Development of a cost-effective ovine antibody-based therapy against SARS-CoV-2 infection and contribution of antibodies specific to the spike subunit proteins

Antibodies against SARS-CoV-2 are important to generate protective immunity, with convalescent plasma one of the first therapies approved. An alternative source of polyclonal antibodies suitable for upscaling would be more amendable to regulatory approval and widespread use. In this study, sheep were immunised with SARS-CoV-2 whole spike protein or one of the subunit proteins: S1 and S2. Once substantial antibody titres were generated, plasma was collected and samples pooled for each antigen. Non-specific antibodies were removed via affinity-purification to yield candidate products for testing in a hamster model of SARS-CoV-2 infection. Affinity-purified polyclonal antibodies to whole spike, S1 and S2 proteins were evaluated for in vitro for neutralising activity against SARS-CoV-2 Wuhan-like virus (Australia/VIC01/2020) and a recent variant of concern, B.1.1.529 BA.1 (Omicron), antibody-binding, complement fixation and phagocytosis assays were also performed. All antibody preparations demonstrated an effect against SARS-CoV-2 disease in the hamster model of challenge, with those raised against the S2 subunit providing the most promise. A rapid, cost-effective therapy for COVID-19 was developed which provides a source of highly active immunoglobulin specific to SARS-CoV-2 with multi-functional activity.


Introduction
The outbreak of coronavirus disease first identified in 2019 , caused by infection with the etiological agent severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), was declared a pandemic on March 11, 2020 (Cucinotta and Vanelli 2020). It continues to blight human public health, and the race to battle it with countermeasures continues apace. Whilst vaccines have made a valiant effort in the control of COVID-19, none of the current vaccines offer sterilising protection and there are major obstacles to overcome for global control of the virus (Kim et al., 2021). Therefore, new antiviral strategies will continue to play an important role in mitigating disease.
One of the first treatments explored early in the pandemic was antibody therapy sourced from previously infected patients. Convalescent human plasma was authorised for emergency use by the U.S. Food and Drug Administration (FDA) in August 2020 for the treatment of hospitalised patients with COVID-19. However, there are at least 176 registered clinical trials assessing convalescent plasma with seemingly contradictory results being reported (Piechotta et al., 2020;Ning et al., 2021). Major problems with convalescent plasma therapy include quality control and standardisation, defining optimal dosing and timepoints and the risk of infection with unknown blood-borne infectious agents (Ning et al., 2021). These issues can be overcome by using intravenous immunoglobulins (IVIG) which are sterile and purified from plasma collected from large pools of donors (Cao et al., 2020). However, the same issue applies with the reliance on human material as the source material.
As an alternative, animal-derived antibodies are more applicable for large scale production and standardisation. Ovine immunoglobulin G (IgG)-based products have been widely used as snake antivenoms (Gutierrez et al., 2011), and have been applied as treatments to infectious diseases including rabies (Redwan el, Fahmy et al. 2009) and tetanus (Redwan el, Khalil et al. 2005). The targeting of polyclonal antibodies to multiple epitopes enables a multitude of effector functions, such as steric hindrance (preventing the virus from attaching to the host cell surface), aggregation (bunching of viruses leading to clearance from the circulation), opsonisation (activation of phagocytic cells) and activation of the complement system (Haurum 2006).
We have previously applied the ovine polyclonal antibody-based approach to develop a cost-effective candidate against Ebola virus (EBOTAb) (Dowall et al., 2016), which demonstrated protection when given several days after challenge in the guinea pig (Dowall et al., 2016) and non-human primate (Dowall et al., 2017) disease models. Based on this success, it was appropriate to utilise this strategy to develop a similar therapy for COVID-19.
For the immunogens to develop the COVID-19 ovine antibody preparations, the full-length spike protein of SARS-CoV-2 was used alongside each of the two individual subunits. The spike protein controls SARS-CoV-2 infectivity with 30-40 spike homotrimers being present on the surface of each virion (Yao et al., 2020). Each spike protein has two subunits, S1 and S2. The S1 subunit contains the receptor-binding domain (RBD) which recognises the host cell receptor, angiotensin-converting enzyme 2 (ACE2) (Lan et al., 2020;Wang et al., 2020), whereas S2 mediates the sequential membrane fusion events allowing the virus to enter the host cytoplasm (Belouzard et al., 2012;Shang et al., 2020).

