key: cord-1000545-5f2fcy60 authors: Karoyan, Philippe; Vieillard, Vincent; Odile, Estelle; Denis, Alexis; Gómez-Morales, Luis; Grondin, Pascal; Lequin, Olivier title: An hACE2 peptide mimic blocks SARS-CoV-2 Pulmonary Cell Infection date: 2020-08-24 journal: bioRxiv DOI: 10.1101/2020.08.24.264077 sha: 197c0998f049345c4e0605c6cfb4013375c100e5 doc_id: 1000545 cord_uid: 5f2fcy60 In the light of the recent accumulated knowledge on SARS-CoV-2 and its mode of human cells invasion, the binding of viral spike glycoprotein to human Angiotensin Converting Enzyme 2 (hACE2) receptor plays a central role in cell entry. We designed a series of peptides mimicking the N-terminal helix of hACE2 protein which contains most of the contacting residues at the binding site and have a high helical folding propensity in aqueous solution. Our best peptide mimic binds to the virus spike protein with high affinity and is able to block SARS-CoV-2 human pulmonary cell infection with an inhibitory concentration (IC50) in the nanomolar range. This first in class blocking peptide mimic represents a powerful tool that might be used in prophylactic and therapeutic approaches to fight the coronavirus disease 2019 (COVID-19). In Brief Helical peptide mimicking H1 helix of hACE2 and composed of only natural amino acids binds to SARS-CoV-2 spike protein with high affinity and blocks human pulmonary cells infection with IC50 in the nM range. Highlights A peptide mimic of hACE2 designed from H1 helix and composed of only natural amino acids show high helical folding propensity in aqueous media. This peptide mimic binds to virus spike RBD with high affinity (sub-nM range). This peptide mimic blocks SARS-CoV-2 pulmonary cells infection with an IC50 in the nM range. This peptide mimic is devoid of toxicity on pulmonary cells. The coronavirus disease 2019 , caused by the severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) has emerged as a pandemic, claiming at the time of writing more than 600,000 deaths and over 14.5 million confirmed cases world-wide between December 2019 and July 2020. 1 Since SARS-CoV-2 discovery 2,3 and identification, the energy deployed by the scientific community has made it possible to generate an extraordinary wealth of information. Whatever, to date, clinically approved vaccines or drugs are lacking. 4 Indeed, no specific drugs targeting the virus are available 5 and many clinical trials have been engaged with SARS-CoV-2 non-specific treatments. 6 The structural and biochemical basis of infection mechanism has been investigated, highlighting that the virus cell-surface spike protein of SARS-CoV-2 is targeting human receptors. 4, 7 Human Angiotensin Converting Enzyme 2 (hACE2) and the cellular Transmembrane Protease Serine 2 (TMPRSS2) have been identified as major actors of the virus entry into human cells. With the goal of preventing the SARS-CoV-2 from infecting human cells, blocking the interaction between hACE2 and the virus spike protein has been validated. Indeed, inhibition of SARS-CoV-2 infections in engineered human tissues using clinical-grade soluble ACE2 was recently demonstrated. 8 Likewise, an engineered stable mini-protein mimicking three helices of hACE2 to plug SARS-CoV-2 spikes 9 was described, but its capacity to block viral infection was not demonstrated. If several in silico designed peptides were proposed to prevent formation of the fusion core, 10 first attempts to design a peptide binder derived from hACE2 proved to be a difficult task, leading to mitigated results. 11 Thus, starting from the published crystal structure of SARS-CoV-2 spike receptor-binding domain (RBD) bound to hACE2, 12 we designed peptide mimics of the N-terminal hACE2 helix which interact with spike protein. We report here the strategy implemented to optimize the design of our peptide mimic, its high helical folding propensity in water and its ability to block cells infection by SARS-CoV-2 with an IC50 in the nM range upon binding to spike RBD with strong affinity. We also demonstrate the non-toxicity of our mimic at concentrations 150 times higher than IC50 on pulmonary cell lines. We first examined the complex between hACE2 and the surface spike protein of SARS-CoV-2 (PDB 6m0j) 12 in order to highlight the important contacts and some relevant characteristics of the interacting hACE2 sequence (Figure 1) . B. C. -S19T20I21E22E23Q24A25K26T27F28L29D30K31F32N33H34E35A36E37D38L39F40Y41Q42S43S44L45- Twenty residues from hACE2 were identified 12 as close contacts, using a distance cut-off of 4 Å. These interactions occur mainly through the N-terminal a-helix H1 of hACE2 ( Figure 1A) . This a-helix, composed of 27 residues (from S19 to L45, Figure 1C ) contains 12 residues (highlighted in red in Figure 1B ) involved in hydrophobic interactions, hydrogen bonds and salt bridges. 