(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Biosynthetic proteins targeting the SARS-CoV-2 spike as anti-virals [1]

['Stéphanie Thébault', 'Unité De Virologie Et Immunologie Moléculaires', 'Inrae', 'Université Paris-Saclay', 'Jouy-En-Josas', 'Nathalie Lejal', 'Alexis Dogliani', 'Centre National De La Recherche Scientifique', 'Architecture Et Fonction Des Macromolécules Biologiques', 'Umr']

Date: 2022-11 

The binding of the SARS-CoV-2 spike to angiotensin-converting enzyme 2 (ACE2) promotes virus entry into the cell. Targeting this interaction represents a promising strategy to generate antivirals. By screening a phage-display library of biosynthetic protein sequences build on a rigid alpha-helicoidal HEAT-like scaffold (named αReps), we selected candidates recognizing the spike receptor binding domain (RBD). Two of them (F9 and C2) bind the RBD with affinities in the nM range, displaying neutralisation activity in vitro and recognizing distinct sites, F9 overlapping the ACE2 binding motif. The F9-C2 fusion protein and a trivalent αRep form (C2-foldon) display 0.1 nM affinities and EC 50 of 8–18 nM for neutralization of SARS-CoV-2. In hamsters, F9-C2 instillation in the nasal cavity before or during infections effectively reduced the replication of a SARS-CoV-2 strain harbouring the D614G mutation in the nasal epithelium. Furthermore, F9-C2 and/or C2-foldon effectively neutralized SARS-CoV-2 variants (including delta and omicron variants) with EC 50 values ranging from 13 to 32 nM. With their high stability and their high potency against SARS-CoV-2 variants, αReps provide a promising tool for SARS-CoV-2 therapeutics to target the nasal cavity and mitigate virus dissemination in the proximal environment.

The entry of SARS-CoV-2 in permissive cells is mediated by the binding of its spike to angiotensin-converting enzyme 2 (ACE2) on the cell surface. To select ligands able to block this interaction, we screened a library of phages encoding biosynthetic proteins (named αReps) for binding to its receptor binding domain (RBD). Two of them were able to bind the RBD with high affinity and block efficiently the virus entry in cultured cells. Assembled αReps through covalent or non-covalent linkages blocked virus entry at lower concentration than their precursors (with around 20-fold activity increase for a trimeric αRep). These αReps derivates neutralize efficiently SARS-CoV-2 β, γ, δ and Omicron virus variants. Instillation of an αRep dimer in the nasal cavity effectively reduced virus replication in the hamster model of SARS-CoV-2 and pathogenicity.

Funding: B.D. was supported by the Agence Nationale de la Recherche (ANR) and by the Fédération pour la recherche médicale (ANR 20 Flash Covid 19 – FRM program). S.T. received salary from the Fédération pour la recherche médicale. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © 2022 Thébault et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Here, we first obtained a series of αReps specific of the receptor binding domain of the spike of SARS-CoV-2. These ligands display high affinities and blocked SARS-CoV-2 infection in vitro. The assembly of αRep through covalent and non-covalent linkages lowers the neutralisation EC50 to the 10 nM range. The αRep F9-C2 fusion protein instilled in the nose was found to limit virus replication and inflammatory response in a hamster model of SARS-CoV-2 infection. Furthermore, the F9-C2 fusion protein and a C2 homotrimer were found as potent inhibitors of SARS-CoV-2 variants including the antigenically distant omicron variant.

As for all coronaviruses, the SARS-CoV-2 spike (S) protein mediates virus entry to permissive cells. The S protein is a trimeric class 1 fusion protein that binds to its cell receptor, angiotensin converting enzyme 2 (ACE2), before undergoing a dramatic structural rearrangement to fuse the host-cell membrane with the viral membrane [ 12 ], [ 13 ]. Fusion is triggered when the S1 subunit binds to a host-cell receptor via its receptor binding domain (RBD). In order to bind to the receptor, the RBD undergoes articulated movements that transiently expose or hide its surface associated to the binding to ACE2 [ 14 ]. The two states are referred to as the “down” and the “up” conformations, in which down corresponds to a state incompetent to receptor binding and up to a state allowing receptor recognition. Due to its key function in the virus cycle, the RBD represents a target to identify binders that block interaction with the host-cell receptor or movements of the RBD between the down to up conformations [ 15 ].

