Smaller, Stronger, More Stable: Peptide Variants of a SARS-CoV-2 Neutralizing Miniprotein

Based on the structure of a de novo designed miniprotein (LCB1) in complex with the receptor binding domain (RBD) of the SARS-CoV-2 spike protein, we have generated and characterized truncated peptide variants of LCB1, which present only two of the three LCB1 helices, and which fully retained the virus neutralizing potency against different SARS-CoV-2 variants of concern (VOC). This antiviral activity was even 10-fold stronger for a cyclic variant of the two-helix peptides, as compared to the full-length peptide. Furthermore, the proteolytic stability of the cyclic peptide was substantially improved, rendering it a better potential candidate for SARS-CoV-2 therapy. In a more mechanistic approach, the peptides also served as tools to dissect the role of individual mutations in the RBD for the susceptibility of the resulting virus variants to neutralization by the peptides. As the peptides reported here were generated through chemical synthesis, rather than recombinant protein expression, they are amenable to further chemical modification, including the incorporation of a wide range of non-proteinogenic amino acids, with the aim to further stabilize the peptides against proteolytic degradation, as well as to improve the strength, as well the breadth, of their virus neutralizing capacity.


Introduction
As of May 2022, the SARS-CoV-2 pandemic has caused more than six million deaths worldwide (https://covid19.who.int/, accessed on 20 May 2022), clearly demonstrating the urgent need for efficient therapeutic and preventive measures to combat the disease and prevent infection with the virus, respectively. In the early phase of the pandemic, therapeutic options in the treatment of COVID-19 were largely limited to alleviating symptoms, mainly through anti-inflammatory intervention [1,2]. Complementary to vaccination, which is considered the most effective means of protecting against severe COVID-19, current therapeutic strategies against SARS-CoV-2 infection include a still low, but growing, number of drugs specifically targeting the virus. The first therapeutics approved for clinical use were monoclonal antibodies that recognize the spike protein, through which the virus contacts its receptor ACE2 expressed on lung epithelial cells [3,4]. Thus, binding of the antibodies to the spike protein prevents attachment of the virus to its host cell [5,6]. In addition, two orally available small molecules have recently been introduced as anti-SARS-CoV-2 drugs. Molnupiravir (Lagevrio ® ) increases the frequency of viral RNA mutations,

