In vitro selection of macrocyclic peptide inhibitors containing cyclic γ2,4-amino acids targeting the SARS-CoV-2 main protease

γ-Amino acids can play important roles in the biological activities of natural products; however, the ribosomal incorporation of γ-amino acids into peptides is challenging. Here we report how a selection campaign employing a non-canonical peptide library containing cyclic γ2,4-amino acids resulted in the discovery of very potent inhibitors of the SARS-CoV-2 main protease (Mpro). Two kinds of cyclic γ2,4-amino acids, cis-3-aminocyclobutane carboxylic acid (γ1) and (1R,3S)-3-aminocyclopentane carboxylic acid (γ2), were ribosomally introduced into a library of thioether-macrocyclic peptides. One resultant potent Mpro inhibitor (half-maximal inhibitory concentration = 50 nM), GM4, comprising 13 residues with γ1 at the fourth position, manifests a 5.2 nM dissociation constant. An Mpro:GM4 complex crystal structure reveals the intact inhibitor spans the substrate binding cleft. The γ1 interacts with the S1′ catalytic subsite and contributes to a 12-fold increase in proteolytic stability compared to its alanine-substituted variant. Knowledge of interactions between GM4 and Mpro enabled production of a variant with a 5-fold increase in potency.

γ-Amino acids can play important roles in the biological activities of natural products; however, the ribosomal incorporation of γ-amino acids into peptides is challenging. Here we report how a selection campaign employing a non-canonical peptide library containing cyclic γ 2,4 -amino acids resulted in the discovery of very potent inhibitors of the SARS-CoV-2 main protease (M pro ). Two kinds of cyclic γ 2,4 -amino acids, cis-3-aminocyclobutane carboxylic acid (γ 1 ) and (1R,3S)-3-aminocyclopentane carboxylic acid (γ 2 ), were ribosomally introduced into a library of thioether-macrocyclic peptides. One resultant potent M pro inhibitor (half-maximal i nh ib it ory c on ce nt ration = 50 nM), GM4, comprising 13 residues with γ 1 at the fourth position, manifests a 5.2 nM dissociation constant. An M pro :GM4 complex crystal structure reveals the intact inhibitor spans the substrate binding cleft. The γ 1 interacts with the S1′ catalytic subsite and contributes to a 12-fold increase in proteolytic stability compared to its alanine-substituted variant. Knowledge of interactions between GM4 and M pro enabled production of a variant with a 5-fold increase in potency. γ-Amino acid residues can induce unique peptide conformations, such as C 14 -helix, C 12/10 -helix and C 12 -turn secondary structures [1][2][3][4][5][6][7][8][9][10] . The γ-amino acids present in natural products play important roles in their biological activities and enable improved proteolytic stability and cell permeability 11,12 . Pepstatin, isolated from Actinomyces, has two γ-amino acid residues, which contribute to its protease inhibitory activity by enhancing binding to the active sites [13][14][15] . Didemnin, an antiviral/antitumor compound, has a γ-amino acid and a macrocyclic scaffold 16 . Given their remarkable bioactivities and conformational rigidity, macrocyclic peptides containing γ-amino acids are an attractive option for incorporation into novel therapeutic peptides.

