Sortase-mediated generation of various macrocyclic peptide scaffolds from low-reactivity electrophilic substrates Authors: Zhe-Min Xia, Hui-Min Wei, Jun-Hao Xue, Ya-Ning Sun, Xiao-Cui Wan, Hua-Rong Tong, Li Zhou, Pan Shi, Ge-Min Fang, Changlin Tian Published: 2026 ABSTRACT Macrocyclic peptides possess numerous advantages, including enhanced tissue penetration, low immunogenicity, good resistance to protease degradation, and high targeting specificity, making them increasingly attractive for drug development. Genetically encoded cyclic peptide libraries enable rapid and convenient de novo discovery of cyclic peptide lead compounds, avoiding the tedious and time-consuming process of isolating natural cyclic peptides. In genetically encoded cyclic peptide libraries, the construction of cyclic peptide frameworks typically requires bioorthogonal reactions to cyclize peptides generated by phages or mRNA. The diversity of the cyclic peptide framework is limited by the type of cyclization reaction. Late-stage cysteine crosslinking is a convenient method for constructing genetically encoded cyclic peptides with different cyclic frameworks. However, enzyme-free cyclization strategies, while versatile, are generally constrained by inherent trade-offs among reaction kinetics, chemoselectivity, and biocompatibility—factors that become particularly critical when preserving the functional integrity of the genetic payload in display platforms. Sortase-mediated peptide ligation and one-pot cyclization (SPLOC) is a mild chemoenzymatic modification strategy that utilizes sortase to introduce a low-reactivity chloroacetyl group into phage-displayed peptides, followed by intramolecular cyclization with a cysteine residue to generate cyclic peptides. Previous sortase-based platforms only used the chloroacetyl group, limiting exploration of a broad chemical space. In this study, we designed and synthesized a series of low-reactivity electrophilic peptide substrates, tested their compatibility with sortase, and generated various cyclic peptide frameworks. Combined with phage display, we discovered several cyclic peptide ligands targeting human IgG1 Fc protein and TROP2, demonstrating the practicality of our approach. The finding of the compatibility of sortase with various low-reactivity electrophilic peptide substrates will provide a useful chemoenzymatic method for constructing cyclic peptide libraries with diverse frameworks. Introduction Macrocyclic peptides are a valuable class of molecules, combining the advantages of traditional small molecules and biomacromolecules.1 They exhibit good tissue penetration and low immunogenicity, with some even capable of crossing cell membranes.2,3 Macrocyclic peptides are also considered to have excellent targeting capabilities for flat protein–protein interaction interfaces, which are usually difficult for traditional small molecules.4,5 The conformationally constrained structure of macrocyclic peptides makes them resistant to degradation by proteolytic enzymes, as exemplified by orally administered cyclosporine.6 Most FDA-approved macrocyclic peptide drugs are based on modification or derivatization of natural cyclic peptides.7 However, the reported structural frameworks of naturally derived cyclic peptides are limited and not applicable to many target proteins. Chemical modification and engineering of natural cyclic peptides typically involve time-consuming and cumbersome multi-step chemical synthesis.8De novo discovery of novel cyclic peptide scaffolds from microorganisms relies on bioinformatics and tedious isolation, purification, and identification procedures, which are insufficient to meet the current needs of peptide drug development.9,10 An effective approach is to develop methods for generating diverse macrocyclic peptide scaffolds and combine them with gene-encoding technologies to construct large-scale cyclic peptide libraries.11–13 Phage display and mRNA display technologies, as the most commonly used gene-encoding techniques, have been widely applied to the construction of cyclic peptide libraries.14–16 Through random