Recombinant proteins
Full-length SARS-CoV-2 spike glycoprotein was produced in Chinese Hamster Ovary (CHO) cells with a His-tag incorporated (REC31868; Native Antigen Company, UK). Subunit SARS-CoV-2 S1 and S2 proteins were produced in human embryonic kidney cells (HEK293) with a sheep Fc-tag (REC31806 and REC31807, respectively; Native Antigen Company, UK).

Sheep
Border Leicester Cross Merino ewes of at least 12 months of age, born in Australia, were obtained from an approved supplier and kept at a controlled farm registered with the Australian Department of Agriculture. The farm operates under extremely high health status and standards of welfare, maintained by regular health inspections and veterinary checks which are carried out weekly and within 3 days of blood withdrawal.

Plasma and IgG production
Plasma and IgG production were conducted by International Therapeutic Proteins Ltd. Sheep were immunised on a 28 day schedule with 0.5 mg recombinant protein delivered subcutaneously across six sites: axillae (x2), groin (x2) and supra-scapula (x2). Six sheep were immunised with full-length spike glycoprotein and three for each of the S1 and S2 subunit proteins. Freunds Complete adjuvant was used for the first immunisation and Freunds Incomplete for subsequent ones. After a priming period that included two immunisations, plasma was collected via plasmapheresis using an automated MCS machine (Haemonetics Corporation, USA). Each collection typically yielded 600 ml plasma which was stored at − 25 • C until further processing. Plasma was collected on a 28 day schedule, 2 weeks following immunisation alongside a 5 ml sera sample for assessing antibody levels at the time of collection. The IgG fraction was purified from the hyperimmune plasma by a series of polyethylene glycol (PEG) fractionations and a Zinc precipitation. The product was formulated at a concentration of 50-60 g/L in 20 mM Sodium Acetate/20 mM NaCl buffer.

Affinity-purification for candidate therapy manufacture
Affinity-purification of antibodies were undertaken by the Native Antigen Company. Columns were made using a ratio of ~5 mg recombinant protein per ~1g of de-hydrated, cyanogen bromide-activated Sepharose (Cytiva) according to manufacturer's instructions, resulting in 5 mL final resin per antigen. Unbound antigen was washed out with 3 × 2 column volumes (CV) of 0.1M NaOAc, pH4.0, 500 mM NaCl, followed by 3 × 2 CV of 0.1M Tris-HCl, pH8.0, 500 mM NaCl. This wash cycle was repeated twice. For the purification of antibodies, the resins were transferred to 10 mL polypropylene gravity flow columns (ThermoFisher).
The PEG precipitated IgG-fraction was pH-adjusted by the addition of 1M HEPES, pH8.0 (1.25 mL per 10 mL of original IgG fraction). After centrifugation (10 min, 4000×g, 20 • C), the clarified supernatant was loaded on the respective antigen column equilibrated in 10 mM HEPES pH 8.0 with a contact time of approx. 1.5 h. The columns were washed with 10 CV of equilibration buffer, followed by 10 CV 10 mM HEPES pH 8.0, 300 mM KCl. Antibodies were eluted with 100 mM glycine, pH 2.5. Protein containing fractions were pooled and adjusted to pH7-8 with 1M Tris-HCl, pH9.0 (1/5th of the original volume) before dialysis into DPBS.

Enzyme-linked immunosorbent assay
Nunc MaxiSorp microtitre plates were coated with 2 μg/ml recombinant protein in bicarbonate buffer overnight at 2-8 • C. Plates were washed 3 times with phosphate buffered saline (PBS) containing 0.05% Tween20 (PBST) and blocked for 1 h at 37 • C with blocking buffer (5% skimmed milk powder in PBS). Plates were incubated for 1 h at 37 • C with prediluted samples (sera collected from sheep bleeds or purified antibody preparations); washed with PBST; and incubated with a donkey anti-ovine IgG horseradish peroxidase (HRP) conjugate (Product 713-035-003; Jackson ImmunoResearch, USA) for 1 h at 37 • C. After further washing, TMB substrate was added and the reaction was stopped by the addition of stop solution before reading the optical density at a wavelength of 450 nm.