Selection of macrocyclic peptide binders to SARS-CoV-2 M pro
To construct a cγAA-containing macrocyclic peptide library, γ 1 and γ 2 ( Fig. 1a) were assigned to GAG and GUG codons using tRNA Pro1E2 CUC and tRNA Pro1E2 CAC , respectively (Fig. 1b), employing the FIT system. Two cyclic β 2,3 -amino acids (cβAAs), (1R,2S)-2-aminocyclopentane carboxylic acid (β 1 ) and (1S,2S)-2-aminocyclopentane carboxylic acid (β 2 ), were introduced at the GUU and UGU codons using tRNA Pro1E2 GAC and tRNA GluE2 GCA , respectively, because we considered that the incorporation not only of cγAA, but also of cβAA, would give more diverse folding possibilities 25 . For the macrocyclization of the library via thioether bonds, we introduced N-chloroacetyl-d-tyrosine ( ClAc y) and d-cysteine (c) at the initiator AUG codon using tRNA fMet CAU and elongator AUG codon using tRNA Pro1E2 CAU , respectively. The thioether bond spontaneously forms between the N-terminal chloroacetyl group on ClAc y and the thiol group of the c residue (Fig. 1c). These amino acids were precharged onto the respective tRNAs using flexizymes. The peptide library comprised a repeat of 11−14 random residues encoded by GWG and NNU codons (W = A or U; N = A, U, G or C) flanked by the cyclizing ClAc y and c residues. Since multiple cγAA incorporations could not be efficiently achieved, as reported in our previous study 29 , the library was designed to have only a single γ 1 or γ 2 appear at the GWG codon. The two cβAAs and the thirteen proteinogenic l-α-amino acids (A, D, F, G, H, I, L, N, P, R, S, T and Y) were assigned at NNU codons. The C-terminal SGGSGG sequence following peptidyl transfer of the P-site peptidyl-tRNA onto the γ-amino group of the A-site γ-aminoacyl-tRNA due to poor compatibility with the peptidyl transferase centre of the ribosome 27,28 . We recently reported that the self-deacylation issue can be circumvented by using structurally constrained cyclic γ 2,4 -amino acids (cγAAs) 29 . The cγAAs were charged on an engineered tRNA, referred to as tRNA Pro1E2 (Supplementary Fig. 1), the T-stem and D-arm of which were optimized for efficient binding to elongation factor Tu (EF-Tu) and EF-P, thereby accelerating accommodation and peptidyl transfer, respectively [30][31][32] . Combination of the flexizyme-assisted 33,34 synthesis of cγAA-tRNA Pro1E2 and the reconstituted Escherichia coli translation system, referred to as the Flexible In vitro Translation (FIT) system 23 , that contains EF-Tu and EF-P with their adjusted concentrations enabled the incorporation of cγAAs into nascent peptide chains 29,35,36 .
Here we report the ribosomal synthesis of a macrocyclic peptide library containing cγAAs and its application to an in vitro display, referred to as the Random nonstandard Peptides Integrated Discovery (RaPID) system 23 . The library contains two types of cγAAs, cis-3-aminocyclobutane carboxylic acid (γ 1 ) and (1R,3S)-3-aminocyclopentane carboxylic acid (γ 2 ) (Fig. 1a). Owing to the constrained rigid conformations of these cγAAs, they can be exquisite building blocks for preorganization of peptide secondary structures. Therefore, screening of macrocyclic peptide libraries containing cγAAs can lead to molecules with potent binding affinities and peptidase resistance. As a biomedicinally important test target protein, we chose M pro , which plays an essential role in processing the polyproteins translated from the viral RNA 37   Article https://doi.org/10.1038/s41557-023-01205-1 the c residue was a linker peptide connected to the 3′ end of mRNA via a puromycin linker. To demonstrate the fidelity of translation, model macrocyclic peptides containing any of cγAA and cβAA (γ 1 , γ 2 , β 1 and β 2 assigned at GAG, GUG, GUU and UGU codons, respectively) were synthesized, and their identities were confirmed by matrix-assisted laser desorption/ionization coupled with time of flight mass spectrometry (MALDI-TOF MS; Supplementary Fig. 2). The macrocyclic peptide library was then applied to the RaPID selection against recombinant SARS-CoV-2 M pro . Translation of the random mRNA library into the peptide library, conjugation of the peptide with the parent mRNA via a puromycin linker and reverse transcription of the mRNA into the cDNA yielded peptide/mRNA/ cDNA conjugates (Extended Data Fig. 1a). The library was first subjected to naked magnetic bead treatment to remove bead binders, and then applied to M pro -immobilized magnetic beads to recover M pro binders. The cDNA moiety of the recovered peptide/mRNA/cDNA conjugate was amplified by the polymerase chain reaction (PCR) and transcribed into the mRNA library for the next selection round (details in Methods). By repeating this affinity selection, the recovery rate of the M pro binders was greatly increased at the third round of selection and, even more obviously, in the fourth round, while the recovery of the bead binders did not increase (Extended Data Fig. 1b). Deep sequencing of the cDNA library at the fourth round revealed that the library was enriched with several families of peptides bearing cγAA (Supplementary Table 1 shows the top 100 sequences). Among these, seven macrocyclic peptides containing cγAA were selected for further analysis of their binding affinities, inhibitory activity and proteolytic stability ( Table 1, GM1−GM7). GM1, GM2, GM4 and GM6 contain a γ 1 residue in their sequences, while GM3, GM5 and GM7 contain γ 2 . Note that the cβAA-containing peptides were not selected because they were not included in the major peptide families with high read numbers. GM1−GM7 were chemically synthesized on a large scale using the standard solid-phase method without the C-terminal SGGSGG linker, and their purities and identities were confirmed by ultra-performance liquid chromatography (UPLC) and MALDI-TOF MS, respectively (Supplementary Figs. 3 and 4).