Screening an αReps phage library allowed the identification of several binders specific of the RBD of the SARS-CoV-2 spike protein. Their binding affinity for the S1 domain was measured by biolayer interferometry. The neutralization activity of selected αReps was evaluated using a pseudo-typed S SARS-CoV-2 neutralization assay and a SARS-CoV-2 infection assay. Competitive binding assays were carried out by BLI to identify αReps recognizing non-overlapping binding sites. Then, αRep derived constructs followed the same characterization steps than their single counterparts. The protective potency of the best candidate was analyzed in vivo in the golden Syrian hamster model.

As an alternative approach to VHH and antibodies, a family of biosynthetic proteins, named αRep, was designed to provide a hypervariable surface on αRep variants ( Fig 1 ) [ 8 ]. αReps are thermostable proteins constituted by alpha-helicoidal HEAT-like repeats (31-amino acids long) commonly found in eukaryotes [ 9 ] and prokaryotes [ 10 ], including thermophiles. Sequences of homologs form a sharply contrasted sequence profile in which most positions are occupied by conserved amino acids whereas other positions appear highly variable generating a versatile binding surface. A large αRep library has been assembled and was demonstrated on a wide range of unrelated protein targets to be a generic source of tight and specific binders. Thus, αReps were previously selected as interactors of HIV-1 nucleocapsid and to negatively interfere with virus maturation [ 11 ].

Series of human neutralizing monoclonal IgG antibodies and nanobodies/VHH fused to a Fc IgG domain able to inhibit SARS-CoV-2 infection have been produced and tested for systemic treatments ([ 4 – 6 ]), but their efficacy by delivery in the nose may not be optimal; their firmness upon nebulization and aerosolization being a main issue for their use as therapeutics. Furthermore, their large-scale production should be economically not affordable in eucaryotic systems and technically difficult to achieve in prokaryotes [ 7 ].

With up to 6 million deaths worldwide in less than two years, the COVID-19 crisis has demonstrated the necessity to better understand and fight the spread and transmission of respiratory viruses. Such knowledge will help to develop new efficient anti-viral strategies to mitigate future epidemics and pandemics. SARS-CoV-2 infection starts in the nasal cavity, the virus replicating at high titres in the olfactory epithelia before reaching the lower respiratory tract where it induces the main pathology [ 1 ]. Infection of the olfactory epithelium leads to massive damage which may explain the high prevalence of smell loss (anosmia) during the COVID-19 pandemic and to environmental dissemination to infect conspecifics [ 2 ], [ 3 ]. Blocking virus multiplication with antivirals delivered in the nose and the upper respiratory tract might therefore allow therapeutic benefit and prophylactic protection.