Receptor binding domain (RBD) binding assay
Anti-RBD antibodies were determined using the fully quantitative Roche Elecsys anti-SARS-CoV-2 S assay (ACOV2 S), a species independent electrochemiluminescent immunoassay (ECLIA). Samples are considered negative if < 0.8 U/ml, or positive if > 0.8 U/ml, with a detection range of 0.400-225,000 U/ml.
A micro-neutralisation assay based on a previously published protocol (Bewley et al., 2021), with minor modifications, was used to assess the neutralising activity of the antibodies. Antibodies were diluted 2-fold over a 12-step dilution range (50 μg/ml -0.02 μg/ml), in duplicate (technical replicates). A fixed concentration of SARS-CoV-2 was added to the diluted antibody. Additional assay wells included virus-free and untreated virus-only controls. The neutralising plates (diluted antibody & virus) were then incubated for 1 h at 37 • C, afterwards the contents were then transferred onto 96-well plates containing Vero-E6 cells (ECACC 85020206) and the virus was allowed to adsorb to the cells for 1 h at 37 • C. The inocula were removed from the wells before the addition of 100 μl overlay media (MEM containing 4% FCS and 1% carboxymethylcellulose (CMC) solution) to each well. Plates were incubated for 24h (Victoria isolate) or 26h (Omicron BA.1) at 37 • C to allow foci to form.
After incubation cells were fixed overnight with 8% (w/v) formalin/ PBS, washed with water and permeabilised with 0.2% (w/v) Triton X-100/PBS at room temperature for 10 min. Cells were washed with PBS, incubated with 0.3% hydrogen peroxide (Sigma) at room temperature for 20 min and washed with PBS. Foci were stained with 50μl/well rabbit anti-nucleocapsid (Sino Biological, 40588-T62) diluted 1:1000 in 0.2% (w/v) Triton X-100/PBS for 1 h at room temperature. Antibody dilutions were discarded, cells washed with PBS and incubated with 50μl/well goat anti-rabbit IgG HRP (Invitrogen, G-21234) diluted 1:4000 in 0.2% (w/v) Triton X-100/PBS for 1h at room temperature. Cells were washed with PBS and incubated with TrueBlue peroxidase substrate (SeraCare, 5510-0030) for 10 min at room temperature then washed with water. Infectious foci were counted with an ImmunoSpot® S6 Ultra-V 367 analyser with BioSpot counting module (Cellular Technologies Europe, Germany). The counted foci data were then imported into R-Bioconductor. Two or three independent experiments were performed for each antibody. The internal positive control for the Victoria isolate was convalescent plasma, donated to UKHSA from the Northern Ireland Blood Transfusion Service (NIBTS), from a patient recovering from a Wuhan-like virus. The internal positive control for the Omicron BA.1 variant was convalescent plasma, donated to UKHSA, from a patient recovering from Omicron infection.
A midpoint probit analysis (written in R programming language for statistical computing and graphics) was used to determine the amount (μg/mL) of antibody required to reduce SARS-CoV-2 viral foci by 50% (IC 50 ) compared with the virus only control (n = 10).
Affinity-purified ovine polyclonal antibodies (3 μl in duplicate) were added to 27 μl BB and serially diluted 1:3 in BB. This was followed by 20 μl of SARS-CoV-2 spike protein-coated magnetic beads (50 beads per μl) to give a final 1:3 serial dilution range starting at 1:20. The serial dilution for NIBSC 20/162 standard started at 1:80. The mixture was incubated at 25 • C for 30 min with shaking at 900 r.p.m. The beads were washed twice in 200 μl wash buffer (BB+0.05% Tween-20), then resuspended in 50 μl BB containing 10% IgG-and IgM-depleted human plasma (prepared per (Alexander et al., 2022)) and incubated at 37 • C for 15min with shaking at 900 r.p.m. Beads were next washed twice with 200 μl wash buffer and resuspended in 100 μl fluorescein (FITC)-conjugated rabbit anti-human C3c polyclonal antibody (Abcam) diluted 1:500 in BB and incubated in the dark. After two more washes with 200 μl wash buffer, the samples were resuspended in 40 μl HBSS and analysed using an iQue Screener Plus® with iQue Forecyt® software (Sartorius, Germany). For each sample, a minimum of 100 beads were collected. Conjugated beads were gated based on forward scatter and side scatter and then further gated by allophycocyanin (APC) fluorescence. The APC fluorescent-bead population was gated and measured for FITC Median Fluorescent Intensity, which represents deposition of C3b/iC3b (gating strategy shown in Supplementary Fig. 1). The NIBSC 20/162 calibrant was plotted as a 4 parameter logistic (PL) curve with 1/Y2 weighting and the linear range calculated. The mean fluorescence intensity (MFI) from each sample was interpolated against the NIBSC 20/162 4 PL curve and the calculated concentration that hit the linear range was multiplied by the dilution factor to assign activity of the sera as Complement Activating Units (CAU).