Sequence Truncation and Cyclization
The 55 amino acid miniprotein LCB1 [19] had been designed through docking of large in silico miniprotein libraries against the RBD structure [26]. LCB1 folds into three consecutive α-helices, two of which interact with RBD, as evident from the cryo-electron microscopy structure of the LCB1-RBD complex (pdb code 7JZU). Addressing the question of whether all three helices of LCB1 are required for its interaction with RBD, we synthesized truncated variants that present helices 1 and 2 (LW25.1), as well as individual helix 1 (LW25.5) and helix 2 (LW25.6), respectively. These peptides, along with an N-terminally acetylated and C-terminally amidated variant of full-length LCB1 (LW25.3), were tested for their ability to inhibit the interaction of soluble recombinant ACE2 with RBD. Interestingly, the inhibitory activity of LW25.3 was essentially preserved in the two-helix peptide LW25.1 ( Figure 1, Table 1), indicating that helix 3 is not required for its inhibitory activity. The individual helices (LW25.5 and LW25.6), on the other hand, were completely inactive, confirming the contribution of both helices to the interaction with RBD ( Figure 1, Table 1). As a further step, C-terminal truncation of helix 2 in the two-helix peptide LW25.1 resulted in the identification of a 35-mer peptide (LW32.4), which fully retained the inhibitory activity of LW25.3 (Table 1). These results were confirmed in a neutralization assay using simian immunodeficiency virus (SIV) pseudotyped with the SARS-CoV-2 spike protein [27], in conjunction with A549 cells expressing human ACE2 and TMPRSS2. In this assay, both LW32.4 and LW25.1, as well as LW25.3, were shown to efficiently inhibit infection ( Figure 1, Table 2), demonstrating the successful size reduction of LCB1 by approximately 30%. peptide LW25.1 resulted in the identification of a 35-mer peptide (LW32.4), which fully retained the inhibitory activity of LW25.3 (Table 1). These results were confirmed in a neutralization assay using simian immunodeficiency virus (SIV) pseudotyped with the SARS-CoV-2 spike protein [27], in conjunction with A549 cells expressing human ACE2 and TMPRSS2. In this assay, both LW32.4 and LW25.1, as well as LW25.3, were shown to efficiently inhibit infection ( Figure 1, Table 2), demonstrating the successful size reduction of LCB1 by approximately 30%.     For the characterization of the interaction of the two-helix peptide LW32.4 with RBD at the level of individual amino acid residues, we generated an alanine scan, i.e., a set of peptide variants, in which all positions of the peptide were individually replaced with alanine. These peptides provided information on the contribution of each amino acid to the interaction with RBD. To compare the energy profiles of LW32.4 and LCB1, the experimental alanine scan data for LW32.4 were then matched with the results of an in silico alanine scan of LCB1. Testing the alanine scan peptides for inhibition of the ACE2-RBD interaction (Figure 2A) identified six residues, i.e., L6, M13, G20, M26, D30 and Y33, whose replacement with alanine resulted in more than 50% loss of inhibitory activity (red bars). All these positions are involved in the LCB1-RBD interface, according to the complex structure (pdb code 7JZU). Additional contact residues, however, including E3, Q7, Y10 and E23, can apparently be replaced with alanine without significant loss of affinity to RBD. Most of these residues (E3, Q7, E23) form specific hydrogen bonds with RBD in the complex structure 7JZU, suggesting that these interactions may be less stable in the shorter peptide LW32.4. The hot spot residues identified in the experimental alanine scan were also found in silico ( Figure 2B), in addition to two more contact residues (Q7 and Y10), whose replacement with alanine was proposed to greatly destabilize the LCB1-RBD complex. In summary, the results obtained using the experimental alanine scan of the truncated peptide LW32.4, in comparison with the computational alanine scan of LCB1, strongly indicate a similar mechanism of interaction with RBD, which is the molecular basis of the inhibitory effect of these peptides on SARS-CoV-2 infection.
The 3D structure of the ACE2-LCB1 complex shows that the three LCB1 helices are interconnected by short loops composed of one (helix 1-2) and four (helix 2-3) amino acids, respectively, and aligned in an antiparallel fashion. Hence, the N-terminus of helix 1 is adjacent to the C-terminus of helix 2. This proximity is most pronounced between residues K2 and G39, which are only 8.7 Å apart. We, therefore, thought to stabilize this proximity covalently by replacing K2 and G39 in LW25.1 with cysteine and introducing a disulfide bridge between these two cysteine residues (peptide LW25.13). Interestingly, the inhibitory activity of this conformationally constrained peptide on the ACE2-RBD interaction was found to be approximately 20-fold and 9-fold, respectively, stronger than that of LW25.3 and LW32.4 (Table 1). A similar improvement was found for the virus neutralization capacity of LW25.13 ( Table 2), indicating that introducing the covalent bridge between the N-and C-termini of the peptide had a stabilizing effect on its bioactive conformation. scan were also found in silico ( Figure 2B), in addition to two more contact residues (Q7 and Y10), whose replacement with alanine was proposed to greatly destabilize the LCB1-RBD complex. In summary, the results obtained using the experimental alanine scan of the truncated peptide LW32.4, in comparison with the computational alanine scan of LCB1, strongly indicate a similar mechanism of interaction with RBD, which is the molecular basis of the inhibitory effect of these peptides on SARS-CoV-2 infection. The 3D structure of the ACE2-LCB1 complex shows that the three LCB1 helices are interconnected by short loops composed of one (helix 1-2) and four (helix 2-3) amino acids, respectively, and aligned in an antiparallel fashion. Hence, the N-terminus of helix 1 is adjacent to the C-terminus of helix 2. This proximity is most pronounced between residues K2 and G39, which are only 8.7 Å apart. We, therefore, thought to stabilize this proximity covalently by replacing K2 and G39 in LW25.1 with cysteine and introducing a disulfide bridge between these two cysteine residues (peptide LW25.13). Interestingly, the

