Automated solid-phase synthesis of metabolically stabilized triazolo-peptidomimetics

1 SCOPE AND COMMENTS

The importance of peptides in medicine has notably grown.¹ Their ability to interact with clinically relevant proteins with exquisite affinity and specificity make them ideal drug candidates. Peptides are also excellent vehicles (vectors) to transport selectively imaging probes and cytotoxic payloads to the site of diseases due to their suitable pharmacokinetic profiles and tissue penetration properties, as well as convenient and economic availability.² However, the use of peptides in the clinic is hampered by their metabolic lability, which can limit their accumulation at the biological target(s).³ As a result, many research groups have reported different methods for the stabilization of peptides, for example, by structural modifications (e.g., N-methylation⁴ or reduction of amide bonds, employment of amide bond surrogates, or incorporation of unnatural amino acids⁵) and/or conformational restraints (e.g., head-to-tail cyclization and stapling⁶). The use of metabolically stable amide bond bioisosters represents an attractive approach to enhance the metabolic stability of a peptide while preserving its biological function(s). Among the examples reported (e.g., sulfonamides and semicarbazides⁷), 1,4-disubstituted 1,2,3-triazoles (Tz) have been shown to be suitable trans-amide bond surrogates.⁸ On the other hand, 1,5-disubstituted 1,2,3-trizoles can serve as analogs of cis-amine bonds.⁹ These metabolically stable Tz-based bioisosters of amide bonds share similarities with the amido functional group in terms of size, planarity, H-bonding properties, and polarity (Figure 1).

The incorporation of triazoles into the peptide's backbone can be accomplished in different ways. An elegant approach relies on the use of solid-phase chemistry for both peptide elongation and triazole incorporation. This way, the peptide is elongated following standard Fmoc/^tBu chemistry until the position of Tz-insertion, and this is accomplished on solid support in two sequential synthetic steps. The first reaction makes use of a diazo-transfer reagent (e.g., imidazole-1-sulfonyl azide¹⁰) for the amide-to-azide conversion at the N-terminus of the peptide. The second reaction consists of the [2+3] dipolar cycloaddition between the introduced azide functionality and an appropriate alkyne derivatives of amino acids.¹¹ The in solution synthesis of α-amino alkynes starting from commercial amino acid derivatives is described elsewhere.¹² Employment of specific metal catalysts yields different regioisomers of the triazole heterocycle. While Copper(I)¹³ is used for the synthesis of 1,4-disubstituted 1,2,3-triazoles, Ruthenium(II)¹⁴ has proven useful to synthesize the 1,5-disubstituted regioisomers. Albeit reported,¹⁵ the introduction of 1,5-disubstituted 1,2,3-triazoles into peptides on solid-phase is challenging, at least in our hands.¹⁶ We therefore focus here on the application of 1,4-disubstituted 1,2,3-triazoles as trans-amide bond surrogates.

The emergence and widespread availability of automated peptide synthesizers offers an opportunity for the rapid synthesis of not only peptides but also peptidomimetics, provided that the chemistry used is compatible with SPPS. Even though the diazo-transfer¹⁷ reaction and CuAAC¹⁸ have been reported for the synthesis of triazolo-peptidomimetics by manual SPPS, they have not yet been adapted to an automated setup. Here, we report a robust protocol for the fast and facile automated solid-phase synthesis of backbone modified triazole-bearing peptidomimetics.

The compatibility of the diazo-transfer and CuAAC reactions with an automated synthesis setup was studied with the model peptide sequence H-GlyΨ[Tz]His-Leu-Nle-NH_2.¹⁹ Unlike in manual synthesis, all solutions of reagents need to be prepared in advance, and the long-term stability of the reagents is essential for automated and programmed reaction sequences. The first attempts to perform the diazo-transfer reaction were conducted in DMF using ISA·HCl (5 equiv.) and DIPEA (10 equiv.) with N-terminally deprotected H-His (Trt)-Leu-Nle-NH₂ bound to a MBHA resin. These conditions did not yield the desired triazolo-peptidomimetic, and an insufficient stability of the diazo-transfer reagent was suspected to be the limiting factor. Subsequent stability studies of ISA·HCl in DMF (HPLC-grade) revealed a significant degradation of the sulfonyl azide to the corresponding sulfonyl amine (HPLC-MS analysis). Only 30% of the diazo-transfer reagent remained intact after 4 h of incubation in DMF at RT (22°C) (Figure S1). Thus, the storage of ISA·HCl solutions in the peptide synthesizer for the insertion of Tz in the N-terminal region of medium or longer length peptides turned out not to be an option. Similarly, DMSO was not able to preserve the integrity of ISA·HCl over time, and thus, these alternative solvents for SPPS were not followed up further. In contrast, ISA·HCl was found sufficiently stable in aqueous solutions with >80% of intact sulfonyl azide remaining after 24 h at RT in the presence of oxygen (air). In order to counteract potential issues with the swelling properties of polystyrene resins in the presence of water, the water content in the final reaction mixture was minimized to achieve a DMF-to-water ratio of 91:9 (total 3 ml DMF from solutions of DIPEA and 300 μl H₂O from the ISA·HCl solution). With the aqueous solution of ISA·HCl and applying the same reaction conditions as described above, the conversion of the N-terminal amine of the peptide to the azido peptide (N₃-His-Leu-Nle-NH₂) was 95%, whereas the crude purity determined by HPLC-MS was >85%. The reaction conditions were further optimized by screening the use of different equiv. of employed ISA·HCl and DIPEA, varying the temperature and time, as well as the utilization of CuSO₄ catalyst.²⁰ Under the optimized conditions as shown in Table 1, approximately 99% conversion of the N-terminal amine of the peptide to an azido-functionality was achieved providing the azido-peptide in 90% purity (Figure 2; Table 1, entry 5).