MSD assay
Samples were determined for their capability at neutralising different variants of concern using the MesoScale Discovery ACE2 assays. Samples were run using the MSD Plate 13 (K15466U) which harbours full-length spike antigens specific to Wuhan and 9 different VOC: Alpha (B.

Hamsters
Golden Syrian hamsters, aged 7-9 weeks (weight range 122-162g), were obtained from a UK Home Office accredited facility (Envigo RMS UK Ltd). Animals were housed in cages in accordance with the requirements of the UK Home Office Code of Practice for the Housing and Care of Animal Used on Scientific Procedures (1986). During procedures with SARS-CoV-2 animals were housed in a flexible-film isolator within a Containment Level 3 facility. Animals were randomly put into groups, with equal allocation of male and female animals. Group sizes of 6 hamsters were used as the minimal number required for statistical significance to be achieved. Access to food and water was ad libitum and environment enrichment was provided. All experimental work was conducted under the authority of a UK Home Office approved project licence that had been subject to local ethical review at Public Health England (now part of the UK Health Security Agency (UKHSA)) Porton Down by the Animal Welfare and Ethical Review Body (AWERB) as required by the Home Office Animals (Scientific Procedures) Act 1986. Humane clinical endpoints consisted of 20% weight loss or severe signs of disease/distress, with animals reaching these limits being culled.

Antibody administration
Affinity-purified antibodies were administered to hamsters after diluting to a concentration of 1 mg/mL with PBS and injecting 2 mL via the intraperitoneal route. On the day before challenge, four groups of hamsters received therapy: antibodies to whole spike, S1 and S2 proteins or PBS control. A separate group received antibodies to whole spike protein on day 3 post-challenge.

Clinical observations
Animals were monitored for abnormal clinical signs twice daily. These were assigned a score based upon the following criteria: 0, normal; 1, behavioural changes; 2, ruffled fur, dehydrated, wet tail; 3, arched, wasp-waisted, eyes shut; and 5, laboured breathing. At the same time each day, animals were weighed.

Sampling
Pharyngeal swabs were taken on day 2, 4, 6, 8 and 9 post-challenge. A dry flocked mini-tip swab (product MW002NF; MWE, UK) was used for sampling before adding to 1 mL Virocult universal transport media (product MW951T; MWE, UK). On day 2 post-challenge, a nasal wash was conducted by instillation of 200 μL PBS into each nare with a flexible feeding tube and collection of fluid extract.
At the end of the study, animals were anaesthetised with isoflurane followed by a lethal dose of sodium pentobarbitone delivered via the intraperitoneal route. During necropsy, a sample of lung was collected for molecular analysis. The thoracic pluck (consisting of lung, trachea and associated structures) and the head were immersed in 10% neutralbuffered formalin for histological examination.

Focus-forming unit (FFU) assay
Pharyngeal swab and nasal wash samples collected at day 2 were quantified for live virus using a FFU assay. Samples were serially diluted before adding, in duplicate, to a VeroE6 cell monolayer in a 96-well flat bottomed culture plates (seeded 24 h before) for 1 h at 37 • C. Samples were removed and overlay media added, then incubated for 24 h at 37 • C. Plates were fixed overnight by adding 20% formalin and then fumigated before staining using the same techniques as for the neutralisation assay.

