Evaluation of a panel of therapeutic antibody clinical candidates for efficacy against SARS-CoV-2 in Syrian hamsters

The COVID-19 pandemic spurred the rapid development of a range of therapeutic antibody treatments. As part of the US government's COVID-19 therapeutic response, a research team was assembled to support assay and animal model development to assess activity for therapeutics candidates against SARS-CoV-2. Candidate treatments included monoclonal antibodies, antibody cocktails, and products derived from blood donated by convalescent patients. Sixteen candidate antibody products were obtained directly from manufacturers and evaluated for neutralization activity against the WA-01 isolate of SARS-CoV-2. Products were further tested in the Syrian hamster model using prophylactic (−24 h) or therapeutic (+8 h) treatment approaches relative to intranasal SARS-CoV-2 exposure. In vivo assessments included daily clinical scores and body weights. Viral RNA and viable virus titers were quantified in serum and lung tissue with histopathology performed at 3d and 7d post-virus-exposure. Sham-treated, virus-exposed hamsters showed consistent clinical signs with concomitant weight loss and had detectable viral RNA and viable virus in lung tissue. Histopathologically, interstitial pneumonia with consolidation was present. Therapeutic efficacy was identified in treated hamsters by the absence or diminution of clinical scores, body weight loss, viral loads, and improved semiquantitative lung histopathology scores. This work serves as a model for the rapid, systematic in vitro and in vivo assessment of the efficacy of candidate therapeutics at various stages of clinical development. These efforts provided preclinical efficacy data for therapeutic candidates. Furthermore, these studies were invaluable for the phenotypic characterization of SARS CoV-2 disease in hamsters and of utility to the broader scientific community.