Results

Selection of αReps binders of the SARS-CoV-2 receptor binding domain An overview of the selection process to generate anti-SARS-CoV-2 αReps specific of the spike is shown in Fig 1. In order to select binders blocking SARS-CoV-2 entry into cells, the RBD (amino acids 330 to 550 of the spike S sequence) was used as a bait for screening. The phage display procedure included three rounds of panning followed by a screening step by phage-ELISA on individual clones. Nucleotide sequencing allowed the identification of >20 independent clones that were retained for further analyses (selected αRep sequences are listed in S1 Fig). His-tagged versions of the anti-RBD αRep were expressed in E. coli and purified. We first explored their affinity for the RBD by biolayer interferometry (BLI) at different concentrations to determine their kinetic rate constants. Fig 2 shows the binding of two most potent anti-RBD ligands, αReps C2 and F9. Their affinity for the RBD was about 0.3 and 1.1 nM, respectively. αRep C7 exhibited an affinity in the 10 nM range. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Selection of αReps based on their affinities and neutralization activities. BLI binding kinetics measurements are shown for F9 (A) and C2 (B). Equilibrium dissociation constants (K D ) were determined on the basis of fits, applying 1:1 interaction model; ka, association rate constant; kd, dissociation rate constant. (C) Pseudo-typed SARS-CoV-2 neutralization assay was shown with selected αRep (C2, F9, C7, G1). An αRep specific to influenza polymerase (H7) was chosen as a negative control. To assess αReps specificity, pseudo-typed VSV-G were incubated with the highest concentration of each αRep (3 μM). Pseudo-type particles entry into cells was quantified by measuring luciferase activity (n = 3, mean ± SEM, two-way ANOVA, *P<0.05). (D) Cell viability of infected cells in presence of dilutions of αReps C2, C7, F9, G1 and H7 was monitored using the CellTiter-Glo Luminescent Assay Kit (Promega). Infected cells (triangle) and mock-infected cells (square) were included in the assay as controls (n = 2, mean is presented). (E) Half maximal inhibitory concentration (IC 50 ) were calculated using “log(inhibitor) vs. normalized response” equation from the neutralization potency curves with GraphPad Prism 8 software. ND: Not done, NA: Not available. https://doi.org/10.1371/journal.ppat.1010799.g002

Identification of neutralizing αReps We next tested the neutralization activity of the best αREPs candidates against SARS-CoV-2 pseudotyped murine leukemia virions (MLV) as previously described [18]; Fig 2C]. These virions only contain the SARS-CoV-2 spike protein on their surface and behave like their native coronavirus counterparts for entry in cells expressing ACE2. Upon cell entry, the luciferase reporter gets integrated into the host cell genome and is expressed, the measured signal being correlated with αRep neutralization properties. C2, F9 and C7 showed a dose-dependent neutralization activity, C2 displaying the highest neutralisation activity. Neither G1, an additional selected anti-RBD αRep, nor an anti-influenza αREP (H7), used as negative control, displayed notable neutralization activity. None of the αReps tested at the highest concentration (3 μM) displayed neutralization activity against vesicular stomatitis virus G pseudo-typed MLV, demonstrating their specificity. We confirmed this neutralization activity using SARS-CoV-2 infection of Vero E6 cells (Fig 2D). C2 showed the highest neutralizing potency with a half-maximal inhibitory concentration (IC 50 ) value of 0.1 μM, while C7 and F9 αReps displayed IC 50 values of 4.8 and 11.7 μM, respectively (Fig 2E). G1 as well as the anti-influenza H7 αRep did not show neutralization activity. We thus identified three potent neutralizing αReps, with C2 and F9 displaying affinity in the nM range. These two lasts αReps were retained for further analyses.

Design of αREP derivates In order to increase avidity and neutralization activity of these RBD binders, we aimed at generating multivalent αReps. We first determined if F9 and C2 recognized non-overlapping binding sites on the RBD to assess their interest to be linked in a fusion protein. Competitive binding assays carried out by BLI showed that C2 and F9 bindings on the RBD did not interfere in a reciprocal manner (Fig 3A and 3B). Competitive binding assays between these two αReps and soluble hACE2 showed that ACE2 binding occurred efficiently after binding of C2 on the RBD. In contrast, binding of F9 on the RBD partially inhibited recognition of hACE2. As a positive control, VHH72 [19] fully blocked hACE2 binding on the RBD (Fig 3C). These results suggest that the neutralization activity of the C2 αRep is not associated to a steric inhibition of the binding of the RBD on ACE2. To map the binding sites of F9 and C2 on the RBD, we carried out competitive binding assays using the neutralizing VHH-72 and VHH H11D4 previously described [19][20] as competitors. As shown by X-ray crystallography data, these two VHHs recognize distant epitopes on the RBD, the binding site of H11D4 partially overlapping the ACE2 binding motif (PDB numbers of the RBD-VHH complexes: 6WAQ and 6YZ5). As expected, VHH-72 and VHH H11D4 did not compete for their binding to the RBD (Fig 3D). αRep C2 and VHH H11D4 competed for the binding to the RBD in a reciprocal manner (Fig 3E) as well as αRep F9 with VHH72 (Fig 3F). These data suggest that αReps F9 and C2 recognize distant binding sites overlapping VHH-72 and VHH11D4 binding sites, respectively, and that a fusion between C2 and F9 αReps may be synergistic. We thus engineered bivalent αReps constructs using F9 and C2 αReps. We also generated trivalent αReps through the addition of a trimerization foldon domain (corresponding to the C-terminal part of T4 fibritin) behind C2 and F9 αReps [21]. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Competitive binding assays. (A and B) BLI experiments showed that C2 and F9 could bind RBD simultaneously. (C) Binding of ACE2 was assessed after a first association phase with αReps C2 and F9, the F9-C2 construct, the VHH72 [19] or with a negative control (NR). F9-C2 and VHH-72 blocked the binding of RBD to ACE2. While F9 inhibited partially ACE2 binding, C2 did not compete with ACE2 binding. (3D-F) Reciprocal competitive binding assays between VHH-72, VHH H11D4, C2 and F9. While C2 and H11D4 competed for binding to the RBD (immobilized on the chip) in a reciprocal manner, F9 and VHH-72 blocked reciprocally their interaction to the RBD. https://doi.org/10.1371/journal.ppat.1010799.g003