Pathological studies
The left lung lobe and a sagittal section of the entire nasal cavity were fixed by immersion in 10% neutral-buffered formalin, processed and embedded into paraffin wax. The nasal cavity was decalcified using an EDTA-based solution prior to embedding. Sections of 4 μm were cut and stained with hematoxylin and eosin (HE) and examined microscopically. In addition, samples were stained using the RNAscope technique to visualise SARS-CoV-2 virus RNA. Briefly, tissues were pretreated with hydrogen peroxide for 10 min (room temperature), target retrieval for 15 min (98-101 • C) and protease plus for 30 min (40 • C) (Advanced Cell Diagnostics). A V-nCoV2019-S probe (Cat No. 848561, Advanced Cell Diagnostics) was incubated on the tissues for 2 h at 40 • C. Amplification of the signal was carried out following the RNAscope protocol using the RNAscope 2.5 HD Detection kit -Red (Advanced Cell Diagnostics).
All slides were scanned digitally using a Hamamatsu S360 digital slide scanner and examined using ndp.view2 software (version 2.8.24). Sections were examined by a qualified veterinary pathologist who was blinded to the animal and treatment groups. A semi-quantitative histopathology scoring system was used to evaluate microscopic lesions in the lung and nasal cavity (reported elsewhere (Dowall et al., 2021)); in addition, 'Nikon NIS-Ar' software (version 5.21.02) was used to perform digital image analysis to calculate the percentage area of pneumonia and quantify the presence of viral RNA in lung sections. For nasal cavity, a semiquantitative scoring system was applied to evaluate the presence of virus RNA: 0 = no positive staining; 1 = minimal; 2 = mild; 3 = moderate and 4 = abundant staining.

Statistical analysis
Statistical analyses were performed using MiniTab, version 16.2.2 (Minitab Inc). A non-parametric Mann-Whitney statistical test was applied to ascertain significance between groups. A significance level below P = 0.05 was considered statistically significant.

Seroconversion of immunised sheep
All sheep immunised with recombinant SARS-CoV-2 proteins exhibited strong antibody responses with the 1st sample collected after two immunisations (Fig. 1). Following subsequent monthly immunisations, the titres were maintained at equivalent high levels. Plasma collected at all of the timepoints were therefore pooled and purified to produce stocks of purified IgG.

Binding of purified antibody preparations to whole spike protein and subunits
Purified IgG pooled from plasmapheresis events was tested alongside antibody preparations that had been affinity-purified to remove nonspecific antibodies. Antibodies generated in response to immunisation with full-length spike proteins and the S1 or S2 subunits all recognised recombinant whole spike protein (Fig. 2a). As expected, antibodies generated from sheep immunised with S2 antibodies produced low levels of S1-specific responses (Fig. 2b) and similarly S1-derived antibodies demonstrated low responses to S2 antigen (Fig. 2c). The affinitypurified preparations consistently gave higher binding compared to the parent purified antibody preparations.

Functional activity of affinity-purified antibodies
Antibodies were tested for their ability to neutralise live viral strains from early in the pandemic (Wuhan-like virus, Victoria) and a recent divergent strain (Omicron BA.1 variant). The antibody raised against the full-length spike protein demonstrated the greatest neutralisaing activity against the Wuhan-like virus with an IC50 value of 49 ng/ml. Antibodies raised against the S1 and S2 spike protein also had neutralising activity with IC 50 values of 1.071 and 0.428 μg/ml respectively (Table 1). All preparations had strong neutralisation activity to both SARS-CoV-2 strains, although ~10-fold higher antibody concentrations were required to neutralise the Omicron BA.1 variant.
To determine the breadth of in vitro activity of antibodies, a range of tests were undertaken. Firstly, binding specific to the receptor-binding site was assessed, with both the full-length and S1-specific antibodies demonstrating strong recognition, whereas the S2-specific antibody did not bind (Fig. 3a). Antibody-dependent complement deposition was observed with all three antibody preparations, with lower but similar levels observed for S1-and S2-specific preparations compared to those derived from sheep immunised with full-length spike protein (Fig. 3b). Using an antibody-dependent neutrophil phagocytosis assay, the fulllength produced antibodies had higher activity than S1-derived antibodies, whereas the S2-derived antibodies were less opsonophagocytotic (Fig. 3c). Antibodies raised against the full-length spike protein reacted strongly to the RBD and spike antigens from different SARS-CoV-2 variants, including alpha (B.1.1.7), beta (B.1.351) and gamma (P.1),

Table 1
Neutralisation activity of two divergent strains of SARS-CoV-2 by ovine antibody preparations. IC 50 values from 2 to 3 independent experiments with 95% confidence intervals around the probit midpoint (IC 50 ) and overall geometric mean IC 50 . whereas the S1-raised preparation cross-reacted only with the RBD of the variants, and S2 showed limited cross-reactivity ( Table 2). The level of interaction of the two subunit antibody preparations were consistently lower than those observed with the full-length preparation, with the S2 antibodies only reaching inhibition levels of 50.7% and below.