chemical synthesis of gene sequences, phage display and mRNA display technologies can easily generate billions of peptides with different sequence characteristics. Constructing a macrocyclic peptide framework requires the macrocyclization of specific residues of the peptides encoded by the phage or mRNA system, which typically involves cross-linking with bioorthgonal reactions.17,18 In mRNA-displayed cyclic peptide library technology, the construction of cyclic peptide frameworks can be achieved through non-natural amino acid insertion techniques, mainly relying on orthogonal tRNA/tRNA synthetases or in vitro aminoacylated tRNA prepared by Flexizyme.19,20 Note that the types of cyclic peptide frameworks that can be generated by mRNA display are limited, possibly due to the specific requirements of the ribosomal translation system for the amino acid types.21 Orthogonal tRNA/tRNA synthetases have also been used to construct phage-displayed cyclic peptide libraries.22,23 Similar to mRNA display systems, this strategy produces a limited number of cyclic peptide frameworks, which restricts the discovery of certain cyclic peptide molecules with unique spatial chemical structures. Another effective technique for constructing gene-encoded cyclic peptide libraries is cysteine-based crosslinking of phage-displayed peptide libraries.24–27 A representative example in this field is the TBMB (tribromomethylbenzene)-based phage library proposed by Heinis et al., which has been widely used to generate bicyclic peptide libraries.28 TBMB undergoes an alkylation reaction with three cysteine residues in the phage-displayed peptide library sequence to form bicyclic peptide molecules.29 This unique bicyclic peptide exhibits enhanced drug properties, including high enzymatic stability, high targeting affinity, and good tissue penetration.30 To further expand the types of cyclic peptide library frameworks, Heinis et al. constructed dozens of cyclic peptide scaffolds using various bireactive electrophilic reagents, providing diverse frameworks for peptide ligand discovery.31 Compared with other approaches, the cysteine cross-linking method is simpler to operate and can construct a large number of different peptide frameworks at low cost.32 Notably, Derda et al. reported that cysteine-based crosslinking chemistry can be combined with other chemical reactions, enabling phages to display more unique pharmacophores.33,34 In addition, thioether-bonded cyclic peptide frameworks generated by cysteine crosslinking have proven to be good scaffolds for developing oral peptide drugs, such as the phage display platform for oral cyclic peptides established by Kong et al.35–39 The late-stage cysteine-crosslinking method provides a convenient way for constructing various types of cyclic peptide library frameworks,40 but it still has some limitations. To ensure the efficiency of the reaction, highly reactive electrophiles, such as bromomethylbenzene derivatives, are often used in the late-stage cysteine-crosslinking method (Fig. 1). These reagents have been shown to affect the infectivity of phages and react with cysteine residues in the phage pIII protein.28,35,41–43 On the other hand, as reported by Dickinson et al., highly reactive electrophiles may also react with nucleic acid bases in mRNA-displayed peptide libraries.44–47 Chemoenzymatic methods are mild late-stage peptide modification strategies that avoid the use of highly reactive electrophiles, which is beneficial for constructing high-quality genetically encoded cyclic peptide libraries. We recently established a peptiligase-based system for the phage display of macrocyclic peptides.48–53 In this system, a low-reactivity chloroacetyl group was introduced into the peptide substrates of the peptiligases. Once the enzymatic ligation reaction occurs, the moiety of the chloroacetyl group reacts intramolecularly with cysteine residues to generate cyclic peptides. However, the types of cyclic peptide frameworks afforded by the chloroacetyl moiety are limited, which is not conducive to exploring more chemical structural spaces. In this