Introduction
In the absence of proven efficacious COVID-19 therapies early in the Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) pandemic, viable options included the use of convalescent plasma and monoclonal antibodies (mAb). The use of convalescent plasma or therapeutic antibodies had seen some success treating a range of viral infections, including Ebola, influenza, and polio, in addition to bacterial pathogens (Amoss and Chesney, 1917;Garraud et al., 2016;Hui et al., 2018;Misasi and Sullivan, 2021). There are now many tools available to leverage antibodies as a rapid response to emerging pathogens. For instance, convalescent plasma can readily be collected from previously infected individuals and isolation of highly potent antibodies is becoming a common practice. Additionally, modern technology allows researchers to rapidly identify pathogen targeted B cells (or epitopes) from hosts previously exposed to the pathogen, from which virus specific mAbs can be isolated and expanded (Sun and Ho, 2020). Another approach is the production of immunogen targeted human polyclonal antibodies in transgenic animals or plants (Kallolimath et al., 2021).
Unlike small molecule therapeutics, the manufacturing processes used to develop mAbs are well established and well-characterized preclinical and clinical studies facilitate their rapid progression to clinical use. Within months of the onset of the COVID-19 pandemic there were several clinical trials using antibody-based therapeutics, demonstrating the capacity and capability to rapidly identify and develop new antibodybased therapeutics (Liu et al., 2020;Piechotta et al., 2021;O'Donnell et al., 2021;Li et al., 2020). The rapid progression of convalescent plasma and mAb products developed by Regeneron and Lilly, and Vir Biotechnology/GlaxoSmithKline (Vir/GSK) (Califf, 2022;Corti et al., 2021) to Emergency Use Authorization (EUA) through pivotal studies was possible due to the generally safe profile of monoclonal antibodies and multiple products demonstrating clinical benefit that outweighed the risks. Additional trials have demonstrated that antibody treatments offer optimal clinical impact for those at high risk that have recently been exposed, or in those exhibiting mild disease, as treatment is less effective in individuals that have progressed to severe COVID-19 disease (Bégin et al., 2021;Lattanzio et al., 2021;Piscoya et al., 2021).
In silico and in vitro research identified specific mAb binding epitopes and categorized mAbs into 4 classes (Class I-IV) based on antibody recognition and blocking of the viral spike protein receptor-binding domain (RBD) to angiotensin converting enzyme 2 (ACE2) and on where the antibody binds to the viral spike protein relative to the RBD (Corti et al., 2021;Barnes et al., 2020;Kumar et al., 2021). An additional class of mAbs binds to the N terminal domain (NTD) of the viral spike protein. While mAbs binding to the NTD are not necessarily neutralizing, they can impact viral fusion by potentially limiting conformational changes in the spike protein (Chi et al., 2020;Zhou et al., 2022).
The long history of safe use of therapeutic antibodies, the rapid development of SARS-CoV-2 specific antibodies, and the critical need for COVID-19 treatments in a crisis moved many candidate therapeutics directly into clinical trials without evaluation of product efficacy in nonclinical or preclinical models. In early 2020 there were no preclinical models that developed severe COVID-like illness. Most models for SARS-CoV-2 infection show only mild disease, subsequently limiting their applicability to hospitalized patients with severe disease. While there is mild to moderate disease in Syrian hamster following infection, disease progression is remarkably consistent between those similarly exposed. Disease severity can be reliably scored based on clinical signs, body weight, viral RNA load, and histopathological parameters (Bednash et al., 2022). This predictable disease reproducibility combined with the use of adequate sample sizes facilitates robust preclinical therapeutic efficacy studies. The cost and resources required to test potential therapeutics in hamsters is also a fraction of the cost required for the nonhuman primate model, a model with a modest and variable disease phenotype (Clancy et al., 2022). The development of mouse and hamster models expressing the human ACE2 receptor could be useful for measuring protection from lethality, however, these models have a high frequency of early onset, fulminant neurologic disease which is atypical of the human condition (Barton et al., 2020;Matschke et al., 2020;Solomon et al., 2020). In addition, recent isolates of SARS-CoV-2 have an N501Y mutation in the spike protein that facilitates binding mouse ACE2 and making typical laboratory mice susceptible to virus infection (Shuai et al., 2021;Lam et al., 2021).
As part of a collaborative effort between government institutions and private industry to accelerate the discovery and development of SARS-CoV-2 therapeutics and further develop in vitro assays and animal models, a panel of therapeutic monoclonal and polyclonal antibodies was assembled and tested for the ability to neutralize the USA-WA1/ 2020 (WA-01) isolate. A subset of the candidate therapeutics tested were split between two institutes and further evaluated in the hamster model of COVID-19 disease, using both prophylactic and post-exposure treatment regimens. These efforts were focused on establishing efficacy datasets for large panels of antibodies in the hamster model that could then be correlated with clinical outcomes. This work also serves as a model for the rapid, and comparative, assessment of candidate therapeutics. In addition, these efforts have further validated the hamster as a valuable model for rapidly evaluating the efficacy of COVID-19 antibody treatments against novel SARS-CoV-2 variants and possible future pandemics beyond COVID-19.

Methods
Work in this project was divided between neighboring Research Institutes. While most of the methods were harmonized to the greatest extent possible, some variations in methodology occurred, and these differences are presented in the Methods section.

Therapeutic antibody candidates
Neutralizing antibody (nAb) therapeutic candidates targeting SARS-COV-2 spike protein were generously donated by participating pharmaceutical companies for the U.S. Government COVID-19 response Therapeutics Research Team efforts. Due to confidentiality agreements with the manufacturers, nAbs are described herein with blinded identification codes. Also included in these studies was a plasma sample from a previously infected donor ("convalescent plasma") as well as a plasma sample from an uninfected donor ("normal plasma").

