A wild boar cathelicidin peptide derivative inhibits severe acute respiratory syndrome coronavirus-2 and its drifted variants

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) poses a clear threat to humanity. It has infected over 200 million and killed 4 million people worldwide, and infections continue with no end in sight. To control the pandemic, multiple effective vaccines have been developed, and global vaccinations are in progress. However, the virus continues to mutate. Even when full vaccine coverage is achieved, vaccine-resistant mutants will likely emerge, thus requiring new annual vaccines against drifted variants analogous to influenza. A complimentary solution to this problem could be developing antiviral drugs that inhibit SARS CoV-2 and its drifted variants. Host defense peptides represent a potential source for such an antiviral as they possess broad antimicrobial activity and significant diversity across species. We screened the cathelicidin family of peptides from 16 different species for antiviral activity and identified a wild boar peptide derivative that inhibits SARS CoV-2. This peptide, which we named Yongshi and means warrior in Mandarin, acts as a viral entry inhibitor. Following the binding of SARS-CoV-2 to its receptor, the spike protein is cleaved, and heptad repeats 1 and 2 multimerize to form the fusion complex that enables the virion to enter the cell. A deep learning-based protein sequence comparison algorithm and molecular modeling suggest that Yongshi acts as a mimetic to the heptad repeats of the virus, thereby disrupting the fusion process. Experimental data confirm the binding of Yongshi to the heptad repeat 1 with a fourfold higher affinity than heptad repeat 2 of SARS-CoV-2. Yongshi also binds to the heptad repeat 1 of SARS-CoV-1 and MERS-CoV. Interestingly, it inhibits all drifted variants of SARS CoV-2 that we tested, including the alpha, beta, gamma, delta, kappa and omicron variants.