study, we prepared a variety of low-reactivity electrophilic peptide substrates, tested their compatibility with the SrtA system, and generated various cyclic peptide frameworks. Combined with the phage display platform, we discovered several bicyclic peptides targeting human IgG1 Fc and TROP2, demonstrating the great potential of this technology in cyclic peptide ligand discovery. The various cyclic peptide frameworks generated using the peptiligase-based method in this study will contribute to the discovery of a variety of valuable cyclic peptide lead compounds. Late-stage cyclization of genetically encoded peptide libraries. (A) The previously reported chemical macrocyclization by cysteine crosslinking with bromomethylbenzyl-based tris-electrophilic reagents. (B) Chemoenzymatic macrocyclization of peptide libraries via sortase-based peptide cyclization by the use of chloroacetyl-based peptide substrates. (C) Sortase-mediated generation of various cyclic peptide frameworks by the use of chloromethylbenzyl-based or fluorinated peptide substrates (this study). Results and discussion Preparation of low-reactivity electrophilic peptide substrates from bis-electrophilic reagents Our study began with the design and synthesis of low-reactivity peptide substrates. In the SrtA recognition motif of LPXTG,54–56 X can be a non-natural amino acid bearing an electrophilic group. Our previous studies have limited the electrophilic group to the chloroacetyl group, which may have restricted the exploration of the structural diversity of cyclic peptides.48,49,51,52 Note that the structural diversity of cyclic peptides is crucial for the discovery of highly bioactive peptides. To facilitate the rapid construction of various cyclic peptide backbones, we employed commercially available thiol-reactive reagents, which are widely used for cysteine crosslinking.31 These bis-electrophilic reagents can provide a variety of cyclic peptide scaffolds, including the structures of phenyl, biphenyl, and their fluorine substituents. Introducing these structures into the X residue of the LPTXG motif allows for further testing of the non-natural promiscuity of SrtA, thereby obtaining a structurally diverse library of cyclic peptides. The bromomethylbenzyl group exhibits high reactivity, potentially leading to phage toxicity and undesirable side reactions.28,35,41–43 To mitigate these potential problems, we decided to prepare chloromethylbenzyl-containing peptides. In comparison with the electrophilicity of bromomethylbenzyl, the chloromethylbenzyl group has lower reactivity, thus avoiding undesirable side reactions with phages or mRNA. We employed a two-step strategy to prepare several peptide substrates for the SrtA system (Fig. 2). First, a short peptide 1 containing the LPCTGR motif was synthesized through Fmoc-based solid-phase synthesis. Then, peptide 1 was treated with 1,2-bis(bromomethyl)benzene to give the intermediate peptide 2a. In the presence of lithium chloride (LiCl),57 peptide 2a was successfully converted to the desired chloromethylbenzyl-containing peptide 3a. Following this method, we successfully prepared four other peptides: 3b, 3c, 3d, and 3e. Chemical synthesis of low-reactivity electrophilic peptide substrates. (A)–(E) synthetic routes, HPLC chromatograms and mass spectra of chloromethylbenzyl-based peptide substrates: 3a–3e. (F) Synthetic routes, HPLC chromatograms and mass spectra of perfluoroarene-based peptide substrates: 3f and 3g. (G) Synthetic routes, HPLC chromatograms and mass spectra of 3h. Reaction conditions of peptide 1 with bromomethylbenzyl-based reagents: 15.0 mM peptide 1 (1.0 equiv.), 2.25 mM bromomethylbenzyl-based reagent (1.5 equiv.), 45.0 mM N,N-diisopropylethylamine (DIPEA, 3.0 equiv.), DMF; reaction conditions of peptide 2a–2e with LiCl: 11.0 mM peptide 2a–2e (1.0 equiv.), 110.0 mM LiCl (10.0 equiv.), DMF/H2O (5/1, v/v); reaction conditions of peptide 1 with perfluoroarene: 3.6 mM peptide 1 (1.0 equiv.), 4.7 mM perfluorobiphenyl (1.3 equiv.) or 14.4 mM hexafluorobenzene (4.0 equiv.), 21.6 mM tris(hydroxymethyl)aminomethane (Tris, 6.6 