Biological activities and stabilities of peptide inhibitors
We first evaluated the binding affinity of GM1−GM7 to M pro by surface plasmon resonance (SPR). All the peptides exhibited low-to-moderate nanomolar kinetic equilibrium (K D ) values (Table 1 and Supplementary Fig. 5). Notably, GM1 and GM4, which contain γ 1 , and GM5, which contains γ 2 , have a conserved four residue motif, yFHγ X (γ X = γ 1 or γ 2 ), at their N termini and showed potent affinities, with K D values of 2.3, 5.2 and 5.2 nM, respectively; GM2 has the same motif, but has a higher K D (71 nM). GM6 and GM7, which had relatively low read frequencies, exhibited weak binding compared to GM1−GM5, indicating selection for tight binding peptides. To evaluate contributions of the cγAA residues to potency, we synthesized peptides where the cγAA was substituted with alanine ( Table 1, GM1γ 1 4A, GM3γ 2 10A, GM4γ 1 4A and GM5γ 2 4A). The thioether-macrocyclic peptides selected by In addition to GM1, GM4 and GM5, which share the conserved yFHγ X motif, GM3 was selected for alanine substitution because of its high read percentage among peptides without the yFHγ X motif. SPR measurements of the alanine mutants revealed that the K D values of GM1γ 1 4A, GM4γ 1 4A and GM5γ 2 4A were comparable or one order higher than those of the original peptides, indicating that the cγAA residue in the conserved yFHγ X motif is important for binding. To our surprise, GM3γ 2 10A exhibited an eightfold stronger binding affinity (K D = 7.5 nM), revealing that the contribution of cγAA to the binding affinity is sequence context dependent. We next evaluated the inhibitory activity of GM1−GM7 against the hydrolytic activity of SARS-CoV-2 M pro using a reported MS-based method 39 . GM1−GM7 exhibited inhibition of M pro ; in particular, the yFHγ X motif containing GM1, GM4 and GM5, which manifest single-digit nanomolar K D values, showed particularly potent inhibition, with half-maximal inhibitory concentration (IC 50 ) values of 40, 50 and 40 nM, respectively (Table 1, Fig. 2a and Extended Data Fig. 2). The peptides with weaker binding affinity, that is, GM2, GM3, GM6 and GM7, had IC 50 values of 1,750-7,930 nM, implying correlation between IC 50 and K D values. The inhibitory activities of the alanine mutants were then determined. Notably, despite GM1γ 1 4A having only a fourfold weaker binding affinity compared with GM1, inhibition was ablated, revealing the importance of the cγAA. By contrast, GM3γ 2 10A, GM4γ 1 4A and GM5γ 2 4A retained inhibition activity, showing the context-dependent effects of the cγAA (IC 50 = 120, 10 and 50 nM, respectively; Table 1, Fig. 2a and Extended Data Fig. 2).
We evaluated the half-life of potent cγAA-containing peptides and their mutants in human serum, because in vivo stability is a critical factor in the development of therapeutic peptides. Each peptide and an uncleavable internal standard peptide were co-incubated in human serum at 37 °C, and the relative amount of the remaining sample peptide was estimated by liquid chromatography/MS (LC/MS). The potent inhibitors GM1 and GM4, containing γ 1 , and GM5, containing γ 2 , exhibit high peptidase resistance with half-lives (t 1/2 ) of 90 h, 126 h and 32 h, respectively (Table 1, Fig. 2b and Extended Data Fig. 3a). Importantly, by contrast, their alanine mutants show substantially shorter half-lives (t 1/2 = 8.3 h, 11 h and 21 h for GM1γ 1 4A, GM4γ 1 4A and GM5γ 2 4A, respectively; Table 1). The enhancement of serum resistance observed for GM4, for instance, was 12-fold. Notably, despite the substitution of γ 1 to an Ala residue not diminishing inhibition relative to GM4, the enhancement in serum stability by the γ 1 residue is substantial. Moreover, the less potent inhibitors GM2 and GM3 have more than nearly three orders of magnitude shorter half-lives (1.1 h and 2.5 h, respectively) compared to GM4, suggesting that the higher (binding and inhibitory) activity of GM4 is reflected in its higher serum stability, possibly due to its more conformationally constrained fold.
To identify peptidase cleavage sites in serum, the products of GM4 and GM4γ 1 4A after 24 h incubation were analysed by LC/MS. In the case of GM4, six fragments (GM4-f1−GM4-f6, Extended Data Fig. 3b) containing the c-(thioether)-Ac yFHγ 1 L motif were detected, but no fragmentation of the motif itself was apparent. With GM4γ 1 4A, three shorter fragments containing c-(thioether)-Ac yF were detected (GM4γ 1 4A-f1−GM4γ 1 4A-f3; Extended Data Fig. 3c), indicating cleavage between F and H of the motif. These results show that the presence of non-canonical residues (one cγAA (γ 1 ), two d-amino acids and a thioether bond) can enable high peptidase resistance by preventing the proteolysis of flanking residues, leading to a remarkable improvement in serum stability.