tide library, we first identified previously characterized cathelicidin genes from the genomes of other species listed on The Universal Protein Resource (UniProt) and then identified additional putative cathelicidins based on homology using BLAST (Supplementary Table 1) 28 .The cathelicidin peptide was isolated and produced by solid-phase synthesis as a pure peptide for each cathelicidin gene.
For the initial library screening, we evaluated the inhibitory potential of each cathelicidin peptide at 50 µM by an in vitro focus forming assay (Supplemental Figure 1) or virus infectivity assay on Vero E6 cells.Using a 50% inhibitory cutoff, our initial assay identified SARS-CoV-2 inhibitory activity in 9 out of 50 candidates in the cathelicidin peptide library including LL-37 and PMAP-36R.We then tested each 50% inhibitory candidate for activity at concentrations ranging from 50 to 0.78 µM using Vero E6 cells overexpressing human ACE2 (Vero hACE2) (Fig. 1 and Supplemental Figure 2).Most candidates failed to achieve > 50% inhibition below 50 µM..Positively charged residues are depicted in blue while hydrophobic residues are depicted in red.Labeled residues along the N-terminal α-helix form a non-polar sector including the weakly hydrophilic threonine.(B-E) Inhibition of SARS-CoV-2 infection by PMAP-36R and its derivatives.PMAP-36R was compared against LL-37 (B), N-terminal truncations (C), C-terminal truncations (D), or a cysteine mutant (E) of PMAP-36R.Peptides at the labeled concentrations were pre-incubated with 100pfu of live SARS-CoV-2 virus (nCoV/USA_WA1/2020) for 1 h at 37 °C before addition to confluent Vero hACE2 cells in a 96-well plate.Infected cells were fixed and quantified by focus forming assay after 48 h.Inhibition of viral infection was calculated based on the percent area of each well staining positively for viral spike protein compared to control wells without peptide inhibitor treatment.Results are representative of 3 independent experiments performed in triplicate.Significance calculated by two-way ANOVA with Bonferonni's correction comparing peptide derivatives against control PMAP-36R (* < .05,** < .01,** < .001,* < .0001).
Vol:.( 1234567890 www.nature.com/scientificreports/Mutation analysis of PMAP-36R identifies key residues for SARS-CoV-2 inhibition.To determine the critical residues in PMAP-36R responsible for its high anti-SARS-CoV-2 activity, we synthesized and tested truncated variants of the parent cathelicidin peptide lacking residues from the N or C termini, as well as a variant without the cysteine residue (Fig. 1A).N-terminal truncates lacking 9, 12, or 14 residues (p9N, p12N, and p14N) displayed a progressive loss of anti-SARS-CoV-2 activity, with no inhibition present in p14N (Fig. 1C).By comparison, the C-terminus was more sensitive to truncation, as loss of the last two residues in p2C immediately impaired function.At the same time, little inhibitory activity remained in p5C and none in p9C (Fig. 1D).The activity of these mutants suggested that viral restriction depended on both N and C terminal amino acids (summarized in Table 1).Interestingly, the final two residues of PMAP-36R on the C-terminus include a singular cysteine, which might be responsible for initiating covalent bonds with other PMAP-36R monomers or part of the SARS-CoV-2 virion.We also made an additional mutant of PMAP-36R by changing the penultimate cysteine to a structurally similar but non-reactive serine residue.The replacement of C36S in pSer had no significant effect on SARS-CoV-2 inhibition compared to PMAP-36R (Fig. 1E).
pSer improves the viral specificity of PMAP-36R by reducing cytotoxicity.Host defense peptides, while exhibiting antimicrobial activity, often exhibit toxicity to mammalian cells.To characterize the potential of PMAP-36R for therapeutic applications, we evaluated its toxicity toward mammalian cells via a hemolysis assay with human red blood cells (RBCs).Of the peptides we tested, LL-37 and PMAP-36R had considerable hemolytic activity.Interestingly, the C36S mutant pSer exhibited significantly reduced hemolytic activity, lysing less than 10% of human RBCs even at 50 µM (Fig. 2A).
In addition to the hemolysis assay, which evaluates the immediate lytic effects, each peptide was evaluated for cytotoxicity after 48 hours of cell culture.Cytotoxicity was tested against Vero hACE2 and HEK293T cells to identify general and cell line-specific sensitivities to cathelicidin peptide treatment.We quantified cell viability by a formazan formation assay.As in the hemolysis results, pSer exhibited decreased cytotoxicity compared to the parent PMAP-36R and LL-37 (Fig. 2B,C).The toxicity from each peptide appeared independent of cell type, producing similar trends in all tested cell lines.In contrast to the hemolysis assay, PMAP-36R has greater toxicity than LL-37 when administered over 48 hours, suggesting that it may act via a slower mechanism than LL37 or differentially target the membranes of RBCs and adherent cell lines.As the toxicity of pSer did not reach saturation at 50 µM, an additional assay was performed with Vero cells evaluated at a 24-hour timepoint (Fig. 2D).These conditions did not substantially alter the observed cytotoxicity compared to Vero cells evaluated at 48 hours.As expected, pSer reached saturation around 100 µM with a calculated TD 50 of 50.41 µM (Table 1).
The therapeutic index (TI) is a measure of safety for therapeutic drugs and, in this context, is the ratio of a compound's IC 50 to its TD 50 .In this case, TI represents the capacity of a cathelicidin peptide to discriminate between SARS-CoV-2 virions and Vero-E6 cell membranes.LL-37 demonstrated no therapeutic potential, with a TI of < 1, as neutralization consistently lagged toxicity.The TI was slightly improved for PMAP-36R at 3.21, indicating a modest targeting of SARS-CoV-2 over Vero E6 cells.However, pSer displayed heightened specificity for SARS-CoV-2 with a TI of 4.48.We named this peptide derivative "Yongshi", after the Mandarin word for warrior.