equiv.). Perfluoroarene-containing peptide macrocycles are a unique class of peptides because fluorine atoms can endow them with unique pharmacological and chemical properties, such as enhanced cell membrane permeability and lipophilicity.58,59 Notably, Pentelute et al. found that perfluoroarene-crosslinked peptides can even cross the blood–brain barrier.60 Furthermore, fluorine substitution is widely used in medicinal chemistry, with hundreds of clinical therapeutic drug molecules containing one or more fluorine atoms. To this end, we designed and synthesized peptide substrates 3f and 3g. As shown in Fig. 2, peptide 3f contains a decafluorobiphenyl group, while peptide 3g contains a pentafluorophenyl group. 3f was obtained by directly treating peptide 1 (1.0 equiv.) with perfluorobiphenyl (1.3 equiv.) in DMF with the catalyst of Tris base. Unexpectedly, in the presence of 1.3 equivalents of hexafluorobenzene, we observed a byproduct consisting of one hexafluorobenzene and two peptide 1 molecules, likely due to the low solubility of hexafluorobenzene in DMF. To address this issue, we increased the amount of hexafluorobenzene from 1.3 equivalents to 4.0 equivalents. Fortunately, 3g was obtained in an ideal yield. Compared to monocyclic peptides, bicyclic peptides exhibit higher enzymatic stability due to their rigid structure. TBMB is a widely used tris-electrophilic reagent for the preparation of bicyclic peptides.28 However, previous studies have shown that TBMB significantly reduces phage infectivity at concentrations above 10 µM, and this side effect becomes more pronounced with prolonged incubation time. To address this issue, we designed and synthesized a peptide substrate 3h bearing two chloromethylbenzyl groups, which reduces phage toxicity. The synthesis of peptide 3h also employed a two-step method. First, peptide 1 underwent cysteine alkylation with excess TBMB (1.5 equivalents) to generate the intermediate peptide 2f. Then, in the presence of LiCl, peptide 2f underwent a bromine-to-chlorine substitution to generate the desired peptide 3h. Sortase-mediated peptide ligation and monocyclization using chloromethylbenzyl-containing peptide substrates We first tested the SrtA-mediated peptide ligation and one-pot cyclization reaction of peptide 3a and peptide 4. 40 µM of 3a and 25 µM of 4 were dissolved in a HEPES buffer (pH 7) containing 1.0 mM tris(2-carboxyethyl)phosphine, followed by the addition of 5 µM SrtA. After one hour, we observed the formation of the desired cyclic peptide product 7, but a significant amount of peptide 4 remained (SI, Fig. S18). The relatively slow reaction rate of 3a and 4 is likely due to the slow intramolecular reaction of the chloromethylbenzyl group with the cysteine residue. Fortunately, when the pH of the reaction buffer was increased from 7.0 to 8.0, the SrtA-mediated reaction of 3a and 4 was completed after one hour (SI, Fig. S19). In the absence of SrtA, peptide 3a did not react with peptide 4, indicating that cyclic peptide product 7 was indeed produced under SrtA catalysis. Unexpectedly, peptide 3a underwent rapid degradation (SI, Fig. S19), likely due to its reaction with tris(2-carboxyethyl)phosphine (TCEP). As expected, peptide 3a no longer degraded when TCEP was not added (SI, Fig. S20). The TCEP-induced 3a degradation is detrimental to the construction of the cyclic peptide library as it could rapidly consume the electrophilic peptide substrate. When the reaction temperature was lowered to 15 °C or 4 °C, the TCEP-induced degradation of peptide 3a remained significant. Fortunately, at a TCEP concentration of 100 µM, 3a hardly reacted with TCEP (Fig. 3D). 