X-ray crystallographic analysis of the GM4:M pro complex
To gain structural insights into how GM4, including its γ 1 residue, interacts with M pro , we obtained an X-ray crystal structure of their complex to 1.7 Å resolution ( Fig. 3a and Supplementary Table 2). The structure reveals GM4 bound with high occupancy and in the same manner at the active site regions of both monomers in the M pro dimer. Excellent electron density was observed for all residues in the GM4 macrocycle (Fig. 3c). The overall conformations of M pro in the GM4 and nirmatrelvir 38 (used for comparison) complex structures are very similar (backbone root mean square deviation, 0.34). The four residues in the yFHγ 1 motif of GM4 occupy the S4, S2, S1 and S1′ substrate binding subsites, respectively 40 (Fig. 3d). H3 GM4 is accommodated in the S1 subsite, forming hydrogen bonds with M pro C145, H164 and H163 (GM4 residues are identified by a subscript ;   Fig. 3e,f). The F2 GM4 residue binds in the S2 subsite, making hydrophobic interactions with M49, M165 and H41. The y1 GM4 side chain forms hydrogen bonds with T190 and Q192 in the S4 subsite (Extended Data Fig. 4a).
The c13 GM4 -thioether link, which enables macrocyclization, and the adjacent G12 GM4 enable a turn connected to P11 GM4 , R10 GM4 and Y9 GM4 ; none of these residues are positioned to make direct contacts with M pro in the structure, though these cannot be ruled out in solution. The backbone of GM4 forms three intramolecular hydrogen bonds, that is between γ 1 4 GM4 and L7 GM4 , Ac y1 GM4 and R10 GM4 , and P11 GM4 and The H41 and C145 catalytic dyad is in orange. γ 1 4 GM4 (magenta) occupies the S1′ pocket; the side chains of H3 GM4 , F2 GM4 and Ac y1 GM4 occupy the S1, S2 and S4 pockets, respectively. e,f, Close-ups of the S1 pocket, showing the H3 GM4 backbone carbonyl in the oxyanion hole (C145, G143 backbone amides). The backbone NH of H3 GM4 is positioned to interact with the H164 backbone CO.
The H3 GM4 imidazole is positioned to hydrogen bond with the H163 side chain and to interact with the E166 side chain, which interacts with S1 of protomer B, as observed in apo-M pro structures. Y9 GM4 is 3.1 Å and 3.3 Å from the E166 and S1 B side chains, respectively. g, Nirmatrelvir with the residues occupying the S1−S4 subsites labelled P1−P4 (the reactive nitrile is indicated with an arrow  Fig. 4b). The side chains of P11 GM4 and Y9 GM4 are arranged such that they 'stack' with each other, an arrangement positioning the phenolic side chain of Y9 GM4 adjacent to the imidazole ring of H3 GM4 , appearing to lock it into the S1 pocket; the phenolic side chain of Y9 GM4 also forms interactions with Ser1 , from the other protomer of the M pro dimer (Fig. 3f). G8 GM4, L7 GM4, N6 GM4 and L5 GM4 form a loop leading to γ 1 4, which is linked to H3 GM4. γ 1 4 GM4 is positioned to make hydrophobic contact with T24 and T25; the side chains of L5 GM4 and N6 GM4 are positioned close to the surface of M pro , with L7 GM4 projecting towards H41, C44 and T45. The observation of the unnatural γ 1 4 GM4 residue, the cyclobutane ring of which was refined in a near planar conformation, in the P1′ position is striking and implies it contributes to the inhibition or lack of efficient GM4 hydrolysis.
At the active site region, the sulfur of the nucleophilic C145 is positioned close (3.4 Å) to the imidazole ε-nitrogen of H41, which acts as a general base/acid in M pro catalysis, and to the carbonyl carbon (4.4 Å) of the amide linking H3 GM4 and γ 1 4 GM4 ; this carbonyl carbon is positioned in a similar manner to the imine (anion) derived by reaction of the nirmatrelvir nitrile with C145 (Fig. 3g,h).
Like the binding modes of substrates (predicted) and inhibitors (as observed by crystallography), the carbonyl of the amide linking H3 GM4 and γ 1 4 GM4 is positioned in the oxyanion hole, making hydrogen bonds with the C145 NH (2.9 Å) and with the G143 NH (2.9 Å). The ~100° angle between the C145 sulfur and the carbon and oxygen of the amide linking H3 and γ 1 4 is close to that predicted for nucleophilic attack onto an amide carbonyl as observed in studies on serine proteases 41 . Despite this apparently catalytically productive arrangement, analysis of the electron density map implies a lack of substantial covalent reaction, consistent with solution studies showing a lack of efficient GM4 hydrolysis by M pro . Lack of reaction appears to be a consequence of multiple interactions made between the peptide macrocycle and M pro , likely involving P11 GM4 , Y9 GM4 and H3 GM4 , resulting in tight binding in a catalytically non-productive manner.
SARS-CoV-2 M pro cleaves after a conserved glutamine residue (P1) in substrates with the following sequence motif: (P2, L/F/V/M (hydrophobic))(P1, Q)↓(P1′, S/A/G/N (small)), where ↓ represents the cleavage site 40,42 . Interestingly, rather than a glutamine (Q), GM4 has a histidine (H3) at the analogous 'P1' position. To investigate the contribution of the H3/P1 residue for activity, we evaluated peptides bearing Q, A or E residues at the analogous position. Among these, GM4H3Q exhibited a sixfold stronger binding affinity (K D = 0.86 nM) and fivefold more potent inhibitory activity (IC 50 = 10 nM; Table 1, Fig. 2 and Supplementary Fig. 5). By contrast, GM4H3A manifested no inhibition, and inhibition of GM4H3E was observed only with the highest tested concentration (25 µM). These results are consistent with the known substrate selectivity of M pro at the P1 position 40 .
In the case of the P1′ residue, M pro prefers substrates with small residues, such as G, A and S, although N is also tolerated 40 . Strikingly, GM4 has γ 1 4 at the P1′ position, likely reflecting the compact size of the γ 1 four-membered-ring main-chain residue. In support of this, GM4γ 1 4A (γ 1 → A) exhibited a comparable inhibitory activity to that of GM4 (IC 50 of GM4γ 1 4A = 10 nM), and GM4γ 1 4N was 20-fold less active (IC 50 = 1,000 nM). Despite the almost equivalent M pro inhibition activities of GM4 and GM4γ 1 4A, the serum stability of GM4 is 12-fold higher (t 1/2 = 126 h) than that of GM4γ 1 4A (Fig. 2b). Furthermore, the most potent mutant GM4H3Q bearing γ 1 4 at P1 also exhibited a comparable serum stability (t 1/2 = 82 h). These results show how the introduction of a nonstandard amino acid at the P1′ position can prolong the serum stability to non-targeted protease-catalysed hydrolysis while maintaining potency against the targeted protease (M pro in our case).