Properties of αREP heterodimers and homotrimers To build the F9-C2 and C2-F9 heterodimers, we inserted a 25 amino acid long flexible linker (GGGGS) 5 between these two subunits (S1 Fig). This linker length (that can reach 8 nm in length) allows the binding of these heterodimers between adjacent RBDs in the trimer, even in the “up” to “down” spike conformers. To generate the homotrimeric C2- and F9-foldon αReps, the foldon sequence was connected to the C-ter of the αREPs through a 16-amino acid long linker (GSAGSAGGSGGAGGSG) (S1 Fig). These linkers would allow cross-links between spikes at the surface of the virus particle. Unable to express efficiently the C2-F9 construct, only the F9-C2 affinity was characterized by BLI experiments (Fig 4A). F9-C2 displayed an equilibrium dissociation constant (K D ) of 91 pM, at least three folds better than that of monomers. F9-C2 also showed a substantially slowed dissociation rate constant of 5.86 x 10-5s-1 owing to enhanced avidity. Circular dichroism revealed melting temperatures of 86.5°, 88.3° and 86.0°C for C2, F9 and F9-C2, respectively, confirming the high stability of this class of protein (S2 Fig). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. The F9-C2 and C2-foldon constructs properties. (A) BLI binding kinetics measurements for F9-C2 to the S1-immobilized biosensor. (B) Pseudo-typed SARS-CoV-2 particles neutralization assay was performed with F9, C2, F9-C2 and C2-foldon constructs (n = 3, mean ± SEM, two-way ANOVA, *P<0.0001). (C) Cell viability of SARS-CoV-2-infected cells in presence of dilutions of F9-C2, C2-foldon, C2, F9 and H7 (an αRep negative control) was monitored using the CellTiter-Glo Luminescent Assay Kit (Promega) (n = 2, mean is presented). Infected cells (triangle) and mock-infected cells (square) were included in the assay. Half maximal inhibitory concentration (IC 50 ) were displayed. (D) SARS-CoV-2 neutralisation by aReps constructs. Virus replication was quantified by qRT-PCR in infected cells treated by C2, F9, F9-C2, C2-foldon (n = 3, mean ± SEM). Half maximal effective concentration (EC 50 ) were shown. https://doi.org/10.1371/journal.ppat.1010799.g004 We next investigated the ability of F9-C2 to block RBD-ACE2 interaction by BLI measurements (Fig 3C). When F9-C2 was bound to the RBD, addition of ACE2 induced no signal shift demonstrating that F9-C2 dimer is a potent inhibitor of spike binding to ACE2, similarly to the VHH72 [19]. We next explored the neutralization activity of F9-C2 and C2- and F9-foldon for comparison with their parental subunits against SARS-CoV-2 spike pseudo-typed MLV (Fig 4B). A synergic effect in neutralisation efficiency was evidenced when the F9 and C2 subunits were covalently linked and when C2 was assembled as a homotrimer. While C2 almost fully blocked entry of SARS-CoV-2 pseudo-type at a concentration of 250 nM, F9-C2 and C2-foldon neutralized infection at 50 nM. We next investigated their viral neutralization potencies in SARS-CoV-2 / Vero E6 cell infection assays by measuring cell viability (Fig 4C) and viral replication (Fig 4D). F9-C2 and C2-foldon were more effective than their monomeric counterparts to protect cells from SARS-CoV-2 infection, with an IC 50 of 12 nM and 3 nM, respectively, while C2 alone neutralized SARS-CoV-2 with an IC 50 of 77 nM. F9-foldon displayed a similar activity than its monomeric counterpart indicating no added value of this construction. Quantification of virus replication confirmed the same trend, with an EC 50 of 18 nM for F9-C2 and 8 nM for C2-foldon, indicating a higher neutralizing activity than C2 (EC 50 of 128 nM). Thus, the covalent linkage between the F9 and C2 subunits or the trimerization of C2 revealed a synergistic effect (~x 10–25) of αREPs oligomerization to neutralize SARS-CoV-2. Since F9-C2 targeted two different epitopes and may be less sensible to spike antigenic shift, we retained this heterodimer for further in vivo analyses.