Cells and viruses
Vero 76 cells were acquired from the American Type Culture Collection (ATCC, Manassas, VA, USA) and Vero E6 cells were acquired from BEI Resources Repository (Manassas, VA, USA). Both cell lines were maintained as recommended by the provider.
The Washington strain of SARS CoV-2 (USA-WA1/2020) was isolated in January 2020 from a patient in Washington State, USA and provided to USAMRIID by the Centers for Disease Control and Prevention (CDC). The Biological Reference Material Repository at USAMRIID passaged the material from the CDC seed stock in Vero76 cells at a multiplicity of infection (MOI) of 0.01. The supernatant was collected and clarified at 10,000×g for 10 min. The stock was subsequently characterized for the presence of contaminants by bacterial growth assays, mycoplasma tests (MycoAlert) (Lonza, Walkersville, MD, USA), endotoxin tests (Endosafe) (Charles River Laboratories, Wilmington, MA, USA), as well as inclusivity and exclusivity sequencing and PCR. The stock was free of contaminants and the virus sequence was shown to be consistent with the published sequence (Accession # MN985325). The stock was plaque titrated and given lot number R4719.

Neutralization assay
The neutralization assay used for these studies has been previously described (Bennett et al., 2021). Briefly, Vero E6 cells were seeded into 96-well plates approximately 24 h prior to assay initiation. The test antibody material was mixed with virus at a MOI of 0.5 and incubated at 37 • C/5% CO 2 for 1 h. The antibody/virus mixture was then added to Vero E6 cells and incubated for 24 h at 37 • C/5% CO 2 . Following incubation, the cells were fixed with 10% neutral buffered formalin (NBF) (Thermo Scientific, Kalamazoo, MI, USA) and the plates were removed from the biocontainment laboratory. Cells were washed twice in 1X phosphate-buffered saline (PBS) diluted with purified water from a 10X stock solution (Thermo Fisher Scientific, Waltham, MA, USA) and permeabilized with 0.25% Triton buffer in 1X PBS (Thermo Fisher Scientific) then stained with a SARS-CoV-2 nucleocapsid protein specific antibody (Sino Biological, Wayne, PA, USA) for 1 h at room temperature. The cells were then washed and probed with an Alexa Fluor 594-conjugated secondary antibody (Thermo Fisher Scientific) by incubation for 1 h at room temperature. The cells were washed and counterstained with Hoechst nuclear stain (Thermo Fisher Scientific) and washed. The number of infected cells per field were quantified using an Operetta optical imaging system (PerkinElmer, Waltham, MA, USA) with a minimum of 4000 cells quantified per well.
To quantify the 50% neutralization titer (NT50), the fluorescence signal was plotted against the log 10 of the antibody concentration. A fourparameter logistical analysis was performed on the full dilution series using Prism (GraphPad Software, San Diego, CA, USA). The regression was performed using all four replicates per dilution, and the precise titer was calculated from the regression curve (Bennett et al., 2021). The 90% response (EC90), or absolute neutralization (NT90), was calculated using the equation: where F = fraction of maximal response and H = hill slope.

Hamster efficacy studies
All Syrian hamsters used in these studies were approximately 6week-old and acquired from Envigo (Indianapolis, IN, USA). Hamsters were delivered to the individual facilities and allowed to acclimate for 7-10 days prior to study start. For each study, hamsters (individually housed at the IRF and group housed at USAMRIID) were assigned to 8 groups by a statistician according to body weight and sex. Each study was blinded, and the hamsters were randomized to treatment. Three candidate antibody therapeutics were evaluated in each study with single treatments administered intraperitoneally (IP) at a standardized dose (10 mg/kg) and a company-selected dose. Mock-IP-treated (sham, phosphate buffered saline [PBS]) and mock-intranasal (IN)-exposed hamsters were included as negative controls. In prophylaxis studies, the hamsters were treated IP with antibody or sham 24 h prior to IN exposure to 5 log 10 PFU SARS-CoV-2, while in therapeutic studies, hamsters were treated 8 h following virus exposure. Each hamster was weighed prior to study initiation to determine their baseline weight for calculation of treatment dose. Following virus exposure, each hamster was weighed, and disease severity was semi-quantitatively scored, based on clinical signs, a minimum of once daily. Half of the hamsters in each group were euthanized on day 3 post-exposure and the other half of each group was euthanized on day 7 (Fig. 1). Blood and lung tissues were collected at necropsy for further analysis. None of the hamsters used in these studies reached pre-determined clinical endpoint criteria that required unscheduled euthanasia.