Protective effect of affinity-purified antibodies against SARS-CoV-2 disease in hamsters
To assess the ability of antibodies to protect against SARS-CoV-2 disease, hamsters received 2 mg antibody 24 h before challenge. In addition, a further group received antibodies specific for full-length spike protein 3 days post-challenge for therapeutic effect. Results demonstrated that in comparison to a PBS control group, hamsters that received the antibodies lost less weight, with the groups that received full-length or S2 specific antibodies showing no evidence of any weight loss (Fig. 4a). When the maximum weight loss of individual animals was analysed, all control animals lost weight, whereas weight loss in other groups was less marked (Fig. 4b). The differences between the PBS group and the full-length and S2 groups were statistically significant (P = 0.0202 and 0.0051, respectively) whereas significance was not reached for the S1 and delayed full-length treated groups (P = 0.0656 and 0.2623, respectively). The clinical score of animals which received either the full-length spike, S1 subunit and S2 subunit antibodies were markedly lower than the PBS control group (Fig. 4c). Observations in the S2 subunit group were only the minimal sign of ruffled fur which was seen in two animals. In the PBS control group, three animals met humane endpoints during the course of the study; but this difference was just outside of statistical significance (P = 0.056, Log-Rank survival test).

Antibody levels at time of challenge
Prior to challenge, a blood sample was collected to enable the concentration of circulating antibody to be evaluated. Results showed that in the full-length spike and S1 subunit groups, one animal had low amounts of antibodies (Fig. 5a). To determine whether this was the cause of a failure to protect against disease, a correlation was plotted with the maximum weight loss which demonstrated that the animals with low ovine-specific antibody concentrations were those which showed a more severe clinical disease (Fig. 5b).

Effects of antibody preparations on viral loads in hamsters after SARS-CoV-2 challenge
Nasal wash and a pharyngeal swab were collected 2 days postchallenge to assess the levels of live SARS-CoV-2. In animals which received antibody treatment prior to challenge, there were lower levels of live virus in the nasal wash and pharyngeal swabs samples (Fig. 6a); however, these did not reach statistical significance with the full-length group being the closest in the nasal wash sample (P = 0.0782, Mann-Whitney test). Pharyngeal swabs were collected at regular times throughout the study and tested for the presence of viral RNA. Levels remained similar across animals who received antibodies compared to the PBS control group until day 8, where significantly lower levels of viral RNA were detected in the animals which received antibodies raised to the S2 subunit (P = 0.0282 on day 8 and 9 timepoints, Mann-Whitney test) (Fig. 6b). At necropsy, viral RNA levels in the lung were assessed with levels in the PBS control animals which met humane clinical endpoints were higher than those which survived to the scheduled end of the study (Fig. 6c); these were excluded from statistical comparison with the groups which received antibodies due to samples being collected at a different timepoint. Viral RNA levels in the lungs were lower in several animals in the groups that received antibody compounds prior to challenge, reaching significance in the S2 subunit group (P = 0.0282, Mann-Whitney test).

Histopathological changes in hamsters given ovine antibody preparations after challenge with SARS-CoV-2
Lesions consistent with infection with SARS-CoV-2 were observed within the left lung lobe and nasal cavity in animals in both the treated groups and the control group. Lung lesions consisted of a multifocal to coalescing broncho-interstitial pneumonia with areas of consolidation. Inflammatory cells, primarily macrophages and neutrophils with some lymphocytes and plasma cells, infiltrated alveolar walls and filled alveolar spaces, with cell degeneration and loss, and patchy alveolar oedema; prominent type II alveolar hyperplasia was noted in some areas. The airways were also infiltrated by similar inflammatory cells. Whilst the bronchi contained fewer inflammatory cells, changes in  Data show percentage inhibition of ACE2 binding using a MesoScale Discovery assay with each antibody preparation being standardised to a concentration of 1 mg/ml.
bronchioles, particularly at the bronchioloalveolar junctions, were of increased severity, with concomitant epithelial cell degeneration and loss. Bronchi and bronchioles were surrounded variably by lymphocytes and other inflammatory cells. Lymphocytes were also noted surrounding blood vessels and occasionally infiltrating the walls. The lesions in the nasal cavity were characterised by presence of exudates (fluid with inflammatory cells, mainly neutrophils but also mononuclear cells) and degeneration and necrosis of the epithelium in the respiratory and olfactory mucosa. Representative images of microscopic changes are shown in Fig. 7. Overall, lesion severity was significantly less in the lung of animals given full-length and S2 ovine antibodies compared to those receiving the negative control (P = 0.0374 and P = 0.0163, respectively, Mann-Whitney test) (Fig. 8b). Furthermore, the percentage of consolidation   in the lung, which represented the severity of pneumonic change, was reduced in the S2 antibody group and reached statistical significance (P = 0.0202, Mann-Whitney test) (Fig. 8a). In the nasal cavity, a reduction in lesion severity only reached significance in the S2 ovine antibody group (P = 0.0250, Mann-Whitney test) (Fig. 8c). Whilst the histological outcomes in the S2 antibody group were significant across the different analysis, the results demonstrate lower levels of lung consolidation and histological scores in many of the animals which received antibody treatment prior to challenge, indicating an effect in the majority of animals but little effect when administration was delayed to 3 days postchallenge (Fig. 8a-c).