12 Our strategy was to design a peptide with high helical folding propensity and retaining most of the binding affinity of hACE2 to spike RBD of SARS-CoV-2, using natural amino acids 13 . Indeed, we preferred not using complex chemical tools known to stabilize a-helix 14,15 in order to limit developability constraints. To design our mimics, we set up an optimization process with the dual aim of optimizing the helical content and limiting the antigenicity, to avoid triggering a neutralizing immune response that would compromise the peptide therapeutic potential. A combination of the Agadir program, 16 an algorithm developed to predict the helical content of peptides, and the semi-empirical method reported by Kolaskar 17, 18 to highlight the number of antigenic determinants (AD), was iteratively used. We observed that the N-terminal sequence of H1 helix, composed of 4 residues (S19TIE22), corresponds to a consensus N-capping box motif (SXXE). 19 A capping box features reciprocal backbone-side-chain hydrogen-bonds favoring helix initiation. Although this sequence does not adopt the H-bonded capping conformation in the crystal structure of the full protein, it could still be an important stabilizing element in the isolated helix when extracted from the protein context. These observations led us to keep 14 residues from the native H1 helix of hACE2 as contact residues or putative stabilizing capping box. The 13 left residues that are not essential for the interaction were considered as possible sites for amino acid substitutions ( Figure 1C ). We thus substituted non-essential positions by Ala and/or Leu residues displaying higher helical folding propensities and calculated the peptide helical content after each substitution (Table S1) . A peptide sequence optimization was then realized to lower the antigenicity while keeping the helical propensity thanks to an iterative residue scanning and calculation of the helical content variation upon new substitutions (Table S2) . This strategy highlighted the influence of the residue N33 in the native sequence. Indeed, if the N33/L33 substitution systematically improved the helical content, it was always at the expense of antigenicity. Conversely, the L33/N33 substitution reduced the antigenicity at the expense of helical content. The solution was found by L33/M33 substitution decreasing the number of AD. The H1 helix of ACE2 adopts a kinked conformation in the crystal structure, leading to a distorted CO/HN hydrogen bond network between H34/D38 and E35/L39 residues. Therefore, we considered introducing a proline as this residue is known to induce local kinks or distortions in natural helices. 20, 21 D38 was classed as a contact residue, while L39 side chain is not involved in any direct interaction. Consequently, L39 position was selected for substitution by proline (peptide P5, Table 1 ). In order to increase the helical content to a maximum level, this iterative study was also applied to longer peptide sequences starting from the 29-mers native one, albeit at the expense of antigenicity. Diverse combinations of N-and C-terminus capping groups were also considered (free extremities or N-acetyl, C-carboxamide groups). Finally, we examined the possibility of promoting additional side chain contacts provided by ACE2 residues that do not belong to H1 helix. Y83 residue appeared as a good candidate as it lies very close in space to A25 in H1 helix. Molecular modeling was carried out on H1 helix analogs in which A25 was replaced by tyrosine or homotyrosine (hTyr) residues ( Figure S1 ). Calculations showed that hTyr residue was able to project the phenol ring in the adequate 3D space to mimic Y83 position. Of note, hTyr is a natural amino acid. 22 Three peptides were selected as controls in our optimization process, P1 (native sequence), P1scr (Scrambled peptide from P1) and Ppen (described by Pentelute & al. in a longer biotinylated and pegylated construct and termed SBP1 as a putative spike binder). 