Animal ethics statement
Animal research was conducted under IACUC approved protocols at both the IRF and USAMRIID in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals. The facilities where this research was conducted are fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International and adhere to principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 2011.

Pseudotype virus assay
The USAMRIID pseudovirion neutralization assay (PsVNA) for SARS-CoV-2 is based on engineered vesicular stomatitis virus (VSV) that express a luciferase reporter protein in place of the virus envelope glycoprotein (G) and has been reported previously (Brocato et al., 2020a;Kwilas et al., 2014). Pseudovirions (PsV) were produced in 293T cells. Cells in T-75 flasks were transfected with plasmid pWRG/CoV-S(opt)Δ 21 and then infected with VSV-luciferase ΔG virus. PsV were harvested after 48 h. To perform the neutralization assay, SARS-CoV-2 WA-1 strain PsV was combined with duplicate serum dilution series overnight. The PsV plus serum mixture was then added to Vero 76 cell monolayers in Fig. 1. Hamster efficacy study overview: in vivo pre-exposure prophylaxis and/or post-exposure treatment Three candidate antibody therapeutics were evaluated in each experiment with 2 doses of single treatments administered intraperitoneally (IP) at a standardized dose (10 mg/kg) and a company-selected dose. Mock-IP-treated (sham, phosphate buffered saline [PBS]) and mock-intranasal (IN)-exposed hamsters were included as negative controls. In these studies, the hamsters were treated IP with antibody/sham 24 h prior to IN exposure to 5 log 10 PFU SARS-CoV-2, while in therapeutic studies, hamsters were treated 8 h following virus exposure. Following virus exposure, each hamster was weighed daily. Half of the hamsters in each group were euthanized on day 3 post-exposure and the other half of each group was euthanized on day 7. Following euthanasia, blood and lung tissues were collected at necropsies for further analysis.
clear bottom 96-well microtiter plates. The plates were incubated 18-24 h and then media was removed, lysis luciferase reagent (Promega, Madison, WI, USA) was added, and flash luminescence data were acquired using a luminometer (Tecan, Morrisville, NC, USA). Neutralization titers were interpolated from 4-parameter curves (GraphPad).
The reciprocal of the interpolated dilution that results in a 50% decrease in luciferase activity was reported as PsVNA50 titer. Geometric mean titers were calculated from duplicates.

NIAID Integrated Research Facility
Tissue sections weighing 50-100 mg were collected in pre-weighed OMNI Bead Ruptor Elite (PerkinElmer, Waltham, MA) beating tubes and were frozen on dry ice. Samples were then stored at − 80 • C until plaque assay/PCR were performed. Samples were reweighed at the time of plaque and PCR assays. Phosphate buffered saline was added to tissue samples achieve a 10% suspension, agitated in Bead Ruptor for 3 min, and then centrifuged at 2500 rpm for 10 min to pellet tissue debris. The homogenate supernatants were transferred to a 2 mL Sarstedt tube and left on wet ice for plaquing or RNA extraction.
Live virus titers were determined by standard plaque assay. Vero E6 cells were plated in 6-well plates at a density of 1 × 10 6 cells/well to ensure at least 90% confluence the following day. Ten-fold serial dilutions (diluted in DMEM [Gibco/ThermoFisher, Waltham, MA, USA] with 5% heat inactivated (HI) FBS (Sigma, St. Louis, MO USA) and 1XAntibiotic-Antimycotic (Sigma) of the tested samples were added, in triplicates of 300 μL/well, to individual wells and the virus was allowed to infect the cells for 1 h at 37 • C/5% CO 2 with rocking every 15 min. The cells were then overlaid with 1 mL of 2.5% Avicel overlay diluted (1:1) (f/c 1.25%) (FMC Biopolymer, Philadelphia, PA, USA) in 1 mL of the 2 × EMEM (Gibco/ThermoFisher, Waltham, MA, USA) containing 4% FBS and incubated at 37 • C/5% CO 2 for 48 h. The overlay was removed, and the cells were then fixed with 0.25% Gentian violet (Ricca Chemical, Arlington, TX, USA) in 10% neutral buffered formalin (NBF) (VWR Radnor, PA, USA) for 30 min at room temperature. The plates were washed with running tap water and the plaques enumerated. The lower limit for this assay is calculated based on the lowest weight of tissues of each assay at a plaque count of 10. Lung homogenized tissue was reported as PFU/g.

