XAV-19, a swine glyco-humanized polyclonal antibody against SARS-CoV-2 Spike receptor-binding domain, targets multiple epitopes and broadly neutralizes variants

Amino acid substitutions and deletions in Spike protein of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants can reduce the effectiveness of monoclonal antibodies (mAbs). In contrast, heterologous polyclonal antibodies raised against S protein, through the recognition of multiple target epitopes, have the potential to maintain neutralization capacities. XAV-19 is a swine glyco-humanized polyclonal neutralizing antibody raised against the receptor binding domain (RBD) of the Wuhan-Hu-1 Spike protein of SARS-CoV-2. XAV-19 target epitopes were found distributed all over the RBD and particularly cover the receptor binding motives (RBM), in direct contact sites with the Angiotensin Converting Enzyme-2 (ACE-2). Therefore, in Spike/ACE2 interaction assays, XAV-19 showed potent neutralization capacities of the original Wuhan Spike and of the United Kingdom (Alpha/B.1.1.7) and South African (Beta/B.1.351) variants. These results were confirmed by cytopathogenic assays using Vero E6 and live virus variants including the Brazil (Gamma/P.1) and the Indian (Delta/B.1.617.2) variants. In a selective pressure study with the Beta strain on Vero E6 cells conducted over 1 month, no mutation was associated with addition of increasing doses XAV-19. The potential to reduce viral load in lungs was confirmed in a human ACE2 transduced mouse model. XAV-19 is currently evaluated in patients hospitalized for COVID-19-induced moderate pneumonia in a phase 2a-2b (NCT04453384) where safety was already demonstrated and in an ongoing 2/3 trial (NCT04928430) to evaluate the efficacy and safety of XAV-19 in patients with moderate-to-severe COVID-19. Owing to its polyclonal nature and its glyco-humanization, XAV-19 may provide a novel safe and effective therapeutic tool to mitigate the severity of coronavirus disease 2019 (Covid-19) including the different variants of concern identified so far.


Introduction
Passive antibody therapies have demonstrated efficacy to reduce progression of mild coronavirus disease 2019  to severe disease if administered early enough in the course of illness [1][2][3] . Three sources of antibodies have so far been assessed. First, passive antibody therapy using the infusion of convalescent plasma (CP) with high SARS-CoV-2 antibody titers in hospitalized patients, administered within 72 hours after the onset of mild symptoms, reduced the relative risk of progression to severe disease by 73% if CP presented a titer of >1:3200 and by 31.4% with lower titer CP 1 . This was true with CP drawn between June and October 2020. However, the CP from patients infected by the original SARS-CoV-2 lineage had poor activity against the Beta variant and this was attributed to three mutations (K417N, E484K, N501Y) in the Spike protein 4 . Among these mutations, E484K has been shown to play a major role to reduce the binding and neutralization 5  However, viral mutations can escape the mAbs which are used to treat the infection of SARS-CoV-2 8 . The Alpha variant is refractory to neutralization by most mAbs which target the NTD of Spike protein and also resistant to several RBD-specific mAbs 9 . Mutations in the Beta lineage (K417N, E484K, and N501Y in RBD), especially mutations of Spike at E484 but also in the N-terminal Domain (NTD; L18F, D80A, D215G, Δ242-244, and R246I in SA variant 10,11 ) reduce neutralization sensitivity or confer neutralization escape from multiple mAbs 4,5,[12][13][14][15][16][17][18][19][20] . Third, polyclonal antibodies produced in their Fab'2 format from horses 21 or in their IgG format from humanized cows 22 or glyco-humanized pigs 23 have also proven efficacy to neutralize SARS-CoV-2. The safety and tolerability in humans of Fab'2 from horses and of humanized IgG polyclonal antibodies has been confirmed recently in different clinical trials (Lopardo et al, 2021 21 , NCT04453384, NCT04469179, Gaborit et al, 2021 24 ), contrasting with unmodified polyclonal antibodies containing wild-type IgG antibodies that induce serum sickness and allergic reactions (including fever and skin rashes) in 20 to 30% of the patients, excepting for those who concomitantly receive immunosuppression and high doses steroids 25,26 . A partial efficacy of anti-SARS-CoV-2 Fab'2 from horses has been reported in those patients with negative baseline antibodies (NCT04494984). The efficacy of humanized or glyco-humanized IgG polyclonal antibodies in COVID-19 is still being investigated (NCT04453384, NCT04928430).
The possible advantage of polyclonal antibodies over mAbs is their recognition of an array of epitopes on the target antigen which should theoretically be not or less affected by antigen variations. Here, we investigated the extent to which XAV-19, a glyco-humanized swine polyclonal antibody previously shown to present neutralizing activity against SARS-CoV-2 Wuhan and D614G (B.1, PANGOLIN lineage) viruses 23
Intermediate R&D preparations of swine glyco-humanized polyclonal antibody against SARS-CoV-2 had been generated, presenting variable anti-SARS-CoV-2 binding activities 23  Wuhan was isolated at The Chinese University of Hong Kong (China).