11 The results highlighting the progression in the helical content and the number of antigenic determinants is reported in Table 1 for the most relevant peptide mimics (see table S1, S2 and S3 for all peptide mimics designed and/or synthesized). These peptides were synthesized on a 5 to 20 mg scale from Fmoc-protected amino acids utilizing standard solid phase peptide synthesis (SPPS) methods on rink amide resin (See Methods). a Amino acid sequence colored as follows: red, contact residue; green, non-essential residue; blue, substituted residue with respect to native sequence. Homotyrosine residue is depicted with underlined letter Y. b Predicted helical content using Agadir program; n.a., not applicable (for hTyr containing sequences). c Predicted antigenicity calculated from the predicting antigenic peptide tool from the immunomedicine group 23 ; n.a., not applicable (for hTyr containing sequences). d Experimental helical content determined by circular dichroism (see Figure 2 ). e % of inhibition on Vero-E6 cells of at least three independent experiments (see Figure 3A ). f For peptides P9 and P10, inhibition of infection was measured in the range 0.1-1 µM. g Determined on VeroE6 and Calu3 cells only for peptides able to inhibit SARSCoV-2/VeroE6 cell infection. h Dissociation constants with SARS-Cov-2 RBD-SD1 measured by BLI for peptides with IC50 in the nM range. The designed peptides highlight an excellent correlation between calculated and experimentally determined helical content by circular dichroism in aqueous media The conformation of synthesized peptides in aqueous solution was investigated by CD spectroscopy. 24 Figure 2 shows the superimposed CD spectra of 12 peptides, including the control ones, i.e. P1 (native sequence), P1scr (Scramble), Ppen. The CD spectra of peptides P1 (native), P1scr (scrambled) are characteristic of a predominant random coil structure with a negative minimum near 200 nm, as expected. Similarly, Ppen described as a helical peptide sequence 11 adopted also a random coil conformation in solution. For all other peptides, the CD spectra exhibit the canonical a-helix signature, with a double minimum around 208 and 222 nm, with the exception of the proline containing P5 peptide. The deconvolution of the CD spectra using DICHROWEB allowed estimating the helical population for each peptide, which is reported in Table 1 . Overall, an excellent agreement was observed between the AGADIR-computed values and the experimental helical population inferred from CD data. The native hACE2 H1 helix sequence (peptides P1, Ppen) has a weak propensity to fold into an a-helix in aqueous solution (below 10%). In contrast, sequence optimization lead to H1 analogs exhibiting a high helical propensity (between 50 and 80%). The introduction of a proline residue in peptide P5 has a strong destabilizing effect on helical conformation (17%) whereas L/hY substitution only led to a slight decrease of helical content (P7 versus P6 and P10 versus P8). which SARS-CoV-2 replicated efficiently. 28 We first observed a dose-dependent reduction in virus titer ( Figure 3B ) and then calculated average median inhibitory concentration (IC50) on VeroE6 and Calu3 cells for P8, P9 and P10, with respectively (IC50) of 800 nM, 300 nM and 60 nM ( Figure 3C) and 94 nM, 76 nM and 84 nM ( Figure 3D) . Importantly, no cytotoxicity was observed in similarly treated uninfected culture cells at 10 µM (concentration more than 150 times higher than IC50 for the most potent peptide mimics, Figure 3E and 3F and SI). Collectively, these data demonstrate the high antiviral potency of peptide analogs P9 and P10. A. B. C. D. E. F. Finally, the peptides able to block the cell infection with IC50 in the nM-sub µM range (P8, P9 and P10) were evaluated for their ability to bind to SARSCoV-2 spike RBD ( Figure 4 ) using Biolayer Interferometry (BLI) with an Octet RED96e (FortéBio). 29 hACE2 and P1 were used as respectively positive and negative controls. Even though this technique presents some drawbacks and at least poor sensitivity when the immobilized sample is a protein of high molecular weight and the binding partners are peptides, it remained useful to identify and rank binding peptides. Of note, only association rates could be quantified accurately, dissociation ones being very slow, highlighting strong binding properties of some peptides. Data indicate that the most efficient peptides P9 and P10 bind with estimated Kd below 1 nM, whereas P1 does not bind to spike RBD in the conditions tested here. The pandemic caused by SARS-CoV-2 is at the origin of an unprecedented health crisis. 