Teixobactin, a novel antibiotic isolated from soil bacterlum Eleftheria terrae by Lewis and co-workers in 2015, is the hope to break the antimicrobial resistance standoff [1]. It exerts an unparalleled activity against Gram-positive bacteria, while without detectable resistance. Unlike other antibiotics, teixobactin binds to the conserved peptidoglycan-pyrophosphate motifs of lipid Ⅱ, thus inhibiting peptidoglycan synthesis and disrupting the membrane integrity [2]. Teixobactin is a cyclodepsipeptide that contains a tedious synthesis amino acid, L-allo-enduracididine (L-allo-End₁₀) (Fig. 1) [3,4]. However, the structure-activity relationship (SAR) studies reveal that L-allo-End can be replaced by other amino acids without significantly altering the antibacterial activity [5,6]. For example, Singh group found that Leu₁₀-teixobactin (Fig. 1) and Ile₁₀-teixobactin displayed highly potent antibacterial activity that comparable to native teixobactin [7,8]. The ease of synthesis and the superior potency of Leu₁₀- and Ile₁₀-teixobactin offered solid structural foundations for the development of teixobactin analogues possessing desirable drug like properties. Based on these results, Nowick group designed and evaluated the O-acyl isopeptide prodrugs of Leu₁₀-teixobactin, resulting in enhanced solubility [9]. Aside from low solubility, another drawback that may hamper the clinical application of these peptides is that they are prone to be hydrolyzed by proteases. However, a comprehensive approach to modifying teixobactin analogues in order to mitigate their susceptibility to protease cleavage, while preserving or enhancing their antibacterial activity, has yet to be established.

A number of modification methods have been developed to improve the enzymatic stability of peptides [10-12]. Among them, N-methylation has been applied in modifying Leu₁₀-teixobactin, leading to backbone N-methylated analogues with significantly decreased or even abolished antibacterial activity, possibly because the conformation of Leu₁₀-teixobactin was altered by the lack of amide protons engaging in intramolecular H-bonds formation [13]. Unlike backbone N-methylation which often influences the H-bond pattern, backbone thioamidation only slightly alters the electron distribution of the peptide owing to a simple oxygen-sulfur exchange in peptide [14]. Thioamide serves as a quasi-isosteric substitute for amide due to the slightly larger size of sulfur compared to oxygen, and the slightly longer C=S bond in comparison to C=O bond [15]. The weak binding of thioamide to solvent water molecules can be attributed to the fact that sulfur acts as a weak hydrogen bond acceptor. Consequently, the introduction of thioamide in the peptide backbone greatly enhances its lipophilicity [11]. In addition, backbone thioamidation can increase the metabolic stability with minimal perturbation of the peptide [16]. Besides backbone thioamidation, the incorporation of fluorine serves as an alternative approach to substantially modifying the biological functionality of peptides. Fluorine has comparable atomic size to hydrogen, allowing for the substitution of hydrogen with fluorine without causing significant structural alterations [17,18]. Additionally, the C-F bond, as the strongest single bond in organic chemistry, exhibits high polarization and reactivity resistance, thereby impeding enzymatic cleavage [19].

To explore whether thioamidation and fluorination would alter the bioactivity of teixobactin analogues, we sought to introduce the thioamide and fluorine-containing groups into Leu₁₀-teixobactin. Considering that the macrocyclic ring and D-amino acids are more tolerable towards enzymatic hydrolysis [20], we hypothesized that either partially or entirely substituting the four L-amino acids (Ile₂, Ser₃, Ile₆, and Ser₇) with thioamide- or fluorine-containing amino acids could enhance enzymatic stability while preserving or augmenting the bioactivity of Leu₁₀-teixobactin. Therefore, we designed and synthesized ten analogues and evaluated their bioactivities in this work (Fig. 1).