Discussion
Our results reveal GM4 and GM4H3Q as highly potent cyclic peptides targeting M pro . Other peptides and peptidomimetic inhibitors of SARS-CoV-2 M pro have been reported with varying potencies (Extended Data Fig. 5). Johansen-Leete et al. recently used the RaPID system to screen thioether-macrocyclic peptides containing standard proteinogenic amino acids in the random region between the cyclizing ClAc y and the downstream l-Cys (ref. 43). A macrocyclic peptide, with yLQY residues at the P3−P1′ equivalent positions, exhibiting an inhibition constant (K i ) value of 14 nM against M pro was identified. The P1 equivalent Q and P1′ equivalent Y residues are on either side of the scissile amide in this peptide, resulting in slow cleavage by M pro .
In the case of the M pro :GM4 complex crystal structure, well-defined electron density was observed for the complete intact peptide in the active site pocket despite the carbonyl of the amide linking H3 GM4 and γ 1 4 GM4 being positioned in the oxyanion hole. This observation both highlights the advantage of introducing a cγAA into the library and implies that GM4 and GM4H3Q in part resist hydrolytic cleavage due to the presence of γ 1 at the P1′ position. Other groups have also reported a substrate-derived cyclic peptide inhibitor of M pro (UCI-1 (ref. 44)) and linear peptide inhibitors of M pro (p13 (ref. 40) and compound 21 (ref. 45)), but the potency is limited to the micromolar range. Peptidomimetic inhibitors, including nirmatrelvir (PF-07321332) and PF-00835231, where the C-terminal residue reacts covalently with the nucleophilic cysteine have been developed 38,42,[46][47][48][49] (Extended Data Fig. 5a,b). X Q , which is present in these inhibitors, is an analogue of the substrate Q/P1 residue wherein the side chain δ-nitrogen is covalently linked to the γ-carbon (Extended Data Fig. 5b, indicated by the blue line) to give a conformationally constrained (S)-γ-lactam, which is well accommodated in the S1 subsite, resulting in improved potency relative to Q in inhibitors 46,50 . GM4 and GM4H3Q contain a related cγAA ring but are non-covalently binding inhibitors, but notably still exhibit low-nanomolar IC 50 values.
Recently, non-covalent small molecule inhibitors against M pro have also been developed by means of virtual screening and optimization of compounds based on structure-activity relationships [51][52][53][54] . Small molecular inhibitors often have an advantage of better cell membrane permeability than macrocycles and thus are able to inhibit M pro in cells. In fact, an impressive M pro non-covalent inhibitor, which is a current clinical candidate, S-217622, exhibited antiviral activity in cells (half-maximal effective concentration (EC 50 ) = 0.37 µM, in Extended Data Fig. 5c) 51 . Another example of a non-covalent inhibitor, compound 23, is reported to have potent in vitro inhibitory activity (IC 50 = 20 nM), but exhibited cellular cytotoxicity (half-maximal cytotoxic concentration (CC 50 ) in Vero E6 cells = 1.15 µM) 52 . This contrasts with S-217622, for which cell cytotoxicity was not reported, presumably because of substantial medicinal chemistry efforts to optimize in vivo properties, including from a toxicity perspective. However, classical small molecule drugs often suffer from insufficient target specificity and unwanted cell cytotoxicity. By contrast, macrocyclic peptide inhibitors generally have remarkably high target specificities, leading to low cytotoxicity; however, at least for intracellular targets, relatively low membrane permeability can be an issue for peptide macrocycles. Our GM4 and related macrocycles show no cytotoxicity at concentrations as high as 100 µM, but as yet do not manifest obvious antiviral activity in cells, likely due to their poor membrane permeability. Johansen-Leete et al. have reported an M pro -inhibiting thioether-macrocyclic peptide using the same technology as that used by us, except without elongation reprogramming (compound 1 in Extended Data Fig. 5a) 43 . The originally discovered peptide showed a potent in vitro efficacy (K i = 14 nM) but no antiviral activity. However, an increase in cellular uptake, by attaching a cell-penetrating peptide (CPP) at its C-terminal region, resulted in modest but observable antiviral activity. It is known that the addition of CPP to a macrocycle can not only increase cell permeability but also increase cytotoxicity depending on the CPP sequence and cell types used for the study 55,56 . Since we have witnessed improvements in in vivo potency by reengineering the macrocyclic peptide backbone structures, such as by N-methylation, in our previous works 57,58 , Article https://doi.org/10.1038/s41557-023-01205-1 our ultimate goal is to generate a non-CPP-modified macrocycle with better cell membrane permeability and potent antiviral activity. In the case of the GM4 peptides, our current efforts are directed at reducing molecular weight while maintaining the critical yFHγ 1 binding motif 59 .
Most importantly, our strategy for the construction of a cγAA-containing macrocyclic peptide library and its application for the RaPID system has demonstrated that unprecedented protease inhibitors containing cγAAs can be generated. It is striking that crystallography reveals that the γ 1 residue of GM4 sits deeply in the M pro active site, and that the compactness of γ 1 plays a critical role in organizing a unique turn conformation, resulting in resistance to proteolytic cleavage of GM4. The general ability of cγAAs in the preorganization of turn conformations thus appears to be advantageous in obtaining potent, compact and stable macrocycles with drug-like properties.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41557-023-01205-1.