Next, we turned our attention to the optimization of the CuAAC on solid support adopted to an automated setup. Similar to the case of the diazo-transfer reagent, the stability of the Cu(I) catalyst in solution is crucial. [Cu(CH₃CN)₄]PF₆ was chosen as Cu(I) source due to its commercial availability and reported suitable shelf-life. However, the weakly bound acetonitrile ligands easily dissociate in solution exposing the metal center and therefore posing the risk of the formation of unreactive Cu(II) species. In order to avoid deactivation of the Cu(I) catalysts by oxidation to Cu(II) upon exposure to air, TBTA²¹ was used as stabilizing ligand. TBTA forms a stable complex with Cu(I) and preserves the oxidation state of the metal while making it accessible for catalysis. The automated CuAAC between the azido group at the N-terminus of the peptide N₃-His-Leu-Nle-NH₂ on solid support and the corresponding α-amino alkyne were done with a pre-prepared equimolar solution of [Cu(CH₃CN)₄]PF₆ and TBTA in DMF (0.01 M). Automated addition of solutions of the catalyst, the α-amino alkyne (Table 2, entry 1: Fmoc-protected propargyl amine as Gly analog), and DIPEA, the reaction was allowed to proceed for 1 h at 70°C. The triazolo-peptidomimetic H-GlyΨ[Tz]His-Leu-Nle-NH₂ was obtained with a conversion of 82% and a purity of 85%. All attempts to shorten the reaction time by varying the temperature and/or the amount of reagents/catalyst did not result in an improvement (Table 2, entries 2–4). Therefore, the initially tested conditions (Table 2, entry 1) were applied for all further experiments. Only in the case of storage of the solution containing the catalyst [Cu(I)(TBTA)] for more than 24 h, a slight decrease of the efficiency of the CuAAC was observed (76% of crude purity; Figure S2). Therefore, no special precautions had to be applied for the automated CuAAC on solid support.

The optimized two-step automated procedure was applied to the synthesis of different triazolo-peptidomimetics in order to demonstrate the general applicability of our new methodology. First, we looked at the application of different resins. Thus, the triazole-containing minimal binding sequence of bombesin¹⁹ was synthesized on MBHA and Wang resins (Table 3, entries 1 and 2). In both cases, the desired triazolo-peptidomimetic was obtained in a total synthesis time of 10–11 h with comparable purities. Thus, no preference for either resin became apparent. In the following, the procedure was also applied to the automated solid-phase synthesis of reported triazole-containing analogs of Leu-enkephalin²² and neurotensin²³ (Table 3, entries 3–4). Together with the above discussed bombesin derivate, the chosen examples represent the application of different α-amino alkynes without and with side chain functionalities as well as of varying steric demands. At the same time, we varied the position of the triazole in the amino acid sequence relative to the resin to investigate the impact of potential steric effects on the yields and to confirm the stability of reagents in solution when stored for different time periods in the peptide synthesizer. Finally, we implemented the methodology for the synthesis of a minigastrin derivative incorporating multiple triazoles (Table 3, entry 5).^{24, 25} In all cases, the desired triazolo-peptidomimetics were obtained within reasonable time (max. overnight), with good conversion for both standard SPPS (Fmoc/^tBu chemistry) and the diazo-transfer/CuAAC steps (S3–S7).

In conclusion, we have developed a robust and versatile protocol for the automated synthesis of triazolo-peptidomimetics on solid support using a microwave-assisted peptide synthesizer. The protocol is applicable to the synthesis of structurally diverse triazolo-peptidomimetics containing one or more 1,4-disubstituted 1,2,3-triazoles in their backbone, and so far, no limitations have been identified. We encourage the peptide chemistry community to apply the new and convenient methodology in the future to get rapid access to this promising class of metabolically stabilized peptidomimetics.

2 EXPERIMENTAL PROCEDURE

2.2 General notes

2.2.1 Amino alkynes

The amino alkynes were synthesized from the corresponding amino acids following a three-step synthetic route. These three steps include the formation of the Weinreb amide, the reduction to the corresponding amino aldehyde, and a Seyferth-Gilbert homologation using the Bestmann-Ohira reagent (Figure 3, blue route). In case of amino acids containing acidic side chain functionalities, an alternative route is recommended (Figure 3, green route). In this case, the carboxylic acid is reduced to the alcohol followed by Swern oxidation. Once the aldehyde is obtained, the homologation is performed with the Bestmann-Ohira reagent as described above. If deprotection of the Fmoc protecting group is observed as a result of the presence of base (K₂CO₃) required for the Seyferth-Gilbert homologation in the last step, the amino group of the obtained crude product can be protected again using Fmoc-ONSu. These synthesis protocols can, at least in our hands, result in the partial racemization of the resulting amino alkynes (likely at the stage of the aldehyde intermediate).²⁶ The presence of enantiomers is best analyzed by chiral-HPLC.²⁷ Alternatively, the amino alkynes can be coupled with an enantiomerically pure amino acid, followed by analysis of the diastereomeric purity of the resulting dipeptide by NMR or HPLC.²⁸ The incorporation of the partially racemized amino alkynes in the peptide backbone via the CuAAC reaction results in the formation of a mixture of diastereomers. The desired product can be readily identified by the comparison of the diastereomeric ratio of the peptide (determined by UV-HPLC) and the enantiomeric ration of the chiral amino alkyne.²⁵ Alternatively, the (partially) racemized amino alkyne can be subjected to (semi-)preparative chiral-HPLC purification prior to use. All diasteromeric mixtures of final triazolo-peptidomimetics studied by us so far could be separated efficiently by HPLC using a C18 column.