Yongshi is active primarily during and after virus infection.
To assess whether the inhibitory effects of Yongshi are mediated by cellular conditioning or direct viral neutralization, we compared the efficacy of viral inhibition when Vero hACE2 cells were treated with Yongshi 1-hour prior to, at the time of, or 1 hour after infection with SARS-CoV-2 and compared these results against cells infected with SARS-CoV-2 that was pre-incubated with Yongshi for 1 hour (Fig. 3).Pre-treatment of cells with Yongshi did not substantially inhibit SARS-CoV-2 infection at any concentration, suggesting Yongshi does not induce a significant antiviral state in host cells as has been described for LL-37 20 .By comparison, neither addition of Yongshi at the time of infection nor 1 hour following infection significantly altered viral inhibition compared to virus pre-incubated with Yongshi.These results further suggest that the inhibition by Yongshi is not dependent on the extended pre-incubation with virus but occurs rapidly at the time of infection.www.nature.com/scientificreports/Yongshi requires chirality for its virus-inhibitory activity.Next, we tested whether peptide chirality had an impact on the effectiveness of Yongshi antiviral activity.To do this, we synthesized a stereoisomer or mirror image variant of Yongshi using D rather than L amino acids.The resulting peptide, D-Yongshi, has an identical sequence and amphipathicity but a mirrored 3D conformation, permitting the dissection of Yongshi's inhibitory activity into sequence-based and structure-based components.Compared to the L-Yongshi peptide, D-Yongshi possessed greatly increased cytotoxicity (Fig. 4A).Although we tested D-Yongshi against SARS-CoV-2, viral inhibition was only observed at concentrations that also induced substantial cell death (Fig. 4B).
Deep-learning sequence comparison and computational modeling predict mimicry of Yongshi to the spike HR1/HR2 regions.What is the molecular mechanism of Yongshi's inhibition of the SARS-Cov-2 virus?A common mechanism of peptide inhibition is mimicry of functionally vital interactions.To find such a clue for Yongshi, we searched its sequence against a representative set of sequences whose structures (and some of their molecular interaction partners) have been determined in the Protein Data Bank (PDB) 33 .For this analysis, classic sequence search tools such as PSI-BLAST 34 and more sensitive HHsearch 35 yield a very small number of hits and do not find any hit related to the proteins of SARS-CoV-2.We then applied our recently developed algorithm SAdLSA 36 , deep-learning based sequence alignment method that shows significantly improved accuracy via learning the structural fold or motif representations encoded by protein sequences 37 .When applied to a recent PDB sequence library (~ 83,000 sequences, see Methods), among the top 1% sequences ranked in comparison to Yongshi, SAdLSA identified the Heptad Repeat 1 and 2 (HR1 and HR2) of SARS Spike protein in PDB entry 1ZV8, HR2 of SARS-Cov-2 in PDB entry 6LVN, as well as HR repeats in respiratory syncytial virus (PDB 3KPE) and Mumps virus (PDB 2FYZ).Figure 6A shows SAdLSA alignments between Yongshi and HR1/2 of SARS, SARS-CoV-2, and MERS, respectively.While the HRs are disordered coiled coils in the pre-fusion stage of the coronaviruses, they form a 6-helix bundle serving as the "fusion core" as observed in the post-fusion conformation of the virus, whereby three HR1s form a central 3-helical bundle, and three HR2s bind to the central core forming the second layer of the complex 38,39 .SAdLSA alignments indicate that Yongshi could mimic either HR1 or HR2, with somewhat higher similarity to HR1 than HR2.The latter has disordered terminal regions that lead to a relatively lower similarity score to Yongshi.Most importantly, Yongshi also exhibits a single helical motif and displays a similar hydrophobic pattern, matching its counterpart in HR1 or HR2.Therefore, we hypothesized that Yongshi could inhibit the SARS-CoV-2 fusion machinery by disrupting the critical HR1/HR2 complex via binding to the HR1 domain.This hypothesis is supported by previous observations that a peptide mimetic of the HR2 of the SARS virus is effective in inhibiting the virus 40,41 .
To evaluate the possibility that Yongshi might act as an HR1 or HR2 mimetic, we built respective structural models for both scenarios using the SAdLSA alignments (Fig. 6A) and a crystal structure of HR1/2 complex of SARS-CoV-2 39 .Figure 6B shows a structural model of two HR1s of SARS-CoV-2 complexed with one Yongshi peptide, replacing the third HR1 with the aligned Yongshi peptide.After energy minimization (see Methods), the complex is stabilized with Yongshi's hydrophobic residues fitting to the complementary hydrophobic core originally formed by HR1s.Eight positively charged lysines or arginines of Yongshi point outwards and form a charged surface patch, dramatically reducing the hydrophobic groove originally reserved for interacting with HR2s and potentially disrupts the HR1/HR2 interactions.Figure 6C shows the second structural model of three HR1s of SARS-CoV-2 in complex with three Yongshi peptides, by replacing three HR2s originally found in the template.Since there are several clashes between lysines and the hydrophobic core, we performed 100 ns molecular dynamics simulations to explore the stability of this putative complex.Figure 6C shows snapshots at the end of the MD simulation.The N-termini of the Yongshi peptides appear flexible and their movements eliminate the originally unfavorable charged-hydrophobic interactions.In contrast, the hydrophobic interactions between the short helix of Yongshi and HR1s are well-maintained throughout the simulations with a root mean square deviation below 3 Å.The modeling and MD simulation suggest that Yongshi could disrupt HR1 and HR2 interactions via two possible mechanisms by mimicking the HR1 helical core and/or competing for binding to HR1, both involving direct interactions with HR1s.We also note that the D950N mutation in the delta variant as well as the Q954H and N969K mutations in the omicron variant are unlikely to impede hypothetical interactions between Yongshi and HR1 since they occur outside the binding interface.