Preparation of flexizymes and tRNAs
Flexizymes (dFx and eFx) and tRNAs for charging cγAAs, cβAAs and d-amino acids were transcribed in vitro using the T7 RNA polymerase from the corresponding template DNAs and prepared by extension and PCR (Supplementary Table 3 for primer sequences). PCR products were purified by phenol/chloroform extraction and ethanol precipitation. The transcription reaction was then carried out using the following mixture: 40 mM Tris-HCl buffer (pH 8.0), 22.5 mM MgCl 2 , 10 mM dithiothreitol, 1 mM spermidine, 0.01% Triton X-100, 3.75 mM NTP mix, 0.04 U µl -1 RNasin RNase inhibitor (Promega, N2615) and 120 nM T7 RNA polymerase (37 °C, 16 h). For transcription of tRNAs, 5 mM GMP was added to the above solution to introduce a monophosphate at the 5′ end of tRNAs. The resulting RNA transcripts were then treated with RQ1 DNase (Promega, M6101) at 37 °C for 30 min and purified by 12% (flexizymes) or 8% (tRNAs) denaturing polyacrylamide gel electrophoresis (PAGE) containing 6 M urea.

Preparation of aminoacyl-tRNAs
Preparation of aminoacyl-tRNA employed the flexizyme method 25,33,29 . The cγAAs, cβAAs and d-cysteine (c) were pre-activated as their 3,5-dinitrobenzyl esters; N-chloroacetyl-d-tyrosine ( ClAc y) was activated as its cyanomethyl ester. The activated amino acids were charged onto the respective tRNAs using flexizymes (dFx for 3,5-dinitrobenzyl ester or eFx for cyanomethyl ester). Aminoacylation was carried out at 4 °C for 16 h for cγAA and cβAA, 6 h for c or 2 h for ClAc y in the following mixture: 600 mM MgCl 2 , 20% DMSO solvent, 25 µM dFx or eFx, 25 µM tRNA and 5 mM activated amino acid. The reaction pH was adjusted by bicine-KOH (pH 8.7) for cγAAs and cβAAs, or HEPES-KOH buffer (pH 7.5) for c and ClAc y. The reaction was stopped by addition of ×4 volume of 0.3 M sodium acetate (pH 5.2) and ×10 volume of ethanol. The resulting aminoacyl-tRNAs were purified by ethanol precipitation.

Translation of peptides
Ribosomal synthesis of the model peptides and the peptide library was carried out using the modified FIT system (37 °C, 40 min) 32 Table 4 for the primer sequences). For the translation of the peptide library, a 1.5 µM mRNA library conjugated to puromycin linker was used.

Production of recombinant SARS-CoV-2 M pro
M pro was prepared as reported, and assays with it were performed exclusively using freshly purified recombinant M pro solution 39,61 . Refrozen M pro samples exhibited reduced activity and were not used.

