min | 76 | 83 : 17 |
| 7 | TBAC | Dropwise for 30 min | 60c) | 97 : 3 |
| 8 | TBAB | Dropwise for 30 min | 40c) | 99 : 1 |
| 9d) | Et3N | Dropwise for 30 min | 61c) | 95 : 5 |
| 10d) | Proton sponge | Dropwise for 30 min | 61c) | >99 : 1 |
| 11d,e) | Proton sponge | Dropwise for 12 min | 78 | 99 : 1 |
a) Isolated yields. b) Dr was determined using chiral SFC analysis. c) NMR yields. d) Amine hydrochloride 18·HCl was used. e) TBHP (2.0 equiv.) and α-ketoacid 13 (1.5 equiv.) were used.
In order to gain the mechanic insight on the additive effect, we conducted similar racemization-tracking experiments to those in Chart 7 (Chart 9). The ee-monitoring of α-ketoacid 13 in the presence of amine 18 and a series of additives revealed that the racemization ratio is consistent with the diastereospecificity trends seen in Table 1 where the addition of amine hydrochlorides provided a slower rate (Chart 9, eq 5). On the other hand, the addition of chloroacetic acid, which has a similar pKa to the α-ketoacids, promoted the racemization. It is noted that the monitoring of the de of the desired amide resulted in a similar trend (Chart 9, eq 6). The results in both Table 1 and Chart 9 suggested that carboxylate would accelerate the racemization of the α-ketoacids, and that halide anions would improve the ratio between the desired reaction pathway and the epimerization pathway. However, the detail behind the effect of the halogen atom is still unclear.
Ee and de values were determined using chiral SFC analysis.
Having the modified protocol in hand, its generality was examined by conducting the dipeptide syntheses in which significant epimerization had been observed in Chart 6 (Chart 10a). All six reaction combinations showed significantly improved diastereomeric ratios and the dipeptides were obtained as single diastereomers (20–22, 24–26). The efficiency of the method was maintained in the presence of functional groups on the AA side chains, and the modified protocol provided serine- and methionine-containing peptides 32 and 33 in good yields and diastereospecificities22) (Chart 10b).
Isolated yields are shown. Dr values were determined using chiral SFC analysis. Yield and dr values in parenthesis are the results using the previous method. a) 1.0 mmol scale. b) The reaction proceeded for 16 h.
Finally, the reaction was applied to the peptide fragment condensations using peptide α-ketoacids (Chart 10c). Two combinations provided the desired tetrapeptides 34 and 3 in excellent diastereospecificities. When conducted using the original conditions, the reactions suffered from terrible epimerization. Collectively, the modified slow addition protocol generally improves the diastereospecificity of the decarboxylative amidation reaction. We are currently expanding the method to allow for the convergent synthesis of longer peptide chains.
A sealed tube was charged with a stirrer bar, amine hydrochloride (0.050 mmol, 1.0 equiv.), proton sponge (10.7 mg, 0.050 mmol, 1.0 equiv.) and TBHP (70% in H2O, 13.7 µL, 0.10 mmol, 2.0 equiv.) in DMF (0.25 mL). To the stirred mixture was added an α-ketoacid solution (0.075 mmol, 1.5 equiv.) in DMF (0.375 mL) dropwise for 12 min with a syringe pump. After stirring for 2 h at room temperature, the reaction mixture was concentrated under reduced pressure. The crude residue was purified using flash column chromatography (SiO2, 10 mL, 0 → 30% EtOAc/hexane) to afford the desired peptide. The dr value of the product was determined using chiral SFC analysis.
This work was supported by JSPS KAKENHI Grants JP16H06384 (Y.T.), JP20K15954 (T.N.), and JP22H02743 (T.N.). The authors gratefully acknowledge the Takeda Science Foundation (T.N.) and the Uehara Memorial Foundation (Y.T.) for financial support and the JSPS Research Fellowships for Young Scientists (N.K.).
The authors declare no conflict of interest.
This article contains supplementary materials.