L-Yongshi but not D-Yongshi, binds heptad repeat 1 of SARS-CoV, MERS-CoV, and SARS-CoV-2.
While computational projections are promising, experimental analysis is critical for confirmation.Using BLASTp, we identified 111 coronavirus spike protein HR1 sequences that share homology with the HR1 of SARS-CoV-2 (Fig. 7) 28 .Out of these, we synthesized three biotinylated HR1 peptides (SARS-CoV-2 HR1, SARS-CoV-1 HR1 and MERS-CoV HR1) and tested their ability to bind SARS-CoV-2 HR2, L-Yongshi, and D-Yongshi peptides.We used bio-layer interferometry to measure the binding of immobilized HR1 peptides to serial dilutions (100-1.56μM) of SARS-CoV-2 HR2, L-Yongshi, and D-Yongshi (Fig. 8).As expected, SARS-CoV-2 HR2 bound to the HR1 of both SARS-CoV-2 and SARS-CoV-1 with a moderate affinity of 96.7 μM and 59.5 µM respectively but showed no binding to MERS-CoV HR1.Interestingly, L-Yongshi showed dose-dependent binding to all three HR1 peptides tested (SARS-CoV-1, MERS-CoV, and SARS-CoV-2), suggesting that Yongshi could potentially act as a pan-coronavirus entry inhibitor.Among the three HR1 peptides, L-Yongshi bound with the strongest relative affinity to the HR1 of SARS-CoV-1 (23.55 μM).L-Yongshi bound to SARS-CoV-2 HR1 with an approximately 4-fold stronger affinity (24.1 μM) than SARS-CoV-2 HR2 (96.7 μM), indicating that it could potentially outcompete HR2 for binding to HR1.As a control, we tested D-Yongshi for binding to all three HR1 peptides, and the D-enantiomer did not show binding to any HR1 peptides up to 100 μM.These binding experiments suggest that Yongshi can both specifically bind to the HR1 of coronaviruses and that it could outcompete HR2 for binding to HR1.The specificity of this interaction is dependent on some secondary structure as the chirality (L-vs.D-enantiomer) is critical for binding.Thus, we posit that Yongshi could act as a viral entry inhibitor by blocking the fusion machinery of coronaviruses.