USAMRIID
Right lung lobes were placed in pre-weighed M tubes (homogenization tubes) and weighed to determine the mass of the lung tissue contained within. Lung tissue was then homogenized in 1.0 mL of cMEM using M tubes on the gentleMACS dissociator system on the RNA setting. M tubes were centrifuged to remove cellular debris (1200×g for 5 min) and supernatants collected in another tube. Ten-fold dilutions of the lung homogenate samples were adsorbed to Vero 76 monolayers (200 μl of each dilution per well) in 6-well plates. Following a 1 h adsorption at 37 • C in a 5% CO 2 incubator, 3 mL per well of agarose overlay ( and 0.2% Fungizone (Sigma Aldrich]) was added and allowed to solidify at room temperature. The plates were placed in a 37 • C/5% CO 2 incubator for 2 days and then 2 mL per well of agarose overlay containing 5% neutral red and 5% HI FBS was added. After one additional day in a 37 • C/5% CO 2 incubator, plaques were visualized and counted on a light box. The lower detection limit for this assay is 2 plaques per well or 100 PFU/mL. Specimens with <100 PFU/mL were given an imputed data value of 70.7 PFU/mL. This value was calculated by dividing the lower limit of the assay, 100 PFU/mL, by the square root of 2, which is a common method to provide a nonzero number below the detection limits of the assay. PFU per milliliter of homogenized lung tissue values and calculated lung weights were used to calculate PFU/g of lung tissue.

Viral RNA determination
Viral RNA determination was performed using an assay developed at USAMRIID and shared with the IRF. Reverse transcription quantitative real-time PCR assays (RT-qPCR) were performed on extracted TRIzol LS diluted samples derived from both lung homogenates and pharyngeal swabs. Briefly, samples in TRIzol LS were extracted using Qiagen's EZ1 extraction robot and Virus Mini Kit 2.0 (Qiagen, Germantown, MD, USA) according to manufacturer's instructions. Samples were tested with 2 PCR assays: One targeting genomic and subgenomic SARS-CoV-2 E gene nucleic acid and one targeting SARS-CoV-2 E gene subgenomic nucleic acid only, as previously described (Corman et al., 2020;Wölfel et al., 2020). Briefly, 5 μL of extracted material, water, or serially diluted standard E gene RNA (Bio-Synthesis, Lewisville, TX, USA) was added to a 20 μL master mix. Assay master mix consisted of 1X Superscript III One-Step RT-PCR System with Platinum Taq (ThermoFisher), 1 μL

Pathology
Tissue samples taken during necropsy were fixed in 10% neutral buffered formalin (NBF) according to facility protocols prior to removal from the biocontainment laboratories. Tissues were routinely processed, embedded in paraffin, mounted on glass slides, and stained with hematoxylin and eosin (H&E). Slides were read by certified veterinary pathologists and semiquantitatively scored using a four-point system based on the distribution interstitial pneumonia: 0% = 0, 1 = ≤25%, 2 = 26-50%, 3 = 51-80%, = >80% within the lung sections examined. The distribution of interstitial pneumonia was semiquantitatively scored as it was the most consistent and salient histopathologic lesion in the golden Syrian hamster model of SARS-CoV-2 that correlated with both clinical disease severity and treatment efficacy, while being the least susceptible to inter-observer variability. In addition to H&E assessments, in situ hybridization (ISH) for the detection of SARS-CoV-2 spike protein mRNA was performed using the V-nCoV2019-S probe (Advanced Cell Diagnostics [ACDBio], Newark, CA, USA). ISH was reported as positive or negative and semiquantitatively scored based on chromogen distribution within the lung tissue section as 0 = 0%, 1 = 1-25%, 2 = 26-80%, or 3 = >80% distribution.