RaPID selection of peptides against SARS-CoV-2 M pro
The random mRNA library was ligated with a puromycin linker at the 3′ end, and then added to the above described translation mixture (Extended Data Fig. 1a, step 1). The peptide library was translated at 37 °C for 40 min in 150 µl (for the first round of selection) or 10 µl (for the second to fourth rounds) of the FIT system and incubated at 25 °C for 5 min to conjugate the translated peptide with the corresponding mRNA-puromycin (step 2). A ×0.04 volume of 500 mM ethylenediaminetetraacetic acid (EDTA; pH 8.0) was then added and incubated (37 °C, 10 min) to dissociate ribosomes from the mRNA-peptide conjugates. Reverse transcription (42 °C, 30 min) used the NNUGWGAUG. R38 primer (5′-TTTCCGCCCCCCGTCCTAACCTCCGCTACCTCCACTCA-3′) and M-MLV reverse transcriptase lacking RNase H activity (Promega, M3682; step 3). The resulting cDNA/mRNA/peptide conjugates were subjected to naked Dynabead streptavidin (Thermo Fisher, DB11206) treatment (4 °C, 15 min) three times to remove bead-binding peptides; the supernatant was then applied to M pro -immobilized Dynabeads (4 °C, 15 min; step 4). The beads were washed with 100 µl ice-cold TBS-T buffer (50 mM Tris-HCl (pH 7.6), 150 mM NaCl, 0.05% (v/v) Tween 20) three times. Note that the removal of bead-binding peptides was not performed for the first selection round. Then, 100 µl of ×1 PCR buffer (10 mM Tris-HCl (pH 9.0), 50 mM KCl, 0.1% (v/v) Triton X-100, 0.25 mM dNTP, 2.5 mM MgCl 2 , 0.25 µM T7.F53 primer (5′-GGCGTAATACGACTC ACTATAGGGTTGAACTTTAAGTAGGAGATATATCCATG-3′) and 0.25 µM NNUGWGAUG.R38 primer) was added to the beads; the cDNAs were eluted at 95 °C for 5 min and PCR amplified to make a cDNA library (step 5). To estimate the recovery rate of cDNA, 1 µl of the elute was mixed with 19 µl of ×1 PCR buffer containing SYBR Green I (Lonza, 50513) and Taq DNA polymerase; amounts of cDNA were quantified by real-time PCR.

Solid-phase peptide synthesis
Macrocyclic peptides were synthesized by standard Fmoc solid-phase peptide synthesis using a Syro I automated peptide synthesizer (Biotage). Fmoc-protected amino acids and coupling reagents were from Merck, Watanabe Chemical Industries or Enamine. NovaPEG Rink Amide Resin (54 mg, 25 µmol) was incubated with N,N-dimethylformamide (DMF; room temperature, 1 h). Each Fmoc-protected amino acid was coupled at 30 °C for 40 min on the resin in a DMF solution containing 0.2 M Fmoc-protected amino acid (6 equiv.), 0.2 M 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (5 equiv.), 0.2 M 1-hydroxybenzotriazole (5 equiv.) and 0.1 M N,N-diisopropylethylamine (12 equiv.). After washing the resin five times with 600 µl DMF, the Fmoc group was deprotected with 600 µl of 40% (v/v) piperidine in DMF at 30 °C for 12 min. Coupling of the Fmoc-protected amino acid and Fmoc deprotection were repeated accordingly. After automated peptide synthesis, 0.2 M chloroacetyl N-hydroxysuccinimide ester (8 equiv.) in N-methylpyrrolidone was added to the resin; the mixture was incubated at room temperature for 1 h with rotation. After washing the resin subsequently with DMF five times and with dichloromethane five times, the resin-bound peptides were treated with 2 ml of a solution of 92.5% (v/v) TFA, 2.5% (v/v) water, 2.5% triisopropylsilane and 2.5% 3,6-dioxa-1,8-octanedithiol at Nature Chemistry Article https://doi.org/10.1038/s41557-023-01205-1 room temperature for 3 h with rotation to deprotect and cleave off from the resin. The resulting linear peptides were precipitated with diethyl ether and dissolved in 10 ml of 80% (v/v) DMSO and 0.1% TFA in water. Following the addition of 200 µl of 0.5 M tris(2-carboxyethyl) phosphine and triethylamine to adjust the pH to 8, the peptide mixture was incubated with rotation at room temperature for 10 h to form a thioether bond between the N-terminal chloroacetamide and thiol group of the downstream cysteine. Macrocyclization of the peptides was confirmed by MALDI-TOF MS; the crude peptides were purified by reverse-phase high-performance LC (Shimadzu). The purities of the peptides were evaluated by UPLC (Shimadzu) using a reverse-phase column (ACQUITY UPLC BEH C18, 1.7 µm, 2.1 × 150 mm; Waters) with a linear gradient from 10% buffer B to 70% buffer B. Buffer A was water with 0.1% (v/v) TFA; buffer B was acetonitrile with 0.1% (v/v) TFA.