34, 35 . The syntheses and the development of long therapeutic peptides (over 30 residues) are no more a challenge, as highlighted by the success story of many GLP-1 analogs. 36 Their possible antigenicity can be evaluated using prediction tools in the design. 37 Of course, even for peptides, the development of a drug at a pandemic speed requires some considerations. Our aim was to design a peptide with reasonable helical folding propensity in water in order to mimic the H1 helix of hACE2 in the protein context, considering this helical folding as a prerequisite to compete with hACE2 upon interaction with viral spike protein. The design was realized using only natural amino acids 38 and avoiding complex tools known and validated to stabilize a-helix. 39, 40 Stabilizing a-helical structure of medium size peptide sequences (up to 15 residues) using only natural amino acids is a hard but achievable task. 41 Our choice was guided by the desire to build a simple peptide easy to produce quickly on a large scale, without technical constraints requiring sometimes laborious development. We also assumed that the use of mostly natural amino acids can facilitate the essential stages of the development of therapeutic tools in the event of success, particularly around pharmacokinetics, pre-clinical and clinical toxicity aspects. This seems to us to be a fair compromise between designing a-helix peptide with optimized binding affinity and developing an effective tool within short deadlines by integrating the constraints of developability. Using a combination of validated methods, we improved the helical folding propensity of the native a-helix extracted from the protein context, thanks to leucine and alanine scanning (See tables 1, S1 and S2), designing and synthesizing a first set of peptides (P1 to P8). These peptides demonstrated to have a high helical content (up to 80% for P6). However, increasing this helical content to maximum level led to increasing mean hydrophobicity and hydrophobic moment (see Table S4 ) that proved to be detrimental to solubility and efficacy. The substitution of a leucine residue by the homoTyrosine residue led to peptide analogue P7 with a slight increase in solubility and a weak efficiency to block SARS-CoV-2 cell infection (IC50=7 µM). In this first generation of peptides, the 27 residue P8 peptide appeared to be highly soluble with high helical folding propensity (70%), and an ability to block SARS-CoV-2 cell infection at 10 µM with an IC50 of 800 nM on VeroE6 cells. In order to improve the potency of our peptides, we designed a new set of mimics combining the properties of peptides P7 and P8, i.e. P9 and P10. If the Leu/hTyr substitution led to a slight decrease of experimental helical content, this was at the advantage of the mean hydrophobicity (see Table S4 ) also highlighted by lower HPLC retention times (see Table S3 ). These peptides proved to be highly efficient in blocking SARSCoV-2 cell infection (100% efficacy at 1 µM) on VeroE6 cells together on pulmonary cells with IC50 in the nM ranges. This blocking property is related to their ability to bind to SARSCoV-2 spike RBD with affinity estimated in the sub-nanomolar range (Tables 1 and S6) . Finally, these peptides proved to be devoid of cell toxicity at 150 times the IC50 concentration (Figures 3E and F and SI) highlighting their therapeutic potential. We The authors declare the following competing financial interest(s): The patent application EP20305449.9 included results from this paper. The authors declare that no other competing interests exist. SARS(-CoV-2), severe acute respiratory syndrome (-coronavirus 2); COVID-19, coronavirus disease 2019; hACE2, human angiotensin converting enzyme 2. AD, antigenic determinant; LCMS, liquid chromatography mass spectroscopy. CD, Circular Dichroism. resin with classical cleavage cocktail TFA/TIS/H2O (95:2.5:2.5). The crude products were purified using preparative scale HPLC. The final products were characterized by analytical LCMS. All tested compounds were TFA salts and were at least 95% pure. The relevant peptides after CD spectra analyses were selected and produced by Genecust France on 20mg scale. Preparative CD spectra were recorded on a Jasco J-815 CD spectropolarimeter equipped with a Peltier temperature controller. Data were obtained at 25°C over a wavelength range between 185 and 270 nm, using a wavelength interval of 0.2 nm and a scan rate of 20 nm/min. Peptide samples were prepared at a concentration of 60 µM in 50 mM sodium phosphate buffer, pH 7.4, in a quartz cell of 1 mm path length. CD experiments were processed and plotted with R program. CD spectra were analyzed using DICHROWEB web server and CDSSTR deconvolution algorithm. 