Discussion
Here we identified a cathelicidin peptide of wild boar origin that inhibits SARS-CoV-2 and exhibits binding to the HR1 of SARS-CoV-2 spike protein.The specificity of this interaction is further supported by the inability of the mirror-image peptide, D-Yongshi to target SARS-CoV-2 or bind to the HR1 peptide.This interaction may contribute to the observed inhibition of SARS-CoV-2 but is likely only one of several mechanisms in addition to direct effects on the viral membrane or induction of an antiviral state in host cells.These results identify how Yongshi can act via both direct and indirect mechanisms, presenting multiple avenues for further engineering of peptide derivatives.Further, as SARS-CoV-2 has undergone significant evolution since our initial studies, we repeated our analysis with the recent XBB.1.16variant and find that Yongshi continues to provide substantial virus inhibition on par with the original wild-type virus (Fig. 9).
For millennia, host defense peptides have co-evolved with their hosts to best protect them against pathogens in their niche.Unlike antibodies or T cell receptors which are highly diverse, and which the individual host makes > 10 7 specificities, host defense peptides are produced in very limited numbers; for instance, humans produce a single cathelicidin LL-37.Yet these host defense peptides work because they often target an Achilles' heel or a conserved motif that is crucial for microbial survival.These targeted motifs are often common among multiple pathogens, even those that do not infect the given species.Because of this feature, if a pathogen contains the motif or element that the peptide acts upon, the host defense peptide can cross-neutralize that bacteria or virus, regardless of the species that encounter it.It is highly unlikely that Yongshi arose to combat SARS-CoV-2 as the virus only recently emerged, but instead Yongshi confers the wild boar protection against some related virus that bears a complementary HR1 peptide.In this context, Juergen Richt and colleagues have reported that five-weekold pigs are not susceptible to infection with SARS-CoV-2 infection.Briefly, they infected pigs with 10 6 TCID 50 dose of SARS-CoV-2 via oral, intranasal or intratracheal route and the virus did not replicate in pigs, and they did not produce antibodies against SARS-CoV-2 either 44 .Whether this protection is due to the Yongshi-related cathelicidin, PMAP-36 remains to be investigated.
Regarding the specific mechanism of HR2 mimicry, several other purpose-built HR2 mimetics have been previously described to inhibit SARS-CoV-2.Early HR2 mimetics designed against SARS-CoV-2 used the core HR2 region (amino acids 1163-1202) with additional amino acid substitutions or peptide modifications to achieve viral IC 50 values in the low micro-molar to high nano-molar range 39,[45][46][47][48] .Recently, a crystal structure of an extended HR1HR2 6-helix bundle led to the design of an extended HR2 mimetic which achieves an impressive IC 50 of 1.5nM against SARS-CoV-2.This activity was dependent on the expression of TMPRSS2 for receptor mediated entry and was maintained against alpha, delta, and omicron variant viruses with similar active concentrations 49 .It is not surprising that HR2 derived peptides can vastly outcompete Yongshi in performance as an HR2 mimetic, however, it is curious that Yongshi achieves a comparable IC 50 to the unmodified, core HR2 region which varies by model system and is reported between 0.98 µM and 33.74 µM [46][47][48] .Given that Yongshi also aligns closely with the heptad repeats of RSV and Mumps, we hypothesize that PMAP-36R may have evolved in response to multiple endemic porcine viruses to achieve broad heptad repeat compatibility.It is unclear whether PAMP-36R is expressed at the level necessary to achieve this activity in nature, but LL-37 has been reported in Figure 7. Phylogenetic tree of HR1 segments in disparate coronaviruses.111 coronavirus spike protein HR1 sequences that share homology with the HR1 of SARS-CoV-2 were identified by BLASTp 28 .Sequences were aligned using MUSCLE and the resultant phylogenetic tree was visualized with ggtree 42,43 .Three of the HR1 sequences (red) were analyzed in detail in panel B and they share a common ancestor sequence near the root of the tree.The inferred breadth of coverage is shaded (blue).The three HR1 peptides we tested for binding with Yongshi peptide are marked in red.humans at 20 µg/mL (~ 4.5 µM) in bronchoalveolar lavage fluid and up to 80 µg/mL (~ 17.8 µM) in nasal secretions.Notably these concentrations are attained in response to systemic and local inflammation, with healthy individuals producing significantly less peptide 50,51 .
In searching for the previous characterization of the Yongshi sequence, we identified that the Yongshi peptide is two amino-acids longer than the mature PMAP-36 cathelicidin, which lacks the N-terminal valine and undergoes a C-terminal amidation reaction which removes the terminal glycine 52 .Although PMAP-36 has not been www.nature.com/scientificreports/studied in the context of viral infection, multiple publications demonstrate a bacterial inhibition curve similar to the inhibition of SARS-CoV-2 by Yongshi.In the context of E. coli, Scheenstra and colleagues demonstrated a minimal bactericidal concentration for PMAP-36 in the range of 5-10 µM.Further, removing up to 11 N-terminal residues or mutation of cysteine to serine did not reduce E. coli killing potential but reduced cell toxicity as measured by both porcine RBC hemolysis and RAW264.7 cell mitochondrial activity in culture 53 .While our model describes a specific interaction between Yongshi and SARS-CoV-2, the parallel effects of mutagenesis against SARS-CoV-2 and E. coli suggest the mechanism of action has shared components whether the target is viral or bacterial.Alternatively, transmission electron micrographs of PMAP-36 treated E. coli document vesicle shedding, a bacterial stress response to lipid asymmetry 53,54 .Combined with our results, this observation may indicate that SARS-CoV-2 is susceptible to lipid asymmetries caused by Yongshi, but not the amphipathic mechanisms of other cathelicidins 55 .L-Yongshi, but not D-Yongshi, inhibits SARS-CoV-2 and its drifted variants.Amino acids can be produced as enantiomers, which are mirror images of each other.Of these two configurations, the L-enantiomer, in contrast to the D-enantiomer, is almost exclusively produced naturally.Because of this, host proteases, for the most part, only demonstrate enzymatic activity against the L-enantiomer of peptides and proteins.In contrast, D-enantiomer peptides and proteins are not susceptible to this natural degradation.However, since only the L-Yongshi demonstrated antiviral activity, we must formulate L-Yongshi appropriately, for therapeutic or prophylactic evaluation in vivo.While Yongshi possesses some inhibitory activity, further modifications to the peptide are necessary to generate a therapeutically viable product.Given the existence of purpose built HR2 mimetics with much lower IC 50 ranges than Yongshi, future designs should further explore the cellular effects of Yongshi or the ability to differentially disrupt viral and host membranes.Additionally, we note that administering Yongshi as a single bolus is sub-optimal in that it increases immediate cytotoxicity and has a short half-life for cellular protection.A better approach may be to design Yongshi as a synthetic cathelicidin to be delivered by an mRNA vector and thereby produced locally in situ.