100 µM TCEP ensured that the cysteine residues in the peptide chain were in a reduced state, which is beneficial for the intramolecular cyclization of chloromethylphenyl and cysteine residues. SrtA-mediated peptide ligation and one-pot cyclization (SPLOC) using low-reactivity peptide substrates. (A) SrtA-mediated ligation and cyclization of peptide 3a with N-terminal glycine peptide 4 to form the desired cyclic peptide product 7. (B) SrtA-mediated ligation and cyclization of peptide 3f with N-terminal glycine peptide 4 to form the desired cyclic peptide product 9. (C) SrtA-mediated ligation and cyclization of peptide 3h with N-terminal glycine peptide 5 to form the desired bicyclic peptide product 11. (D) HPLC chromatograms (210 nm) of the SrtA-mediated reaction of peptide 3a and peptide 4. (E) HPLC chromatograms (210 nm) of the SrtA-mediated reaction of peptide 3f and peptide 4 (* is from the hydrolysis of 3f by SrtA). (F) HPLC chromatograms (210 nm) of the SrtA-mediated reaction of peptide 3h and peptide 5. Conditions 1#: 4 (25 µM), 3a (40 µM) SrtA (5 µM), pH 8.0, TCEP (100 µM), 37 °C; conditions 2#: 4 (25 µM), 3a (40 µM) SrtA (0 µM), pH 8.0, TCEP (100 µM), 37 °C; conditions 3#: 4 (25 µM), 3f (40 µM) SrtA (5 µM), pH 8.0, TCEP (100 µM), 37 °C; conditions 4#: 4 (25 µM), 3f (40 µM) SrtA (5 µM), pH 9.0, TCEP (100 µM), 37 °C; conditions 5#: 4 (25 µM), 3f (250 µM) SrtA (5 µM), pH 9.0, TCEP (100 µM), glutathione (1 mM), 37 °C (peaks 3f and * were truncated to allow better visualization of the desired product 9); conditions 6#: 5 (25 µM), 3h (40 µM) SrtA (5 µM), pH 7.0, TCEP (100 µM), 37 °C; conditions 7#: 5 (25 µM), 3h (40 µM) SrtA (5 µM), pH 8.0, TCEP (100 µM), 37 °C. We then tested SrtA-mediated peptide ligation and one-pot cyclization of 3b with peptide 4. In the presence of 100 µM TCEP and 5 µM SrtA, 3b reacted rapidly with 4 to generate the desired cyclic peptide product (SI, Fig. S24). As expected, peptide 3d also reacted with peptide 4 under the same conditions to generate the desired cyclic peptide product (SI, Fig. S27). We also tested SrtA-mediated peptide ligation and one-pot cyclization of 3c with peptide 4. Compared to 3a, 3c was more sensitive to TCEP: even at a concentration of 100 µM TCEP, 3c was still degraded by TCEP (SI, Fig. S25). To solve this problem, we increased the concentration of SrtA from 5 µM to 20 µM. Encouragingly, the reaction of 3c with peptide 4 was completed within 10 minutes, during which almost no degradation of 3c was observed (SI, Fig. S26). We finally tested SrtA-mediated peptide ligation and one-pot cyclization of 3e with peptide 4. Notably, the biphenyl group in peptide 3e enhances the structural rigidity of the resulting cyclic peptide.61,62 As expected, in the presence of 20 µM SrtA, 3e reacted rapidly with peptide 4 to generate the desired cyclic peptide product (SI, Fig. S28). These results demonstrate that the SrtA system is compatible with many non-natural amino acid residues at the X site of the LPXTG motif. It is worth mentioning that the combination other types of commercially available bis-electrophilic reagents with the SrtA system holds promise for providing more cyclic peptide scaffolds. Sortase-mediated peptide ligation and monocyclization using perfluoroarene-containing peptide substrates We first tested SrtA-mediated peptide ligation and one-pot cyclization of perfluorobiphenyl-containing peptide 3f and peptide 4. 3f and 4 were dissolved in HEPES buffer (pH8.0), and SrtA (25 µM) was added. After incubation at 37 °C for one hour, we observed a little amount of the desired cyclic peptide product 9, although the linear ligation intermediate 8 was detected (Fig. 3E). Note that most of peptide 3f underwent hydrolysis under SrtA catalysis, suggesting that the low yield of the desired 9 may be due to the slow intramolecularly aromatic nucleophilic substitution (SNAr) between the cysteine residue and the fluorine atom in the perfluorobiphenyl moiety.63 In our previous studies,48 the rapid intramolecular reaction between the chloroacetyl group and the cysteine residue significantly promoted sortase-mediated peptide ligation and cyclization. Based on this, we hypothesized that enhancing the SNAr reaction of the cysteine residue and the perfluoroaryl group would favor the SrtA-mediated peptide ligation and one-pot cyclization of 3f and 4. We increased the concentration of peptide 4 from 40 µM to 500 µM. An excess of 4 should promote SrtA-mediated peptide ligation and suppress SrtA-mediated hydrolysis of the ligation intermediate 8. After incubation at 37 °C for 