Discussion
Our work has established proof-of-concept evidence that hyperimmune ovine immunoglobulins can confer protective effects against SARS-CoV-2 infection using a stringent preclinical animal model. Antibodies were raised against three antigens: whole spike protein and its two constitutive parts, the S1 and S2 subunits. This contrasts with others who have focused on generating responses to the RBD region. An equine polyclonal preparation using the RBD of SARS-CoV-2 as the immunogen demonstrated neutralisation activity but without efficacy testing in a disease model (Zylberman et al., 2020). Similarly, genetically-modified pigs have been used to generate polyclonal glyco-humanised antibodies to develop a candidate product named 'XAV-19' which is being assessed in phase 2 trials, but similar to the equine preparation were immunised with only the RBD (Gaborit et al., 2021). Whilst RBD-specific antibodies have been demonstrated to be immunodominant in SARS-CoV-2 infection (Premkumar et al., 2020), antibodies which bind the S1 subunit outside of the RBD have also been shown to confer neutralising capacity (Chi et al., 2020;Jiang et al., 2020). Therefore, our approach also covers the areas outside of the RBD.
Antibody-based therapeutics suffer the risk of mutational escape of the target antigen, such as spike protein . Studies have demonstrated that SARS-CoV-2 variants with mutations in the RBD can be readily selected (Weisblum et al., 2020). For monoclonal antibodies, this can be overcome by use of noncompeting antibodies (Baum et al., 2020), but the approach of targeting a broad area of the spike protein with polyclonal sera reduces mutational escape risk even further. Indeed, given the rise of variants with mutations in the spike protein (Harvey et al., 2021) the outbreak will continue to present challenges for control, and even more so for those therapies which focus on only a few antigenic epitopes. Given that there are at least 18 monoclonal antibodies against SARS-CoV-2 spike protein that are in clinical trials (Kumar et al., 2021), mutational escape poses a real risk to their effectiveness.
After immunisation of sheep, the antibody titre was at the highest level at week 6 demonstrating the rapid turnaround times achievable with this approach. This is earlier to when sheep were immunised with an Ebola virus antigen where the antibody levels reached maximum levels after 10 weeks (Dowall et al., 2016). When antibodies were affinity-purified, their binding levels to the specific antigen were all Top row, lung-multifocal to patchy areas of pneumonic consolidation (arrows) (H&E); middle row, nasal cavity-inflammation and degeneration of the mucosa with variable luminal exudate (asterisks). Inset, higher power images of nasal epithelium (×800 magnification) (H&E); lower row, nasal cavity-staining for SARS-CoV-2 viral RNA in the mucosa and luminal exudate (in situ hybridisation). consistently increased, confirming removal of unspecific antibodies from the preparation. The three preparations all showed similar binding levels to their respective immunisation antigen but differences in functional activity. The full-length specimen demonstrated strong binding to the RBD and had the strongest neutralising activity against live Wuhan-like virus (Victoria) and Omicron BA.1 in vitro. Hamsters given this preparation also demonstrated a milder clinical disease, with a significant reduction in maximum weight loss and reduction in lung histopathology. However, whilst the S2 antibody did not bind to the RBD, it still exerted neutralising activity. Interestingly, hamsters receiving S2-specific antibodies showed the same significant protective effects seen with the full-length antibodies, but with additional parameters also being significantly improved compared to control animals, including reduction in viral RNA in pharyngeal swabs at day 8 and 9 post-challenge, reduction in RNA levels in lung tissue taken at necropsy, lower levels of pneumonia and a reduced pathology score in both the lung and nasal cavity. The observation of S2-specific antibodies demonstrating neutralisation activity has been documented for SARS-CoV (Zeng et al., 2006). The S2 subunit contains structures that are critical for virus entry into cells, such as the fusion protein, and thus has conserved sequences which may explain the reactivity observed in some samples collected prior to the COVID-19 pandemic (Shrock et al., 2020). The S2-specific neutralisation may be due to several mechanisms, including the S2 subunit altering the structural change of the spike affecting the binding between the virus and its receptor which may prevent formation of a complementary surface between the two required for efficient neutralisation (Li et al., 2005). Other mechanisms include the S1 subunit and receptor binding being reversible so the virus may fail to stably dock without subsequent insertion of the fusion peptide into the host cell membrane functioned by the S2 subunit or simply due to the bulk of antibodies binding the S2 region hindering binding to the receptor (Zeng et al., 2006).