Methods
Identification of zoonotic cathelicidins.Cathelicidin peptide sequences were identified using the Uni-Prot database and analogous peptide sequences were identified using the nBLAST databases.Sequences were analyzed for homology to the antimicrobial domain of Human Cathelicidin LL-37, as variation from LL-37 was preferred.Putative peptide sequences were modeled using I-TASSER for antimicrobial peptide-like secondary structural features, analyzed for net charge, and then searched for preexisting characterization 56 .Sequences with 11-50 amino acids that were not previously described for antiviral activity were chosen as candidates.

Synthesis of cathelicidin peptides and SARS-CoV-2 peptides.
Unlabeled and biotinylated zoonotic cathelicidin peptides and SARS-CoV-2 peptides were produced by standard Fmoc synthesis by Genemed Synthesis Incorporated.Peptides were purified to > 95% purity by HPLC as confirmed by mass-spectroscopy.Recombinant human LL-37 was purchased from Anaspec (Catalogue#: AS-61302).Recombinant OVA peptide was purchased from Invivogen (Catalogue#: vac-sin).For viral inhibition and cell toxicity assays, lyophilized peptides were first reconstituted in DMSO to a concentration of 10 mM before further dilution in DMEM or PBS.

SARS-CoV
Data visualization and analysis.Data visualization and analysis for SARS-CoV-2 peptide inhibition, human RBC hemolysis toxicity, and cell viability assays was performed using GraphPad Prism 9.2.0.Outliers were defined as greater than 1.5 the interquartile range above or below the boundary of the interquartile range and excluded from analysis.Graphical axes are cut at 0% and 100%, data points beyond this range are not visualized but are included in the calculation of the mean and standard error.IC 50 and TD 50 values were calculated by S-curve regression of the plotted data.SARS-CoV-2 peptide inhibition assay. 2 × 10 4 Vero E6 cells were seeded to each well of a poly-L lysine coated 96-well tissue culture plate (Corning, Catalogue#: 354516) and allowed to adhere overnight.The following day, SARS-CoV-2 virus was diluted in DMEM without FBS or P/S/A such that 50 µL contains 100 pfu of virus.50 µL of virus in DMEM was then combined with 50µL of peptide dilution in DMEM containing 2% Cell viability assay. 2 × 10 4 Vero E6 or HEK293T cells were seeded to each well of a 96-well tissue culture plate (CellTreat, Catalogue#: 229197) and allowed to adhere overnight.24 h later, cell culture media was aspirated and replaced with 100 µL peptide containing or control DMEM with 1% FBS and 1X P/S/A.Cells were then incubated at 37 °C and 5% CO 2 for 48 h.After incubation, 0% viability control wells received 1µL of NP-40 (Sigma-Aldrich, Catalogue#: I3021-50ML).5 min after the addition of NP-40, 20 µL of CellTiter 96-Aqueous One (Promega, Catalogue#: PAG3580) was added to each well and briefly mixed on an orbital shaker.Cells were incubated for 1 h at 37 °C and 5% CO 2 .Plates were briefly shaken to mix formazan products before the transfer of supernatants to a clear bottom 96-well assay plate (Costar, Catalogue#: 3912).Formazan concentrations were measured at 490 nm on a Biotek Synergy 2 Multi-Mode Microplate Reader.The percentage of viable cells was calculated by linear regression using DMEM-only and DMEM + 1% NP-40 as 100% and 0% cell viability controls, respectively.Statistical significance was calculated by two-way ANOVA with Bonferroni's correction.
Modeling of Yongshi and SARS-CoV-2 interactions.Initial complex models of HR1/Yongshi (2HR1/1Yongshi and 3HR1/3Yongshi) were built using SAdLSA sequence alignment and a crystal structure of HR1/HR2 complex (PDB 6LXT) as the template.Full atoms were modeled by Scwrl4 60 and VMD 61 .Each structure was explicitly solvated in a water box, neutralized with additional ions, and minimized for 2000 conjugate gradient steps with NAMD version 2.14 62 and CHARMM36 forcefields 63 .The 3HR1/3Yongshi complex was further equilibrated by 100 ps equilibration under constant pressure and temperature conditions (NPT) and followed by 100ns equilibration under constant volume and temperature conditions (NVT).An integration time step of 2fs was employed.Full electrostatics were computed using the particle-mesh Ewald (PME) method 64 .The calculations were done on a workstation with 24 Intel Xeon Gold 6226 CPU cores and 4 Nvidia RTX6000.Run time is 62 hours for the MD simulation.All proteins were diluted in a running buffer of 20 mM HEPES, 150 mM NaCl, 0.02% Tween-20, pH 7.4.Biotinylated HR1 peptides were loaded to streptavidin sensors at 5 μg/mL to a threshold of 1.5 nm.After washing in the running buffer, the loaded sensors were moved to serial dilutions of SARS-CoV-2 HR2, Yongshi, or D-Yongshi for 100 s of association and 200 s of dissociation in running buffer.Sensors were depleted of remaining bound molecules by 3 rounds of 5 s washings in 500 mM NaCl followed by 5 s washings in running buffer.
A heterogenous ligand binding model was fit to the collected data in Data Analysis HT software v11.1.1.39(Sartorius).Raw data and fittings were exported to GraphPad Prism v9.1.Approximate steady state calculations were performed by plotting response at the end of association to the concentration of protein used and fit to a hyperbola.