Virus Variant Selectivity
In order to assess the breadth of the antiviral activity of LW25.3, LW32.4 and LW25.13, the peptides were tested in the ACE2-RBD inhibition and virus neutralization assays, respectively, using RBD proteins and pseudovirus from wild type (wt) virus, as well as the alpha, beta, delta and omicron variants of SARS-CoV-2. Interestingly, while the delta variant was susceptible to inhibition by LW25.3, LW32.4 and LW25.13, the beta and omicron variants were found to be largely resistant against all three peptides ( Table 2). These two virus variants share specific mutations in the RBD, i.e., K417N/T, E484K/A and N501Y [28,29], suggesting that these positions are involved and critical for the interaction with LW25.3 and LW32.4, respectively. In fact, the structure of the LCB1-RBD complex indicates contacts of LCB1 with K417 and N501 of RBD. The only discrepancy in the inhibitory activity among the peptides was observed for the alpha virus variant, which, as previously reported for LCB1 [22,23], was well neutralized by LW25.3, but not by LW32.4. Re-introducing two positions (M36 and K37) at the C-terminus of the truncated peptide LW32.4, however, completely restored the inhibitory activity against the alpha variant (peptide LW32.2). Likewise, the cyclic peptide LW25.13, which also contains M36 and K37, was shown to neutralize the alpha variant at low nanomolar concentrations as well.
For a detailed analysis of the relevance of the above-mentioned mutations in the RBD of various virus variants, these mutations were introduced either individually, or in combination, into the spike protein of wt SARS-CoV-2. The inhibitory effect of LW25.3, LW 32.2 and LW25.13 on pseudoviruses expressing these spike protein variants was then used to gauge the consequences of individual mutations, or combinations thereof ( Figure 3). Interestingly, the effect of individual mutations E484K and N501N, respectively, on the neutralizing capacity of the peptides, appears to be negligible, as the IC 50 values for these variants are not, or only slightly, different from those for wt virus. A combination of E484K and N501N resulted in an approximately three-fold (LW25.13), nine-fold (LW32.2) and 1.4fold (LW25.3), respectively, reduction in inhibitory activity, which, however, was still in the low nanomolar range. A stronger effect was seen for the mutation K417N, which is present in the beta and omicron virus variants, and which decreased the susceptibility of the virus to inhibition by the peptides by a factor ranging from six (LW25.3) to 25 (LW32.2). The strong effect of the K417N exchange may be a consequence of the loss of a salt-bridge between K417 of RBD and D30 of LCB1-derived peptides. This effect was even more dramatic when K417N was combined with either E484K or N501N, or both, as virus variants containing these combinations of mutations could no longer efficiently be neutralized by any of the peptides. While the K417N + E484K variant was still neutralized, albeit with eight-to 64-fold higher IC 50 values, the K417N + N501Y and K417N + E48K + N501Y variants were largely resistant to all three peptides, confirming the critical role of the K417N mutation, which, in combination with N501Y, abrogates susceptibility towards inhibition by the peptides. These two mutations are present in the beta and omicron variants of SARS-CoV-2, which may explain the resistance of these two variants against the peptides. Furthermore, the data illustrate the notion that it takes more than one mutation in the spike protein to confer a change in susceptibility of a virus variant for an inhibitor. Apart from that, it should also be noted that the cyclic peptide LW25.13 is consistently the most potent inhibitor among the tested peptides. These two mutations are present in the beta and omicron variants of SARS-CoV-2, which may explain the resistance of these two variants against the peptides. Furthermore, the data illustrate the notion that it takes more than one mutation in the spike protein to confer a change in susceptibility of a virus variant for an inhibitor. Apart from that, it should also be noted that the cyclic peptide LW25.13 is consistently the most potent inhibitor among the tested peptides. In conclusion, introducing a range of individual mutations, as well as combinations thereof, in the spike protein of SARS-CoV-2, in conjunction with our peptide inhibitors, we were able to dissect the contribution of these mutations to the loss of susceptibility to inhibition by the peptides, which helps elucidate the molecular basis of susceptibility of the various known SARS-CoV-2 variants to antiviral molecules.

Proteolytic Stability
Oral availability of protein and peptide drugs is generally hampered by their inherent susceptibility to proteolytic degradation, as well as poor absorption due to their molecular size [30,31]. Therefore, these molecules are typically administered through intravenous, intramuscular or subcutaneous injection. Alternative parenteral administration routes include oral inhalation, which is preferred for the delivery of drugs to the lungs, as it results in high tissue levels shortly after application [31][32][33]. As SARS-CoV-2 primarily infects cells in the upper respiratory system and the lungs, an application of potential antiviral drugs for the treatment of SARS-CoV-2 infections through inhalation appears beneficial. While proteases are far less abundant in lung tissue, as compared to the gastro-intestinal tract, some proteases are also expressed by lung cells, including the primary cellular receptors of SARS-CoV-2, i.e., ACE2 and TMPRSS2, which are proteases presented on the surface of lung epithelial cells [34][35][36]. Furthermore, the aspartic endopeptidase cathepsin D is considered the main protease produced in lung tissue [36,37]. As the substrate specificity of cathepsin D is very similar to that of pepsin, which In conclusion, introducing a range of individual mutations, as well as combinations thereof, in the spike protein of SARS-CoV-2, in conjunction with our peptide inhibitors, we were able to dissect the contribution of these mutations to the loss of susceptibility to inhibition by the peptides, which helps elucidate the molecular basis of susceptibility of the various known SARS-CoV-2 variants to antiviral molecules.