30 min, most of peptide 4 disappeared, generating the linear ligation intermediate 8, but the desired product 9 remained negligible (SI, Fig. S30). On the other hand, the pH of the reaction solution was increased from 8.0 to 9.0. As shown in Fig. 3E, we observed the formation of the desired 9, but the yield was not high. Considering that the perfluorobiphenyl-based crosslinking of cysteine-containing peptides is usually performed with Tris base, we replaced the HEPES buffer with Tris buffer. Disappointingly, the Tris buffer did not promote the intramolecular SNAr reaction of the thiol group with the perfluorobiphenyl moiety (SI, Fig. S32). At pH 9.0, cysteine residues tend to be oxidized by air, so we added glutathione (1.0 mM) to the system. Encouragingly, the yield of the desired cyclic peptide 9 increased significantly (Fig. 3E). Under similar conditions, peptide 3g also reacted smoothly with peptide 4 to generate the desired product (SI, Fig. S34). Decafluorobiphenyl-crosslinked peptides are a unique class of molecules with the potential to cross the blood–brain barrier. Derda et al. introduced decafluoro-diphenylsulfone into phage display, but decafluorobiphenyl is not suitable for late-stage cyclization modification of phage-displayed peptides because its intermolecular reaction with cysteine-containing peptide is quite slow.63,64 In our SrtA system, the SNAr reaction between the cysteine residue and perfluorobiphenyl occurs intramolecularly, which can expand the framework of genetically encoded fluorinated cyclic peptide libraries. Sortase-mediated peptide ligation and bicyclization of the TBMB-derived peptide substrate Bicyclic peptides exhibit enhanced protease stability and target binding affinity. In particular, the TBMB-based method proposed by Heinis et al. has been widely used for constructing bicyclic peptide libraries. We performed SrtA-mediated peptide ligation and one-pot bicyclization of peptide 3h and peptide 5. At pH 7.0, the yield of the SrtA-mediated bicyclization of 3h and 5 was approximately 50% (Fig. 3F), likely due to the lower reactivity of chloromethylbenzyl with cysteine. Upon increasing the pH to 8.0, we observed complete conversion of 3h and 5 to the desired bicyclic peptide product 11 (Fig. 3F). Note that, at pH8, we observed the side reaction of peptide 3h with TCEP (0.1 mM). When the SrtA concentration increased from 5 µM to 20 µM, the bicyclization of 3h and 5 was completed within 30 minutes (SI, Fig. S37). In the absence of SrtA, we did not observe the formation of the desired bicyclic peptide product 11 and the TCEP-induced 3h degradation (SI, Fig. S36), indicating that the formation of 11 is indeed catalyzed by SrtA. In traditional TBMB-based bicyclization reactions, TBMB exhibits high reactivity and causes nonspecific side reactions and phage toxicity. In our SrtA-based system, the chloromethylbenzyl group in peptide 3h exhibits low reactivity and does not react with the cysteine-containing peptides in the absence of SrtA, thus avoiding the phage toxicity and undesired side reactions. All optimized reaction conditions are summarized in Table 1. Table 1Optimized conditions for SrtA-mediated peptide ligation and one-pot cyclizationEntryPepa,b,c[SrtA]/µMTCEP/µMdpHTime/minYield/%13a/451008608423b/451008609033c/4201008309043d/451008608753e/420f1008309363f/4e51009180g6073g/4e20h10091804083h/520f10083089a “Pep” refers to the two peptide substrates used in the reaction.bThe concentrations of 3a–3e and 3h were 40 µM, whereas those of 3f and 3g were 250 µM.c4 and 5 were used at a concentration of 25 µM.dHigh concentrations of TCEP led to the degradation of chloromethylbenzyl-based electrophilic peptides.eGlutathione (1.0 mM) was added to the mixture to prevent oxidation of 4 or 5 at pH 9.fA higher concentration of SrtA was used to accelerate the reaction to prevent TCEP-induced degradation of electrophilic peptides.gDue to the slow intramolecular SNAr reaction between the thiol group and the perfluorobiphenyl moiety, the reaction time was extended to 180 min.hSince 3g exhibits lower activity than 3f, 20 µM SrtA was required