Figure 2 .Figure 3 .
Figure 2. The PMAP-36R cathelicidin derivative pSer possesses reduced cytotoxicity.(A) Human RBC hemolysis assay showing reduced lytic activity of pSer compared to PMAP-36R or LL-37.Peptides at the labeled concentrations were incubated with human RBCs for 1 h at 37 °C in 1X PBS buffer.RBC lysis was determined by the spectral absorbance (490 nm) of cell supernatants relative to PBS-only and TritonX-100 containing controls.(B-C) MTS formazan formation assay with (B) Vero hACE2, and (C) HEK-293 T cell lines showing reduced cell death in wells receiving pSer compared to PMAP36-R.Peptides at the labeled concentrations were incubated with cells in 1% FBS containing DMEM for 48 h prior to the addition of MTS substrate.The percentage of cell death was calculated relative to untreated and NP-40 receiving control wells.Results are representative of 3 independent experiments performed in triplicate.(D) MTS formazan formation assay with the Vero A2T2 cell line showing saturation of toxic effect at 100 µM.Peptides at the labeled concentrations were incubated with cells in 1% FBS containing DMEM for 24 h prior to the addition of MTS substrate.The percentage of cell death was calculated relative to untreated and NP-40 receiving control wells.Results are representative of 2 independent experiments performed in duplicate.Significance calculated by two-way ANOVA with Bonferonni's correction comparing pSer against control PMAP-36R (* < .05,** < .01,** < .001,* < .0001).

Figure 4 .
Figure 4.The D-enantiomer of Yongshi loses SARS-CoV-2 specificity.(A) Toxicity towards Vero hACE2 cells was evaluated by MTS formazan formation assay.Peptides at the labeled concentrations were incubated with cells in 1% FBS containing DMEM for 48 h prior to the addition of MTS substrate.The percentage of cell death was calculated relative to untreated and NP-40 receiving control wells.(B) Inhibition of SARS-CoV-2 infection of Vero hACE2 cells.Peptides at the labeled concentrations were pre-incubated with 100pfu of live SARS-CoV-2 virus (nCoV/USA_WA1/2020) for 1 h at 37 °C before addition to confluent Vero hACE2 cells in a 96-well plate.Infected cells were fixed and quantified by focus forming assay after 48 h.Inhibition of viral infection was calculated based on the percent area of each well staining positively for viral spike protein compared to control wells without peptide inhibitor treatment.Results are representative of 3 independent experiments performed in triplicate.Data for L-Yongshi reproduced from Figs. 1 and 2. Significance calculated by two-way ANOVA with Bonferonni's correction comparing D-Yongshi against control L-Yongshi (* < .05,** < .01,** < .001,* < .0001).

Figure 5 .
Figure 5. Yongshi retains inhibitory activity against emergent SARS-CoV-2 variants alpha, beta, gamma, kappa, and delta.Dilutions of the Yongshi peptide were tested for inhibition of SARS-CoV-2 and its drifted variants.Peptides at the labeled concentrations were pre-incubated with 100pfu of the indicated SARS-CoV-2 variant virus for 1 h at 37 °C before addition to confluent Vero hACE2 cells in a 96-well plate.Infected cells were fixed and quantified by focus forming assay after 48 h.Inhibition of viral infection was calculated based on the percent area of each well staining positively for viral spike protein compared to control wells without peptide inhibitor treatment.Results are representative of 2 to 3 independent experiments performed in triplicate.Significance calculated by two-way ANOVA with Bonferonni's correction comparing A.1 inhibition against each variant (* < .05,** < .01,** < .001,* < .0001).