Virus neutralization
A panel of 16 therapeutic antibody candidates were tested in a live virus neutralization assay against the progenitor WA-01 isolate of SARS-CoV-2. Tested antibodies consisted of single monoclonal antibodies (mAbs), mAb cocktails, purified polyclonal IgG, and convalescent plasma. In this assay, most mAbs and mAb cocktails had similar NT 50 ranging from 13 to 543 ng/mL (Fig. 2). One candidate therapeutic (Y) was found to be more effective at neutralizing the WA-01 SARS-CoV-2 isolate than other antibodies with an NT 50 of 13 ng/mL. There was also a subset of four polyclonal candidate antibody therapeutics that were less potent, as might be expected, with NT 50 s calculated between 160,851 (AE) and 332,171 (AJ) ng/mL.

In vivo assessment
The infection of hamsters with SARS-CoV-2 (WA-01) is not a lethal model. In these studies, the loss of body weight relative to day 0 was used as a primary determination of viral pathogenicity with therapeutic efficacy indicated by continued weight gain in these juvenile hamsters. Secondary outcome measures included the presence of viral RNA (subgenomic RNA) and infectious virus in lung homogenates on day 3 and the distribution of interstitial pneumonia in lung samples collected from animals sacrificed at days 3 and 7 post-exposure. In these studies, all uninfected controls continued to gain weight until reaching the predetermined day 7 post-exposure study end, whereas mock-treated, virus-exposed controls lost weight through approximately day 6 postexposure and had begun gaining weight when euthanized at day 7 (Fig. 3, mock groups). Virus-exposed, untreated control hamsters had evidence of virus and viral RNA in lung tissues (Fig. 4, mock groups) and had an interstitial pneumonia at day 3 that became moderate to severe, but in the reparative stage, at day 7 post-exposure. (Fig. 5, mock groups).
In these studies, groups of hamsters were administered a single antibody treatment at a standardized dose of 10 mg/kg or a manufacturer selected dose (varied by candidate) delivered by IP inoculation. Importantly, most clinical stage products are delivered by IV infusion in humans resulting in an almost immediate maximum blood concentration (C max ) while the IP route in hamsters typically required 2-3 days to reach C max (Al Shoyaib et al., 2019;Stauft et al., 2021;Gupta et al., 2016;Yadav et al., 2021). Two different antibody administration times − 24 h prior to virus exposure or 8 h post-virus-exposure were investigated in separate studies at each treatment dose. In each study, serum collected 3 days after virus exposure was evaluated for the presence of SARS-CoV-2 specific antibody using a pseudovirus neutralization assay (Bennett et al., 2021). SARS-CoV-2 specific antibody was detected in serum samples from treated hamsters on day 3 post-exposure in a dose dependent manner (Fig. 6). Titers were generally similar between groups receiving the standardized dose of 10 mg/kg (Fig. 6), yet titers varied in a dose-dependent manner in company selected doses, as would be expected. For the +8h study, neutralizing activity was not detected in either the untreated mock-exposed or sham-treated, virus-exposed negative control hamsters at days 3 or 7 post-exposure. For the − 24h study, there were 5 positives (titer ≤313) in the mock exposure group and one in the mock treatment group on day 3 (titer 743). There was also one positive (low in the normal plasma group (titer 58). Background activity (i.e. false positives) is very rare in this assay, so this cluster of rare events in the − 24h studies is perplexing. We were unable to determine an assignable cause to exclude these data, so they were included in the analysis. There were 4 outliers that did not have neutralizing activity on day 3 whereas the other animals in those groups had high titer activity (titers >1000). These outliers were subsequently removed from statistical analysis. It is possible that these 4 hamsters inadvertently did not receive antibody, received only a partial dose, or the antibody was injected into the GI tract, not the peritoneal cavity.
SARS-CoV-2 specific antibody titers in serum collected on day 3 from hamsters that were administered convalescent plasma and product AE were low in both the prophylactic and post-exposure treatment studies (Fig. 6). The low titers achieved following administration of convalescent plasma are not surprising, as it is an unpurified antibody product and the virus specific IgG concentrations are low. Overall, these results confirmed that serum antibody was present at day 3, following administration the IP antibody treatments, with titers ranging from a PsVNA50 >10 5 to <100, depending on treatment group.