Binding ELISA
The target antigen (SARS-CoV-2 Spike RBD-HIS protein, Sino Biological Europe) was immobilized on Maxisorp plates at 1µg/ml in carbonate/bicarbonate buffer at 4°C overnight.
After washing, saturation was performed with PBS-Tween-BSA for 2h at room temperature.
Samples were diluted into PBS-Tween and added into the plate in duplicate, incubated 2h at RT and washed 3 times. Bound pig IgGs were revealed with a secondary anti-pig-HRPconjugated antibody (Bethyl Laboratories, USA) diluted in washing buffer, at 1:1000, incubated 1h at RT and washed 3 times. TMB reagent was added in the plate, incubated up to 20 minutes in the dark and the reaction was stopped with H2SO4. Reading was performed at 450nm.

Spike/ACE-2 neutralization assay
An assay was developed to assess the properties of anti-SARS-CoV-2 Spike antibodies to inhibit binding of ACE-2 to immobilized Spike. SARS-CoV-2 Spike S1 (either Wuhan, Alpha or Beta) was immobilized on Maxisorp plates at 1 µg/ml in carbonate/bicarbonate buffer pH 9.0 at 4°C overnight. The plates were washed in PBS-Tween-0.05% and saturated with PBS-Tween-0.05%-2% skimmed milk for 2h at room temperature (RT). Anti-Spike RBD antibodies diluted in PBS-Tween-0.05%-1% skimmed milk were then added and incubated for 30 min.
Then, ligand human ACE-2-mFc tag (Sino Biological; 125 ng/ml final concentration) was added in the same dilution buffer. After 1h incubation at room temperature and 3 washes, the mouse Fc tag was revealed with a specific HRP-conjugated anti-mouse IgG secondary antibody (diluted in in PBS-Tween-0.05%-1% skimmed milk powder at 1:1000, incubated 1h at RT and washed 3 times). TMB reagent was added into the plate, incubated 6 minutes in the dark and the reaction was stopped with 50 µl 1M H2SO4. The plate was read at 450 nm.

Determination of XAV-19 target epitopes
A peptide microarray analysis has been performed using the PepStar TM system (JPT Peptide Technologies, Berlin, Germany). Fifty-three purified synthetic 15-meric overlapping peptides derived from the SARS-CoV-2 Spike S1 RBD domain (sequence from BDSOURCE accession number NC_045512.2), with an additional C-terminal glycine (added for technical reasons) were covalently immobilized on glass surface. Full-length human IgG, mouse IgG and preimmune pig IgG were co-immobilized on microarray slides as assay controls. XAV-19 sample used in the analysis is the clinical drug substance batch BMG170-B06, hybridized at the dilution of 100 µg/ml (the same applied for control swine IgG) for 1 hour at 30°C on microarray slides.
After sample incubation, secondary fluorescently labeled mouse anti-pig-IgG antibody diluted 1:5000 was added in the corresponding wells and left to react for 1 hour. Finally, a tertiary fluorescently labeled anti-mouse-IgG antibody at 1 μg/ml was incubated for 1 hour to detect bound anti-pig-IgG secondary antibody. After washing and drying, the slides were scanned with a high-resolution laser scanner at 635 nm to obtain fluorescence intensity profiles. The slides were scanned with a receiver gain of 900 V and images quantified to yield a mean pixel value for each peptide.
The epitope mapping process comprised a proteolytic digestion step of the SARS-CoV-2 Spike RBD protein, the isolation of resulting peptides by XAV-19 affinity chromatography and the LC-MS/MS analysis of the eluted peptides. In short, the RBD protein was reduced, alkylated and digested with an enzyme/protein ration of 1:50 during 3h at 37°C with the endopeptidases trypsin, chymotrypsin or Arg-C. Digestion products were immunocaptured on Sepharose 4B on which XAV-19 IgG has been immobilized, during 2h at room temperature. Binding yields on Sepharose was 80%. Columns were then washed (ammonium bicarbonate 25mM) and elution performed with a Glycine/HCl 50 mM pH2 buffer. Eluted fractions were then resolved by C18 inverted phase chromatography and tandem analyzed (MS/MS) to measure peptide masses.
Data were compared with theoretical masses resulting from an in silico RBD digestion with the corresponding enzyme. Titration of viral stock was performed on Vero E6 by the limiting dilution assay allowing calculation of tissue culture infective dose 50% (TCID50). The neutralizing activity of XAV- 19 was assessed with a whole virus replication assay using the five SARS-CoV-2 isolates.