to facilitate the enzymatic ligation reaction. Optimized conditions for SrtA-mediated peptide ligation and one-pot cyclization “Pep” refers to the two peptide substrates used in the reaction. The concentrations of 3a–3e and 3h were 40 µM, whereas those of 3f and 3g were 250 µM. 4 and 5 were used at a concentration of 25 µM. High concentrations of TCEP led to the degradation of chloromethylbenzyl-based electrophilic peptides. Glutathione (1.0 mM) was added to the mixture to prevent oxidation of 4 or 5 at pH 9. A higher concentration of SrtA was used to accelerate the reaction to prevent TCEP-induced degradation of electrophilic peptides. Due to the slow intramolecular SNAr reaction between the thiol group and the perfluorobiphenyl moiety, the reaction time was extended to 180 min. Since 3g exhibits lower activity than 3f, 20 µM SrtA was required to facilitate the enzymatic ligation reaction. Selection of peptide macrocycles targeting IgG1 Fc and TROP2 To verify the practicality of this method, we used the TBMB-derived peptide substrate 3h to generate bicyclic peptides via SrtA-mediated phage library bicyclization and screened for bicyclic peptide ligands of target proteins. We first selected commercially available human IgG1 Fc protein as the target protein for phage selection. A peptide library in the form of GXmCXnC (m + n = 9, m = 3, 4, 5 or 6) was fused to the N-terminus of the M13 pIII protein, and 7 × 108 transformed colonies were obtained after plasmid electroporation. The SrtA-mediated phage modification was performed under conditions similar to those summarized in Table 1. Briefly, phage particles (∼2 × 1012 pfu) were incubated in TBS buffer (pH 8.0) containing 1 mM TCEP for 20 min, followed by a 10-fold dilution with HEPES buffer (pH 8.0, 2.0 mM CaCl2). SrtA (20 µM) and 3h (50 µM) were then added to initiate the chemo–enzymatic reaction, and the mixture was incubated at 37 °C for 1 h. Biotinylated IgG1 Fc was immobilized on magnetic beads and then co-incubated with the phage-displayed bicyclic peptide library. After four rounds of selection, we observed a 50-fold increase in phage recovery. Compared to magnetic beads coated only with avidin, IgG1 Fc-coated beads captured nearly 100-fold more phages, indicating that phage enrichment specifically targets IgG1 Fc. We randomly selected 50 phage clones for sequencing. Sequencing analysis showed that the enriched peptide sequence belongs to the format of GX4CX5C. We synthesized the enriched bicyclic peptide 12 (Fig. 4F) and determined its affinity for IgG1 Fc to be 8.95 µM (SI, Fig. S70), using surface plasmon resonance (SPR). Note that the bicyclic peptide 12 differs significantly from the previously reported IgG1 Fc-binding peptides, suggesting that its binding epitope may differ from that of previously reported IgG1 Fc-binding peptides. We believe that our bicyclic peptide holds promise as a potential tool for ligand-directed IgG1 conjugation.65,66 Phage screening for macrocyclic peptides targeting human IgG1 Fc and TROP2. (A) Schematic diagram of phage display of macrocyclic peptides (SPLOC: SrtA-mediated peptide ligation and one-pot cyclization; SAV: streptavidin-coated magnetic beads). (B) Construction of a phage displayed bicyclic peptide library targeting human IgG1 Fc. (C) Construction of a phage displayed bicyclic peptide library targeting TROP2. (D) Phage titers after 4 rounds of screening for human IgG1 Fc (R4′ is the negative selection with avidin-coated beads). (E) Phage titers after 4 rounds of screening for TROP2 (R4′ is the negative selection with avidin-coated beads). (F) Chemical structure of bicyclic peptide 12. (G) Chemical structure of bicyclic peptide 13. (H) Determination of binding affinity of peptide 13 to TROP2 by SPR. We further validated the practical application of this method using TROP2 as the target protein. A peptide library in GX5CX4C format was fused with the pIII protein of the M13 phage plasmid, and 1 × 109 transformed colonies were obtained by electroporation. The packaged phages were then treated with SrtA (20 µM) and peptide 3h (50 µM) to convert them