Unlike commercially available Fmoc-protected fluorine-containing amino acids, to replace the amide bond of Leu₁₀-teixobactin by thioamide bond at Ile₂, Ser₃, Ile₆ or Ser₇ position, two thioamide precursors, Fmoc-Ile^S-NBt (NBt: benzotriazolide) and Fmoc-Ser^S(O^tBu)-NBt, were prepared according to the previous report (Scheme S1 in supporting information) [21]. With the Fmoc-protected thiobenzotriazolides and fluorine-containing amino acids in hand, the total synthesis of these peptides was carried out. Cyclization is one of the key steps in the synthesis of cyclopeptides. Given that the steric hindrance of Ala₉ is smaller than that of Ile₁₁ bearing a branched side chain, the cyclization site was selected at Ala₉ and Leu₁₀ junction, resulting in a linear precursor with C-terminus Ala₉ and N-terminus Leu₁₀ residues [22]. In addition, the esterification between D-Thr₈ and Ile₁₁ is another key step for the synthesis of this kind of cyclodepsipeptides. To circumvent O→N acyl migration during removal of Fmoc protecting groups from D-Thr₈ [23], the esterification step should be carried out after the assembly of the main peptide chain. Three generic strategies for the synthesis of teixobactin and its analogues have been developed, including a convergent strategy [24], a stepwise Fmoc-SPPS and solution-phase cyclization [25,26], and a stepwise Fmoc-SPPS synthesis and "on-resin" cyclization [27]. In this research, a modified Fmoc-SPPS and solution-phase cyclization strategy was applied to synthesize the Leu₁₀-teixobactin analogues efficiently. The synthetic route is displayed as Scheme 1. The synthesis was commenced with the loading of Fmoc-Ala-OH onto 2-chlorotritylchloride (2-CTC) resin (initial substitution is 1.0 mmol/g). The residual active sites on the resin were capped by using MeOH/DIEA/DCM. Then, Fmoc deprotection was performed by using 20% piperidine in DMF and the free amine was detected by using the Kaiser Test. According to the standard Fmoc-SPPS procedure (SI), the peptide chain was elongated from Fmoc-D-Thr-OH to Boc-N-Me-D-Phe-OH. To be noted, Fmoc-D-Thr-OH without any protection at the side chain was directly used in the coupling reaction to avoid the additional deprotection procedure needed to release the free hydroxyl group before the esterification step [28]. Fortunately, O-acylation of D-Thr₈ was not observed in the subsequent rounds of SPPS. In addition, to reduce the epimerization of the thioamidated peptides during removal of the Fmoc protecting groups, shorter deprotection time was applied (5 min × 2) [21]. Additionally, immediate coupling of the next amino acid to the peptidyl resin following Fmoc group removal is necessary to prevent the instability of the thioamino acids. Next, esterification of the hydroxyl group was successfully accomplished on the solid support by using excess Fmoc-Ile-OH, DIC and DMAP [29]. After Fmoc deprotection, Fmoc-Leu-OH was coupled to the peptidyl resin. Remarkably, DKP (2,5-diketopiperazine) side product was not detected during removal of Fmoc groups from Leu₁₀, which formation is a frequently occurring side reaction in SPPS [30]. Then, the side chain protected linear peptide was cleaved off from the resin by using TFE/AcOH/DCM. After purification, the linear peptide was cyclized between C-terminus Ala₉ and N-terminus Leu₁₀ with HATU/DIEA under high diluted condition (1 mg/mL, in DCM) [31]. Finally, global deprotection of the cyclodepsipeptide was performed by using TFA/DCM. The crude product was purified by semi-preparative RP-HPLC. According to this procedure, we synthesized Leu₁₀-teixobactin and ten analogues, the structures of which were confirmed by HRMS and ¹H NMR (supporting information).

After successful synthesis these peptides, the minimum inhibitory concentration (MIC) assay was conducted to assess the antibacterial activity of them. Levofloxacin and Leu₁₀-teixobactin were chosen as a positive control and a benchmark for activity, respectively. The MIC values are listed in Table 1. As observed, among the thioamidated analogues, 1a exhibited the most potent antibacterial activity, which was comparable or slightly superior to that of Leu₁₀-teixobactin. In contrast, 1b displayed slightly reduced antibacterial activity, while 1c, 1d and 1e showed greatly decreased or even completely lost antibacterial activity. These data indicated that thioamide substitution at Ser₇ position was the most tolerable, followed by Ile₆, and then Ile₂ and Ser₃, suggesting that the impact of thioamidation on the antibacterial activity of Leu₁₀-teixobactin depends on the specific amino acid and the site being substituted. According to the previous SAR studies, Ser₇ of teixobactin mainly binds to the pyrophosphate group of lipid Ⅱ through the backbone amide and side chain hydroxyl group, Ile₆ primarily serves as a membrane anchor with its hydrophobic side chain [32]. Furthermore, hydrogen bonds can be established between specific amino acid residues within the β-sheet interfaces of teixobactin analogues, such as Ser₃ and Ser₇, Ile₂ and Ile₅. In an aligned fibril structure, additional hydrogen bonds may form between Ser₇ and N-Me-D-Phe₁^I, as well as between Ser₃ and D-allo-Ile₅ [25]. Additionally, the presence of Cα-Cα contacts has been found between residues Ile₂-Ser₇, Ile₂-Ile₆, and Ser₃-Ser₇ in the antiparallel β-strands of [R4L10]-teixobactins [33]. Therefore, we propose that the backbone thioamidation in 1a may not only decrease the H-bonds formation between Ser₇ and water molecules, but also Ser₇ and Ser₃, Ser₇ and N-Me-D-Phe₁ to increase lipophilicity, which facilitates Ser₇ to bind with the pyrophosphate group of lipid Ⅱ, resulting in good antibacterial activity. On the other hand, the small conformational change induced by thioamidation in 1b interferes the binding of Ile₆ to membranes, leading to the slightly reduced antibacterial activity. As reported, the N-terminal residues are important for the formation of antiparallel β-sheet of teixobactin analogues that bound to the target [25]. In this regard, thioamidation of N-terminal residues, Ser₃ or Ile₂, possibly may not only result in the disruption of the dimer, but also in the impairment of the supramolecular assembly of 1c and 1d, thus causing the greatly decrease in their antibacterial activity. These findings indicated that O to S substitution in amide bond of N-terminal residues impacts the antibacterial activity of Leu₁₀-teixobactin analogues, while thioamide substitution at Ser₇ is found to be tolerable, providing 1a with good antibacterial activity. In addition, similarly enhanced antibacterial activity of Leu₁₀-teixobactin and 1a was observed in the presence of polysorbate 80, which could prevent peptide binding to plastic surfaces [1,34].