Proteolytic Stability
Oral availability of protein and peptide drugs is generally hampered by their inherent susceptibility to proteolytic degradation, as well as poor absorption due to their molecular size [30,31]. Therefore, these molecules are typically administered through intravenous, intramuscular or subcutaneous injection. Alternative parenteral administration routes include oral inhalation, which is preferred for the delivery of drugs to the lungs, as it results in high tissue levels shortly after application [31][32][33]. As SARS-CoV-2 primarily infects cells in the upper respiratory system and the lungs, an application of potential antiviral drugs for the treatment of SARS-CoV-2 infections through inhalation appears beneficial. While proteases are far less abundant in lung tissue, as compared to the gastro-intestinal tract, some proteases are also expressed by lung cells, including the primary cellular receptors of SARS-CoV-2, i.e., ACE2 and TMPRSS2, which are proteases presented on the surface of lung epithelial cells [34][35][36]. Furthermore, the aspartic endopeptidase cathepsin D is considered the main protease produced in lung tissue [36,37]. As the substrate specificity of cathepsin D is very similar to that of pepsin, which is far less expensive than cathepsin D, and is, therefore, often used as a cathepsin D surrogate in biomedical research [38][39][40][41], we examined the susceptibility of the linear peptide LW32.4 and the cyclic variant LW25.13 to proteolytic degradation by pepsin. Based on the preference of pepsin for peptide bonds with hydrophobic/aromatic amino acids at the P1 position, the sequence of the peptides contain a range of potential pepsin cleavage sites. Interestingly, though, LW32.4 was hydrolyzed only at position L31 ( Figures 4A and S2). More importantly, the cyclic peptide LW25.13 was shown to be largely stable towards pepsin for at least 60 min ( Figure 4B), illustrating that cyclization of peptides can protect against degradation not only by exopeptidases, but also endopeptidases, such as pepsin. is far less expensive than cathepsin D, and is, therefore, often used as a cathepsin D surrogate in biomedical research [38][39][40][41], we examined the susceptibility of the linear peptide LW32.4 and the cyclic variant LW25.13 to proteolytic degradation by pepsin. Based on the preference of pepsin for peptide bonds with hydrophobic/aromatic amino acids at the P1 position, the sequence of the peptides contain a range of potential pepsin cleavage sites. Interestingly, though, LW32.4 was hydrolyzed only at position L31 ( Figures 4A and S2). More importantly, the cyclic peptide LW25.13 was shown to be largely stable towards pepsin for at least 60 min ( Figure 4B), illustrating that cyclization of peptides can protect against degradation not only by exopeptidases, but also endopeptidases, such as pepsin. Apparently, the more compact structure of the cyclic peptide LW25.13, as compared to LW32.4, shields the L31-I32 peptide bond in LW25.13, stabilizing it against attack by pepsin. This effect was also seen for neutrophil elastase, a protease expressed in inflamed lung tissue [36,42,43], which rapidly hydrolyzed LW32.4 ( Figures 4C and S3), while LW25.13 was still largely intact after 60 min exposure to the enzyme ( Figure 4D). Based on these data, LW25.13 may be a better candidate, compared to the linear peptide LW32.4, for the development of a novel therapeutic approach against SARS-CoV-2 infection.
Plates were developed as described in 3.2. Percentage inhibition values were calculated using the following formula: in which blank1 is a sample without spike RBD, reference is a sample without peptide, and blank2 is a sample with neither spike protein RBD, nor peptide. The program Origin Version 2020 (OriginLab Corporation) was used for curve fitting and calculation of IC 50 values.