Prophylactic treatment
In studies examining the ability of prophylactic antibody treatment (− 24 h) to prevent disease, body weight loss was minimal in most hamsters and those that lost weight had begun to recover by day 4 (Fig. 3A). Area under the curve (AUC) analyses over the course of the 7day studies found convalescent plasma demonstrated reduced efficacy, as did the 3 mg/kg dose of treatment D, and the standardized dose (10 mg/mg) of treatment AE (Fig. 7 AUC ratio).
However, despite a lack of prolonged body weight loss in most treated hamsters, many had evidence of interstitial pneumonia (Figs. . 5A and 7), with hamsters analyzed at day 3 post-exposure typically scoring lower than those examined at day 7. Candidate treatments I, Z, S and AQ each largely prevented the development of interstitial pneumonia, while convalescent plasma and candidate treatments Y, AE, AC, D and AL were less effective using at least one of the selected doses. A caveat, the confidence intervals and subjective semi-quantitative scoring system used, reduce the value of these analyses.
The concentrations of viable virus or viral RNA in lung tissue and semiquantitative interstitial pneumonia histopathology scores were highly correlated (Fig. 8). Untreated, infected hamsters had viral titers in the range of 7-8 log 10 PFU/g tissue while most treated hamsters had titers <7 log 10, with some having no detectable virus (Fig. 4A). Viral RNA levels correlated very well with viable virus titer data with some treated hamsters having no detectable virus or viral RNA in the lungs (Fig. 4A).
These data demonstrate that nearly all antibodies were effective at preventing significant disease in this model when given 24 h prior to

Fig. 2. Neutralization of authentic SARS-CoV-2 (the progenitor WA-01 isolate) by a panel of 16 therapeutic antibody candidates
The 50% neutralization titers were calculated referencing "virus only" controls for each run. A fourparameter logistical analysis was performed on the full dilution series using Prism (GraphPad Software, San Diego, CA, USA). Analysis was done in quadruplicate. The regression was performed using all four replicates per dilution and the precise titer was calculated from the regression curve. Most monoclonal therapeutics tested had similar neutralization curves with EC50 values around 10 2 ng/mL. Polyclonal products (AE, AG, AJ, Q) require higher concentrations with EC50 values around 10 5 ng/mL. virus exposure. Two candidate therapeutics (AE and D), along with convalescent plasma, were not effective at preventing weight loss at the lower dose tested, while 4 candidate therapeutics (AQ, I, S and Z) were nearly uniformly effective at preventing disease when used as a prophylactic treatment, especially at the higher dose tested (Fig. 7).

Therapeutic treatment
Delivery of antibodies as a therapeutic is a more stringent method for determining protective efficacy, especially due to the delayed PK expected when administered IP versus the IV route used in the clinic. In these studies, hamsters were treated 8 h following virus exposure and half were sacrificed on days day 3 post-exposure and the other half were euthanized on day 7, with samples collected for lung histopathology and viral load. Body weight loss or decreased gain was generally more pronounced and more prolonged in the treatment group than in the prophylaxis studies (Fig. 3B). Although modest body weight recovery was evident in most of the treatment groups, many of the treated hamsters failed to recover body weight when compared to uninfected controls by day 7. Unlike the prophylactic treatment studies, AUC analyses found that body weights in all virus-exposed, treated groups were well below those of uninfected controls (Fig. 7). Several treatment groups, including

Fig. 4. Lung tissue virus titer and viral load detected by qPCR
Virus titers (upper panel) and subgenomic viral RNA (lower panel) from lung tissues collected at day 3 post-exposure were measured after (A) prophylactic or (B) post-exposure administration of antibodies in SARS-CoV-2-exposed hamsters. There was significant viral shedding observed in mock-treated controls. Minimal viral shedding was observed on day 7. Samples with PFU/g (titer) or subgenomic RAN gene copies/mg (gene copies) calculated to be below the assays lower limit of quantification (LLOQ) are represented using unfilled circles, while values above the LLOQ are represented using filled circles.