We report two animals which underwent antibody administration via the intraperitoneal route which did not translate into increased levels of circulating antibody in the blood. These animals were maintained in the dataset and analysis due to undergoing all procedures as planned and no substantial reason for exclusion outside of the subsequent measurement of antibody levels. It is not known why the antibody failed to enter the circulation. Using the intravenous route, which is the likely delivery route for use in human administration, is possible in small animal models but requires either very small samples being delivered or catheterisation of animals (Dowall et al., 2013).
In providing interim recommendations on assessing the effectiveness of convalescent plasma therapy, the Association for the Advancement of Blood & Biotherapies (AABB) indicate that the two most important factors are the quality of the product, as measured by neutralising antibody titre, and the disease state of the patients (Cohn et al., 2021). Our findings that the antibodies raised against the S2 subunit which were not highly neutralising and did not bind to the RBD demonstrate that other mechanisms are also involved in protective antibody function. Antibodies which do not neutralise may allow the virus-antibody complex to enter cells via endocytosis where other humoral effector functions may be performed such as antibody-dependent cell-mediated cytotoxicity (ADCC), antibody-mediated phagocytosis, aggregation and potentially immune activation (Excler et al., 2014;Mayr et al., 2017). Our results demonstrated S2 antibody binding to complement, and this activation of the complement system may aid in opsonophagocytosis or induce viral aggregation which can reduce virus levels (Santiesteban-Lores et al., 2021).
In addition to testing as a prophylactic, we also assessed giving the affinity-purified full-length spike Ab 3 days post-challenge to look at therapeutic effect. This timepoint was chosen due to clinical signs first being present in infected hamsters at this timepoint (Dowall et al., 2021). In humans, convalescent plasma has shown clinical benefit when used within 72 h from the onset of symptoms Libster et al., 2021). However, the disease progression in hamsters occurs over a shorter timeframe to disease observed in humans, with weight loss reversed after 6 days and live virus being detected in swabs until day 4 (Sia et al., 2020). To fully assess the prophylactic properties of the antibodies, earlier timepoints should be considered.
One issue with antibodies against SARS-CoV-2, whether induced through vaccination or with direct delivery, is the risk of antibodydependent enhancement (ADE) as has been reported with SARS-CoV and MERS-CoV (Wen et al., 2020). A common mechanism of ADE is that the antibody promotes entry into host cells through the Fc receptor and complement receptor. Given that these antibodies are raised in sheep, this is reduced by no human Fc receptor being present. However, the risks could be further mitigated by further refinement of the antibodies into Fab or F(ab)2 fragments.
Whilst there are many vaccines for COVID-19, this may not be an option for immunocompromised and vulnerable populations where effective therapies are needed, particularly due to being prone to more severe COVID-19 (Focosi and Franchini 2021). Therefore, the results reported within presents an additional avenue applicable for both the current COVID-19 pandemic and future infectious disease threats.

Funding
This research was funded jointly by the National Institute for Health Research (NIHR) and the UK Research and Innovate (UKRI) through the COVID-19 Rapid Response Call for a project on 'Development of an ovine polyclonal immunoglobulin therapy against COVID-19' (Ref MC_PC_19077).

Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Sandra Smith and Neville Pope are employees of International Therapeutic Proteins Ltd. Gareth Humphries and Holger Schuhmann are employees of the Native Antigen Company. All other authors declare no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.