Virus Neutralization Assay/Generation of Virus Mutants
Generation of mutated SARS-CoV-2 Spikes: Point mutations were introduced into the wt spike expression plasmid by polymerase chain reaction (PCR). For that purpose, the wt spike DNA sequence was amplified by the Phusion ® High-Fidelity polymerase (New England Biolabs) using overlap extension of specific primers introducing the K417N, E484K or N501Y single mutations. After purification, the PCR fragments were digested with AgeI and Bsu36I (New England Biolabs) restriction enzymes and digested PCR products were gel purified and ligated into the wt spike expression plasmid using T4 DNA Ligase (ThermoFisher Scientific), according to the manufacturer recommendations. The multi-ple mutated SARS-CoV-2 spikes K417N + E484K, K417N + N501Y, E484K + N501Y and K417N + E484K + N501Y were generated by restriction enzymes digestion of the single mutated spike and ligated into the corresponding expression plasmids. XL-Gold ® Ultracompetent bacteria (Stratagene) were transformed with the ligation reactions and single colonies were screened for the respective mutations in the region encoding spike. All expected nucleotide substitutions in the wt spike DNA sequence were confirmed by sequencing.
Pseudoviruses neutralization assay: The neutralizing activity of the peptide variants was investigated by a lentiviral pseudotype assay, essentially as described previously [27,44]. In brief, pseudoviruses were produced by co-transfecting HEK293T cells with an expression plasmid encoding the respective spike variants, an SIV-based vector construct encoding luciferase and the SIV-based packaging plasmid. The supernatants of transfected cells containing the pseudoviruses were harvested, spun down for 5 min at 2000× g to pellet cell debris, 0.45 µM sterile filtered and stored at −20 • C for further use in neutralization assays. To investigate the peptide variants neutralization potency, 1.5 × 10 4 A549 cells overexpressing human ACE2 and TMPRSS2 protease were seeded in 150 µL D10 medium into flat bottom 96-well plates one day before transduction. Serial dilutions of the peptide variants were performed in 60 µL D1.5 medium, followed by the addition of 60 µL of the respective lentiviral pseudovirus stock and incubated for 1 h at 37 • C. Subsequently, the previously seeded A549 cells were washed once with 200 µL phosphate buffer saline (PBS) per well, followed by the addition of 100 µL of the pseudoviruses: peptides mix and A549 cells were incubated for 48 h in the CO 2 incubator at 37 • C. Finally, transduced A549 cells were washed twice with 200 µL PBS per well, and incubated with 50 µL ONE-Glo™ reagent (Promega) in 100 µL PBS for 3 min at RT. The luciferase intensity was, finally, measured on a microplate luminometer (VICTOR X5 from PerkinElmer, Rodgau, Germany). The luciferase activity from cells transduced with SIV pseudotyped particles lacking surface envelopes was subtracted as background from all sample values, while the luciferase intensity of cells transduced with pseudoviruses in absence of inhibitory molecules was considered as the reference value of transduction.

In Silico Alanine Scan
The PDB structure of the LCB1-RBD complex (7JZU) was pre-processed utilizing the pdb-tools software kit [45] and the optimize tool of the FoldX forcefield [46]. The Positionscan Mutagenesis (PSSM) tool of FoldX [47] was used to mutate every residue to Ala and calculate the ∆∆G of interaction energy between protein and ligand upon mutation. Large negative ∆∆G differences correspond to interface residues, for which a replacement by alanine results in a significant destabilization of the protein-protein interface [48].

Conclusions
Based on a previously reported, computationally designed three-helix miniprotein (LCB1), we have generated and characterized truncated and conformationally constrained peptide variants of LCB1. The inhibitory effect on the interaction of the receptor binding domain (RBD) of the SARS-CoV-2 spike protein with ACE2, as well as the SARS-CoV-2 neutralizing potency, of the peptide presenting full length LCB1 (LW25.3) was shown to be maintained in LW25.1 and LW32.4, which present only two of the three helices of LCB1.
Moreover, the RBD-ACE2 inhibitory, as well as virus neutralizing capacity, of a cyclic variant (LW25.13) of the truncated peptides was found to be approximately ten-fold stronger than that of LW25.3. This cyclic peptide was also shown to be considerably more stable towards proteases, such as pepsin and neutrophil elastase, as compared to the linear peptide LW32.4. Taken together, these data illustrate how the activity, as well as stability, of bioactive peptides can be improved through chemical modification, such as truncation and cyclization.
Furthermore, using these peptides in conjunction with synthetic virus variants containing individual mutations, or combinations thereof, we were able to dissect the role of these mutations, which are present in various virus variants of concern, to the susceptibility of the virus to neutralization by the peptides.
Finally, as the peptides reported here were generated through chemical synthesis, rather than recombinant protein synthesis, they are amenable to further chemical modification, including the incorporation of a wide range of non-proteinogenic amino acids. Such modifications can be expected to further stabilize the peptides against proteolytic degradation, as well as to improve the strength, and the breadth, of their virus neutralizing capacity.