Cytopathogenic Effect (CPE) assay
XAV-19 was subjected to serial two-fold dilution ranging from 50 µg/ml to 0.05 µg/ml in fresh medium. 50 µl of these dilutions were incubated with 50 µl of diluted virus (2 x 10 3 TCID50/ml) per well in a 96-well plate at 37°C for 60 min in 8 replicates. Hundred µl of a Vero E6 cell suspension (3 x 10 5 cells/ml) were then added to the mixture and incubated at 37°C under an atmosphere containing 5% CO2 until microscopy examination on day 4 to assess CPE. An infectivity score has been assigned on each well: 0, no cytopathic effect; 1, a fraction of cells Viable cells were quantified with CellTiter-Glo 2.0 luminescent cell viability assay.

Antibodies escape study
XAV-19 was subjected to serial two-fold dilutions ranging from 50 µg/ml to 0.2 µg/ml in fresh medium. Fifty µl of XAV-19 dilutions were incubated with 50 µl of diluted virus (2 x 10 3 TCID50/ml) per well in a 96-well plate at 37°C for 60 min in duplicates. Hundred µl of a Vero E6 cell suspension (3 x 10 5 cells/ml) was then added to the mixture and incubated at 37°C under an atmosphere containing 5% CO2. A no antibody control was included to account for any cell culture adaptations of each SARS-CoV-2 variants. Virus replication was monitored on day 4 by screening for cytopathic effect. The supernatants were collected from wells with the highest antibody concentration displaying evident CPE. Fifty µl of these supernatants were passed under the same or greater fresh XAV-19 concentrations as before, until five passages. RNA extraction was also performed on these supernatants with NucliSENS EasyMag (BioMerieux) according to manufacturer's protocol. Sanger sequencing was performed on the last passage (5 th ) of each variant.

Human ACE-2 mouse model
The protocol of the animal experiments was described in the previous study 27 . Balb/c mice were first infected with 10 8 TCID50 of adenovirus carrying human ACE-2 protein intranasally.

XAV-19 binding to SARS-CoV-2 Spike correlates with neutralizing potency
To assess whether binding strength to SARS-CoV-2 Spike of IgG antibodies present in XAV-  Figure 1B).

XAV-19 target epitopes on SARS-CoV-2 Spike protein
Two orthogonal methods have been used to identify the epitopes recognized by XAV-19 on the S1-RBD protein. First, all 15-meric peptides overlapping by 10 amino acids (thus a total of 53 peptides) were spotted on glass slides and hybridized with XAV-19 or pre-immune swine GH-pAb control antibodies. The resulting heatmap plot revealed that all peptides, though to varying extents, could be specifically recognized by antibodies contained in XAV-19 ( Figure 2A). Since not all peptides can be linked together with XAV-19 antibodies when contained in the Spike protein, for steric hindrance, a second investigation was undertaken to identify which peptides in the Spike RBD domain in a more native configuration are recognized by XAV-19 antibodies.
The assay was based on the recognition by XAV-19 antibodies of SARS-CoV-2 Spike RBDderived peptides obtained by proteolytic digestion (three enzymes tested: trypsin, chymotrypsin and Arginase-C) and an isolating step of the resulting peptides by affinity chromatography (XAV-19 being immobilized on Sepharose) followed by an LC-MS/MS analysis of the eluted peptides. This epitope mapping analysis revealed several recognition areas on the S1-RBD protein. The peptides where the two methods gave the strongest overlapping hits, thus most probably representing dominant target epitopes, were amino acids 347-355 and 445-461.
Amino acids 409-417, 462-473 and 530-535 were also found to be protected in the LC-MS/MS analysis although less recognized in the peptide array ( Figure 2B). Interestingly, 6 amino acids described to directly interact with human ACE-2 28 are located in these regions.