Since the synthesis of fluorine-containing serine amino acid is difficult, we focused on selective modification of Ile₆ and Ile₂, and synthesized five fluorine-containing analogues (1f-1j). The MIC assays of these analogues were carried out in the presence of polysorbate 80. It was found that 1f-1i exhibited good activity against the tested pathogens. Among them, 1h showed the best antibacterial activity, which was comparable or slightly inferior to that of Leu₁₀-teixobactin. Compared to 1h, 1g displayed retained or slightly decreased antibacterial activity, while 1f and 1i manifested 2–4 times decrease in activity. These data demonstrated that the subtle structural changes of Ile₆ with the replacement of fluorinated amino acid would not completely damage the activity. The difference of antibacteria activity between fluorinated analogues 1f-1i could be closely related to the structural similarity of the fluorine-containing amino acids with Ile. These fluorine-containing residues are analogues of Leu-with hydrogen and/or methyl group at the side chain are replaced by fluorine. The size of the trifluoromethyl group approximates twice than that of a methyl group so the steric effects of the trifluoromethyl group is close to isopropyl group, or even larger sec‑butyl and cyclohexyl groups [35]. The better antibacteria activity found for 1h indicated that TfAbu is more structurally similar to Ile, compared to fLeu, DfAbu and TfnorVal incorporated in 1f, 1g and 1i, respectively. 1j with fluorine substitution at both Ile₂ and Ile₆ caused 8–16 times loss of the activities in comparison with 1f with the same fluorine substitution at Ile₆. This observation may be attributed to the structural difference between the side chain of fLeu and Ile₂, which possibly influences the interaction with the membrane lipid [36]. These findings indicated that replacement of the Ile by fluorine-containing amino acids is tolerable at position 6, while less tolerable at position 2, which is consistent with the result of backbone thioamidation.

1a was then selected for further cytotoxicity and plasma stability investigation (supporting information) since it exhibited the most potent antibacterial activity among all the analogues. No obvious cytotoxicity (≥80% viability) was detected in HUVEC and HeLa cells treated with analogue 1a for 48 h at concentrations up to 100 µmol/L (121.77 µg/mL), which was well above the MIC values (1–2 µg/mL, 61–122 times) (Fig. S1 in supporting information). As expected, 1a had an extended half-life (t_1/2 > 72 h) compared to Leu₁₀-teixobactin, indicating that the protease stability of 1a was improved owing to the presence of the thioamide group at Ser₇ (Fig. 2). Since proteases exhibit a notable degree of accuracy and specificity in the hydrolysis of peptide bonds [37,38], the decreased degradation rate of 1a could be attributed to the three-dimensional structure changes caused by oxygen-sulfur exchange at Ser₇ which may alter the recognition of 1a by proteases within the plasma. These results demonstrated that the incorporation of a thioamide at Ser₇ into Leu₁₀-teixobactin is not only absence of cytotoxicity but sufficient to prevent enzymatic hydrolysis.

In conclusion, ten thioamidated and fluorinated Leu₁₀-teixobactin analogues were designed and synthesized based on the bioisosterism principles. The conducted SAR investigations revealed the significance of the N-terminal Ser₃ and Ile₂ residues in Leu₁₀-teixobactin for its antibacterial efficacy, as substitutions at these positions resulted in a substantial decrease in bioactivity. Furthermore, the Ile₆ residue exhibited tolerance towards both backbone thioamidation and fluoridation. Notably, the Ser₇ residue displayed the highest tolerance for thioamidation. In contrast to prior knowledge, our findings suggest that Ser₇ and Ile₆ of Leu₁₀-teixobactin can undergo additional modifications using structurally similar groups to furnish more potent analogues. The newly developed compound (1a) exhibited favorable antibacterial activity, enhanced enzymatic stability, and absence of cytotoxicity, thus has great potential for further exploration in antibacterial drug discovery.

The experiment of plasma stability was performed under the guideline of Institutional Ethical Committee of Animal Experimentation of Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Yanye Fan: Writing – original draft, Methodology, Investigation, Data curation. Jingjing Chen: Investigation. Bichun Chen: Investigation. Jinyu Bai: Investigation. Bowen Yang: Investigation. Feng Liang: Project administration, Conceptualization. Lijing Fang: Writing – review & editing, Project administration, Funding acquisition, Conceptualization.

This research is supported by the National Natural Science Foundation of China (No. 21977111), the Natural Science Foundation of Guangdong Province (No. 2023A1515011765), Shenzhen Science and Technology Program (Nos. JCYJ20220818101404010, JCYJ20220818100412028). The authors wish to acknowledge the Peking University Shenzhen Graduate School for the assistance of Mass facility. WuXi AppTec (Shanghai) Co., Ltd. is thanked for MIC assays.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110075.