F9-C2 prophylaxis limits SARS-CoV-2 infection in vivo In order to evaluate if F9-C2 nasal instillation prophylaxis was effective to limit SARS-CoV-2 infection in vivo, we first examined αRep stability in the nasal cavity of rodent (in mice). Using western blot analysis, we observed a slight decline in the amount of F9-C2 at 1h post-administration and did not evidence protein degradation (S3 Fig). We next used Syrian golden hamsters known to reflect the infection in human [22]. We focused on the nasal cavity as we choose to examine how a local treatment could limit the start of the infection in a physiological context. We pre-treated the hamsters with 0.6 mg of F9-C2 distributed between the two nostrils 1h prior to infection with 5.103 TCID 50 of SARS-CoV-2 of the circulating European strain in 2020 (harbouring the D614G mutation in the spike protein) (Fig 5A). After such treatment, we observed the presence of infiltrated αReps on the surface of the epithelium layer, indicating an efficient absorption of the molecule (S4 Fig). The group treated with the non-neutralizing αREP G1 lose weight starting from day 2. Treatment with F9-C2 limited weight loss and the difference with G1 treatment almost reach significance at 3 dpi (P = 0.057, 2-way ANOVA, Fig 5B). During the 3 days following infection, virus titres in nasal swabs were lower in the group treated with F9-C2 than in the G1-treated group (Fig 5C, two-way ANOVA, P<0.0001). In the olfactory turbinates where virus starts to replicate at high titres, amount of viral RNA was significantly lower at 1 dpi in F9-C2 treated animals (Mann Whitney, P = 0.0286, Fig 5D), consistent with a tendency of lower expression of inflammation markers, in particular IL-6 and TNFα. At 3 dpi, no significant differences were observed for viral RNA and inflammation markers between the two treatments. Thus, while we observed a significative difference in infectious particles present in nasal swabs between the two conditions (treated with G1 or F9-C2, at day 3 post-infection), we did not reveal a difference in the amount of viral RNA in the nasal epithelium, suggesting an erosion of the F9-C2 inhibitory activity at late times post infection. Next, we examined if the lower amount of viral RNA at 1 dpi in the nasal cavity of F9-C2-treated animals was reflected at the histological level. While the virus was present in large patches of the epithelium in the G1-treated animals, it was only present in small stretches in F9-C2-treated animals (Fig 5E). We measured the infected area in the rostral part of the nasal cavity at 1 dpi where the infection starts. The difference between G1 and F9-C2 treated animals was close to significance (Mann Whitney, P = 0.057, Fig 5F). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Efficacy of F9-C2 αRep prophylaxis in SARS-CoV-2 infection in a golden Syrian hamster model. (A) Overview of the experiment design. 6 mg/kg of αReps were delivered intranasally in hamsters 1h prior to infection with 5.103 TCID 50 of SARS-CoV-2. (B) Evolution of animal weight (n = 4, mean of the relative weight to 1-day prior infection ± SEM, two-way ANOVA). (C) Evolution of virus titre in nasal swabs (n = 4, mean of TCID 50 ± SEM, two-way ANOVA, ****P<0.0001) (D) Quantification of RNA encoding SARS-CoV-2 protein E, IL-6, TNFα, Ncf2 in the olfactory turbinates, relative to viral infection, inflammation and neutrophil respectively (normalized to ß-actin, mean ± SEM, Mann–Whitney *P<0.05). (E) Representative images of the infected olfactory epithelium area treated by G1 or F9-C2 in the rostral zone of the nasal cavity (1 dpi) showing respectively a strong and partial infection. (F) Measurement of the extent area of infection in the dorso-medial part of the hamster nose. Values represent the mean of infected area (Arbitrary Unit ± SEM, Mann–Whitney *P<0.05). https://doi.org/10.1371/journal.ppat.1010799.g005