ACE-2
XAV-19 was tested in a Spike/ACE-2 binding competition assay, where the Spike protein was of the original Wuhan type or contained the RBD mutations N501Y, N439K, Y453F described in the Alpha and Beta variants, or the mutation E484K show to induce resistance to mAbs 29 .
Variants expressing a combination of mutations present in the Spike Alpha (HV69-70 deletion, Y144 deletion, N501Y, A570D, D614G, P681H) or Beta (K417N, E484K, N501Y, D614G) were also tested. All single mutation forms of the Spike could be fully neutralized at concentrations not significantly different (slightly lower for the E484K mutation) from the Wuhan type ( Figure 3A). XAV-19 also demonstrated a 100% inhibitory activity on the 2 Spike proteins fully representative of the Alpha and Beta variants, similar to the Wuhan Spike, with IC50 values of 6.4, 4.0 and 4.5 µg/ml, respectively ( Figure 3B). Bamlanivimab, tested in parallel, demonstrated a potent inhibitory capacity against the Wuhan and the Alpha variants, with a IC50 value of 0.01µg/ml but, as described 30 , failed to inhibit binding of Beta Spike to ACE-2, even at high concentration ( Figure 3C).

Neutralization of live SARS-CoV-2 variants in CPE
The neutralizing effect of XAV-19 was determined by CPE assays in 3 different platforms

Absence of XAV-19 induced SARS-CoV-2 variant selection in vitro
Antibodies, by applying a selective selection pressure, can favor outgrowth of resistant novel variants with unknown properties, especially in conditions where their neutralizing potency is suboptimal. To investigate whether XAV-19 is susceptible to generate such variants, Vero E6 cells were infected with 100 TCID50/50 µl of either B.1D614G, Alpha, Beta, Gamma or Delta strains and maintained over 5 passages (20 days) with culture medium or culture medium containing increasing concentrations of XAV-19. After passage 5, emergence of antibodies escape mutants was evaluated by Sanger sequencing. The data ( Table 1)  were found with XAV-19. These mutations probably represent adaptations to the culture conditions. None of these mutations has been described as resistant to antibodies.

Reduction of viral load by XAV-19 in human ACE-2 expressing mice
Balb/c mice with human ACE-2 expression in the lung were infected with 10 5 PFU SARS- When administered intraperitoneally to human-ACE-2 mice challenged intranasally with SARS-CoV-2 viruses, XAV-19 induced a dose-dependent reduction of viral load in the lung, demonstrating the potential to localize to infected tissues. The 94% to 98% reduction in the viral load noticed here is in agreement with data obtained with different neutralizing antibodies in ferrets, hamster or mouse models 32,33,[34][35][36][37][38][39][40] or obtained with the REGN-CoV2 antibody cocktail in a comparable animal model 41 . In a few studies, however, the viral reduction factor Early after Covid-19 outbreak onset, many labs have been able to rapidly develop neutralizing antibodies. One year later, as of mid-2021, more than 93 clinical trials assessing the safety and benefit of mAbs and 8 of polyclonal antibodies are listed in the Clinicaltrials.gov repository (NCT04610502,  NCT04838821,  NCT04514302,  NCT04834908,  NCT04834089, NCT04518410, NCT04453384, NCT04453384). In late 2020, variants of concern started to spread in the population and now cause the majority of infections. However, antibodies that are now assessed clinically have been mostly raised against the initial, Wuhan strain. It has therefore become essential to revisit their potential to also neutralize variants.      28 .