Binding kinetics analysis of peptides by SPR
The binding affinities of selected peptides and M pro were analysed by SPR using a Biacore T200 instrument (Cytiva) at 25 °C. The composition of the running buffer was 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05% (v/v) Tween 20 and 0.1% (v/v) DMSO. Biotin-tagged M pro was immobilized on a Biacore sensor chip CAP (Cytiva) to a surface density of 1,500-2,000 response units following the immobilization protocols of the Biotin CAPture Kit (Cytiva). The kinetic constants were determined by a single-cycle kinetics method by the injection of five different concentrations (twofold dilution series) of each peptide at a flow rate of 30 µl min -1 . Binding sensorgrams were fitted to the standard 1:1 interaction model and analysed using Biacore evaluation software.

Solid-phase extraction coupled to MS inhibition assays
Inhibition of M pro was measured by solid-phase extraction purification coupled to MS analysis using a RapidFire 365 high-throughput sampling robot (Agilent) connected to an iFunnel Agilent 6550 accurate mass quadrupole TOF spectrometer as reported 39 . In brief, the cyclic peptides were dispensed in an 11-point, threefold dilution series across 384-well plates using an acoustic Echo Dispenser machine (LabCyte). Formic acid and DMSO were used as positive and negative controls, respectively. The assay was adapted for a reaction volume of 20 µl with a final condition of 75 nM M pro and 4 µM 37-mer substrate (ALNDF-SNSGSDVLYQPPQTSITSAVLQ/SGFRKMAFPS-NH 2 ). Reactions were incubated (15 min) and then quenched by the addition of 10% (v/v) aqueous formic acid (5 µl per well). C-terminal product peptide-derived SGFRKMAF-NH 2 (10 µl per well) was employed as the internal standard; data were extracted and processed as reported.

Serum stability assays
A synthetic macrocyclic peptide (10 µM) and an internal standard peptide (10 µM) ( Supplementary Fig. 6a for its structure) were mixed and incubated in human serum (Cosmo Bio, 12181201) at 37 °C for up to 100 h. We have previously shown that the standard peptide is peptidase resistant, with no degradation occurring under these experimental conditions 25 . At each time point (0, 0.5, 1, 2, 4, 8, 24, 50 and 100 h), 4 µl of the mixture was removed and quenched by adding 12 µl methanol. Following centrifugation (15,000g, 25 °C, 3 min), 10 µl of the supernatant was mixed with 40 µl of 1% (v/v) TFA. Following centrifugtion (15,000g, 25 °C, 3 min), the supernatant was collected for LC/MS analysis, which used a reverse-phase column (ACQUITY UPLC BEH C18, 1.7 µm, 2.1 × 150 mm; Waters) and a Xevo G2-XS QTof system (Waters) with a linear gradient from 1% B to 60% B. Buffer A was water with 0.1% (v/v) formic acid; buffer B was acetonitrile with 0.1% (v/v) formic acid. The percentages of remaining peptides were determined by the peak area integration of the chromatograms.

Crystallization, data collection and structure solution
M pro was thawed and diluted to 6 mg ml -1 using 20 mM HEPES (pH 7.5) and 50 mM NaCl. GM4 was diluted into the protein solution to a final concentration of 10 mM, which was incubated for two hours at room temperature prior to dispensing plates. The drop composition was 0.15 µl protein ligand solution, 0.3 µl 11% (v/v) polyethylene glycol (PEG) 4K, 0.1 M 2-(N-morpholino)ethanesulfonic acid (MES) pH 6.5 and 0.05 µl M pro crystal seed stock. The M pro crystal seed stock was prepared by crushing M pro crystals with a pipette tip; suspending them in 30% PEG 4K, 5% (v/v) DMSO and 0.1 M MES pH 6.5; and vortexing for 60 s with ~10 glass beads (1.0 mm diameter, BioSpec Products). The reservoir solution was 11% (v/v) PEG 4K, 5% (v/v) DMSO and 0.1 M MES pH 6.5. Crystals were grown by the sitting drop vapour diffusion method (20 °C) and appeared within 24 h, reaching full size within 36 h; they were harvested after one week.
Data collection and structure determination. All diffraction data were collected at 100 K with a wavelength of 0.9762 Å on beamline I03 at the Diamond Light Source. Data were processed using Dials 62 via Xia2 63 and Aimless 64 within CCP4i2 (ref. 65). The datasets were phased using Molrep 66 using the M pro apo structure (Protein Data Bank 6YB7). Ligand restraints were generated using AceDRG 67 . Crystal structures were manually rebuilt in Coot and refined using Refmac 68 and PDBredo 69 .

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Coordinates and structure factors have been deposited in the Protein Data Bank (PDB) under accession code 7Z4S. Other data supporting this study are available in the Supplementary Information. Source data are provided with this paper.