Figure 6 .
Figure 6.Deep-learning sequence alignment algorithm and computational modeling predict stable binding interactions between Yongshi and HR1.(A) Sequence alignments between Yongshi and HR1/HR2 sequences of SARS-CoV-1, SARS-CoV-2, and MERS-CoV by SAdLSA 36 .< D > denotes the mean of predicted distance between aligned residues, and SeqID is Sequence Identity.Identical and statistically favorable amino acid substitutions are denoted by "*" and "+", respectively.The digits below each alignment indicate the probability that the predicted structure for each aligned amino acid pair will be less than 3 Å, e.g., the number 6 predicts a probability score between 60 to 70%.Note that SARS-CoV-1 and SARS-CoV-2 have identical sequences in their HR2 regions aligned to Yongshi.(B) A structural model of two SARS-CoV-2 HR1s in complex with a single Yongshi peptide.The two HR1s are shown in the surface representation, and the Yongshi peptide is shown in ribbon backbone and licorice sidechain representations in the upper snapshot, and the same surface representation in the lower snapshot.The lower snapshot is slightly rotated from the upper snapshot to show a large positively charged surface patch of Yongshi as pointed by a yellow arrow.The color codes for amino acids are hydrophobic (white), polar (green), positive (blue), negative (red).(C) A structural model of three SARS-CoV-2 HR1s in complex with three Yongshi peptides after 100 ns of MD simulation.The upper snapshot is shown in cartoon representation for three HR1s (blue, cyan, grey), and ribbon representation for Yongshi (yellow).The lower snapshot is in the same orientation as the upper one but shown in the same representation as snapshots in (B). https://doi.org/10.1038/s41598-023-41850-7

Figure 8 .Figure 9 .
Figure 8. Yongshi binds to HR1 peptide with higher affinity than HR2.Biotinylated HR1 peptide from SARS-CoV-2, SARS-CoV-1, or MERS-CoV was bound to a streptavidin-coated bio-layer interferometry sensor and incubated with serial dilutions of either SARS-CoV-2 HR2, Yongshi, or D-Yongshi starting from 100 μM.(A) Representative plots of peptide binding over 100 s of association followed by 200 s of dissociation.(B) Regression analysis of the peptide binding response at each concentration after 100 s of association.Relative steady state affinities were derived from the K D of the hyperbola.

Table 1 .
Calculated IC50and TD 50 values for PMAP-36R derivatives and control LL-37 or OVA peptides.IC 50 and TD 50 values were calculated for each peptide by S-curve regression.

Table 2 .
Calculated IC 50 values for Yongshi against drifted SARS-CoV-2 variants.IC 50 values were calculated for each variant by S-curve regression.*The IC 50 of A.1 is calculated from a set of experiments performed at the same time as the other variants and therefore varies slightly from the IC 50 presented in Table 1.Vol:.(1234567890) Scientific Reports | (2023) 13:14650 | https://doi.org/10.1038/s41598-023-41850-7 FBS and 2X P/S/A and incubated at 37 °C and 5% CO 2 for 1 h.Media was then aspirated from Vero E6 cultures and replaced with 100µL of virus-containing media per well.Plates were then incubated at 37 °C and 5% CO 2 for an additional 48 h.Cells were fixed by removal of virus media and incubation with 2% paraformaldehyde in PBS.Cells were permeabilized by 1X PBS + 0.1% saponin (Sigma, Catalogue#: 47036-50G-F) and 0.1% BSA (Sigma, Catalogue#: A2153-500G).Cross-reactive SARS-CoV-1 clone CR3022 monoclonal antibody (Abcam, Catalogue#: ab273073-200ug) was used to detect SARS-CoV-2 spike protein expression in combination with Goat anti-human IgG-HRP secondary antibody (SouthernBiotech, Catalogue#: 2045-05).Virus infected cells were visualized by adding TruBlue Peroxidase Substrate (SeraCare, Catalogue#: 50-78-02) and the plates were scanned and the spots counted using a custom Matlab script we developed in house.All infection assays were executed in a BSL-3 lab environment following appropriate handling and disposal guidelines.Cells were briefly mixed on an orbital shaker then incubated at 37 °C for 1 h with 5% CO 2 .After incubation, cells were pelleted in a benchtop centrifuge at 500 rcf for 3 min and supernatants were collected for reading at 490 nm on a Biotek Synergy 2 Multi-Mode Microplate Reader.The percentage of RBC lysis was calculated by linear regression using the PBS and PBS + 0.1% TritonX-100 as 0% and 100% lysis controls, respectively.Statistical significance was calculated by two-way ANOVA with Bonferroni's correction.