24 Accordingly, the viral titer of SABA95 P2 stock is about 5.3 10 5 PFU/mL. A novel coronavirus outbreak of global health concern A Novel Coronavirus from Patients with Pneumonia in China Structural basis of receptor recognition by SARS-CoV-2 Structural basis for the recognition of SARS-CoV-2 by full-length human ACE2 Science Inhibition of SARS-CoV-2 Infections in Engineered Human Tissues Using Clinical-Grade Soluble Human ACE2 An engineered stable mini-protein to plug SARS-Cov-2 Spikes In silico design of antiviral peptides targeting the spike protein of SARS-CoV-2, Peptides, 2020, 130, 170328. B. Han, Y.; Král, P. Computational Design of ACE2-Based Peptide Inhibitors of SARS-CoV-2 The first-in-class peptide binder to the SARS-CoV-2 spike protein Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor Stabilized Helical Peptides: A Strategy to Target Protein−Protein Interactions A Highly Stable Short r-Helix Constrained by a Main-Chain Hydrogen-Bond Surrogate Elucidating the folding problem of ahelical peptides using empirical parameters, II. Helix macrodipole effects and rational modification of the helical content of natural peptides A semi-empirical method for prediction of antigenic determinants on protein antigens Helix Stop Signals in Proteins and Peptides: The Capping Box Substituted prolines: syntheses and applications in structure-activity relationship studies of biologically active peptides Proline-induced distortions of transmembrane helices Homotyrosine-Containing Cyanopeptolins 880 and 960 and Anabaenopeptins 908 and 915 from Planktothrix agardhii CYA Two Detailed Plaque Assay Protocols for the Quantification of Infectious SARS-CoV-2 An orally bioavailable broad-spectrum antiviral inhibits SARS-CoV-2 in human airway epithelial cell cultures and multiple coronaviruses in mice Cultured Human Airway Epithelial Cells (Calu-3): A Model of Human Respiratory Function, Structure, and Inflammatory Responses Comparative tropism, replication kinetics, and cell damage pro ling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study, The Lancet Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation Synthetic therapeutic peptides: science and market Therapeutic peptides: Historical perspectives, current development trends, and future directions Peptide therapeutics: Current status and future directions The emergence of peptides in the pharmaceutical business: From exploration to exploitation The future of peptide development in the pharmaceutical industry. PharManufacturing: The International Peptide Review Glucagon-like peptide-1 analogue s: An overview Stabilized Helical Peptides: A Strategy to Target Protein−Protein Interactions A Highly Stable Short r-Helix Constrained by a Main-Chain Hydrogen-Bond Surrogate A single-stranded amphipathic alpha-helix in aqueous solution: Design, structural characterization, and its application for determining alpha-helical propensities of amino acids Cell supernatants were collected at 48 h post-infection for Elisa assay using SARS-CoV-2 (2019-nCoV) Nucleoprotein / NP ELISA Kit from Sino biological, according the manufacturer's instruction, and standard plaque assay Methodology: 96-well plates were used to seed 10 4 CALU-3 or Vero-E6 cells per well The sensortip was then dipped into 100 nM hACE2 (Sanyou Biopharmaceuticals Co. Ltd, His Tag) or 1 µM of any tested peptide to measure association before being dipped into a well containing only running buffer composed of DPBS (Potassium Chloride 2.6mM, Potassium Phosphate monobasic 1.5mM, Sodium Chloride 138mM Peptides were produced manually synthesized from Fmoc-protected amino acids utilizing standard solid phase peptide synthesis (SPPS) methods. Solid-phase peptide syntheses were performed in polypropylene Torviq syringes (10 or 20 mL) fitted with a polyethylene porous disk at the bottom and closed with an appropriate piston. Solvent and soluble reagents were removed through back and forth movements. The appropriate protected amino acids were sequentially coupled using PyOxim/Oxyma as coupling reagents. The peptides were cleaved from the rink amide