into a bicyclic peptide library. Similar to the protein of IgG1 Fc, biotinylated TROP2 was immobilized on magnetic beads and then co-incubated with the phage-displayed bicyclic peptide library. Pleasingly, after four rounds of selection, the number of captured phages increased by 6250-fold. Compared to avidin-coated beads, TROP2-coated beads captured 1000-fold more phage particles, indicating significant enrichment of phages targeting TROP2. Sequencing of 40 randomly selected phage clones revealed three enriched bicyclic peptides. All peptides were chemically synthesized, and one peptide showed a binding affinity of 6.01 µM for TROP2 as determined by SPR assay (SI, Fig. S71). Given that all three enriched peptides contain the GWLP motif at their N-terminus, we constructed a secondary phage library containing a peptide form of GWLPX2CX4C. After three rounds of phage screening, we obtained 11 bicyclic peptides. We chemically synthesized these 11 bicyclic peptides and determined their TROP2 binding affinity. One of the bicyclic peptides, 13, showed a binding affinity of 1.77 µM (Fig. 4G). All together, these results demonstrate the potential of our method for discovering cyclic peptide ligands for target proteins. Conclusions To summarize, we report a strategy for preparing a variety of macrocyclic peptide frameworks from low-reactivity electrophilic substrates using sortase-mediated peptide ligation and one-pot cyclization (SPLOC). The electrophilic group used in our previous sortase-based platform was limited to the chloroacetyl group, which has restricted the exploration of chemical spaces related to cyclic peptides. In this study, we prepared a variety of chloromethylbenzyl-based electrophilic peptide substrates using commercially available bis-electrophilic reagents and demonstrated that SrtA can efficiently recognize these substrates, generating cyclic peptides including monocyclic and bicyclic peptides. Furthermore, we tested the compatibility of fluoroarene-containing peptide substrates with SrtA and successfully prepared perfluorinated cyclic peptides. Notably, these fluorinated cyclic peptides are monocyclic peptides crosslinked by hexafluorobenzene or perfluorobiphenyl. Due to the low reactivity of perfluoroarene, it is difficult to perform cysteine crosslinking of phage-displayed peptide libraries with fluorobenzene or perfluorobiphenyl. Different from the bromomethylbenzyl-based electrophiles, the chloromethylbenzyl-based electrophiles used here do not react with cysteine-containing peptides in the absence of sortase, thus avoiding phage toxicity and undesirable side reactions. Combined with phage display technology, we constructed phage-displayed bicyclic peptide libraries and screened several cyclic peptide ligands targeting human IgG1 Fc or TROP2. Note that, considering the advantages of bicyclic peptide in terms of enzymatic stability, here we have only verified the SrtA-mediated phage library bicyclization by using the TBMB-derived peptide substrate. The SrtA-mediated phage library monocyclization by the use of other peptide substrates still requires further verification in the future. It is anticipated that the compatibility of various low-reactivity electrophilic peptide substrates with the SrtA-based platform will further expand the application of ligase-mediated peptide ligation and one-pot cyclization in constructing genetically encoded macrocyclic peptide libraries. Author contributions The idea of the study was designed by G.-M. Fang. This sortase-based phage display platform was established by Z.-M. Xia and H.-M. Wei. The low-reactivity electrophilic substrates were synthesized by Z.-M. Xia, H.-M. Wei, J.-H. Xue, Y.-N. Sun and Y. Hou. The sortase-mediated peptide ligation and one-pot cyclization were performed by Z.-M. Xia and H.-M. Wei. The binding affinity of the peptides was measured by Z.-M. Xia and H.-M. Wei. The manuscript was written and checked by Z.-M. Xia, H.-M. Wei, R.-H. Tong, P. Shi, L. Zhou, G.-M. Fang and C. Tian. Conflicts of interest The authors declare there to be no conflicts of interest. Supplementary Material