Repeated F9-C2 treatments further limit SARS-CoV-2 infection in vivo In order to improve the efficiency of the treatment, hamsters were treated with F9-C2 (0.6 mg per dose) 1h prior infection and on days 1 and 2 post-infection (Fig 6A). F9-C2 treatments limited weight loss and the difference reach significance at 3 dpi when compared to controls (P = 0.015, 2-way ANOVA, Fig 6B). During the 3 days post-infection, virus titres in nasal swabs were lower in the group treated with F9-C2 than in the control group (Fig 6C, two-way ANOVA, P<0.0001). Viral RNA was significantly lower in olfactory turbinates at 1 dpi and 3 dpi in F9-C2 treated animals (Mann Whitney, P = 0.0286, Fig 6D) when compared to an irrelevant αRep. This observation correlates with lower expression of inflammation markers (IL-6, TNFα and Ncf2, the last one being related to neutrophil infiltration). We observed less damage of the olfactory epithelium accompanied with a reduction of immune cell infiltration (revealed by the iba1+ marker) and desquamated cells in the lumen of the nasal cavity for animals treated by F9-C2, especially at 1 dpi (Figs 6E and S5). The infected area in the rostral part of the nasal cavity at 1 dpi was significantly reduced in F9-C2 treated animals compared to controls (Mann Whitney, P = 0.0286, Fig 6F). These results suggested that repeated injections of F9-C2 significantly reduce the spread of the virus up to 3 days post-infection. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Efficacy of F9-C2 αRep repeated treatments in SARS-CoV-2 infection in a golden Syrian hamster model. (A) Overview of the experiment design. 6 mg/kg of αReps were delivered intranasally in hamsters 1h prior to infection with 5.103 TCID 50 of SARS-CoV-2. The treatment was repeated on 1 dpi and 2 dpi for the group examined at 3 dpi. (B) Evolution of animal weight (n = 4, mean of the relative weight to 1-day prior infection ± SEM, two-way ANOVA). (C) Evolution of virus titre in nasal swabs (n = 4, mean of TCID 50 ± SEM, two-way ANOVA, ****P<0.0001) (D) quantification of RNA encoding SARS-CoV-2 protein E, IL-6, TNFα, Ncf2 in the olfactory turbinates, relative to viral infection, inflammation and neutrophil respectively (normalized to ß-actin, mean ± SEM, Mann–Whitney *P<0.05). (E) Representative images of the infected olfactory epithelium area treated by F9-C2 or H7 in the dorso-medial zone of the nasal cavity (1 dpi) showing respectively a partial infection with a low number of Iba1+ immune cell infiltration and a strong infection associated with damage of the olfactory epithelium and Iba1+ cell infiltration as well as desquamated cells in the lumen of the nasal cavity (white asterisk). (F) Measurement of the extent area of infection in the dorso-medial part of the hamster nose. Values represent the mean of infected area (Arbitrary Unit ± SEM, Mann–Whitney *P<0.05). https://doi.org/10.1371/journal.ppat.1010799.g006

[END]
---
[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1010799

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/