Fig. 5. Interstitial pneumonia (IP) lung pathology scores
Lung sections taken at days 3 and 7 post exposure were examined by light microscopy and semiquantitatively scored on a scale ranging from 0 to 4. Scoring based on the percentage of interstitial pneumonia: 0: Below lower limit of perception (LLOP, estimated by model); 1: LLOP -≤25%; 2: 26-50%; 3: 51-80%; 4: ≥80%. Protection from SARS-CoV-2-induced pneumonia was determined following prophylactic (A) or post-exposure (B) antibody administration. convalescent plasma and candidate treatments Z, AC and D, showed little therapeutic benefit in preventing weight loss when compared to infected, mock-treated hamsters. Two candidate therapeutics, M and AQ, showed the most benefit, particularly at the highest dose tested.
Unlike the prophylaxis trial where many hamsters sampled at day 7 had less interstitial pneumonia than those sampled at day 3, in the therapeutic trial nearly all hamsters had some evidence of pneumonia (Fig. 5B). For candidate treatment AQ, despite showing some therapeutic benefit based on body weight loss, most hamsters with the group had evidence of marked pneumonia histopathologically. Interestingly, many of the treated hamsters had low levels of viable virus in their lungs at day 3 post-infection (Fig. 4B). With the exception of a few hamsters in the groups treated with candidate treatments AQ, I, M and S viral loads did not correlate with levels of viral RNA in most cases (Fig. 8), as viral RNA levels were high in most hamsters. While there were good correlations between measured parameters in animals treated prophylactically (− 24h), the only correlations in the treatment study were between body weight AUC and day 3 PCR and plaque assay data. This demonstrates that the therapeutic studies produced a less robust dataset for evaluating treatment efficacy.

Discussion
The use of therapeutic antibodies in the treatment of viral diseases is common and began with the use of convalescent plasma over a century ago (Garraud et al., 2016). The development of technologies that can rapidly identify and expand human B cells that produce functional mAbs has made a significant impact on the development of targeted therapeutics. Other technologies, such as the development of transgenic animals or plants producing fully human IgG, have provided an alternative means of producing antibody-based therapeutics without the need for blood collection from previously infected or vaccinated individuals. The COVID-19 pandemic highlighted these capabilities with efficacious antibody-based therapeutics made available for clinical trials within 6

Fig. 6. Pseudovirus neutralization antibody measured in serum collected on day 3
The serum dilution required to achieve 50% neutralization of SARS-CoV-2 pseudovirus, an indication of the presence of a nAb therapeutic in the hamster serum, were measured in prophylactic (A) or post-exposure (B) experiments in SARS-CoV-2-exposed hamsters. In treated hamsters, most serum samples showed 50% neutralization of pseudovirus at high dilutions, indicating that antibodies were present in the serum at day 3. months of the pandemic's onset. These products included purified IgG from humans or transgenic animals, mAb cocktails, and individual mAb therapeutic approaches (Corti et al., 2021).
In the studies reported here, the Therapeutics Research Team, as part of the U.S. government COVID-19 response (Operation Warp Speed), undertook an effort to evaluate a set of candidate therapeutic antibodies, produced by multiple entities, in both cell culture and in vivo studies. The objectives were to directly compare neutralizing titers using a common assay, and protective efficacy using both prophylactic and therapeutic approaches in the hamster model of SARS-CoV-2 infection. Clinical data from trials designed to assess the efficacy of antibody therapies for treating COVID-19 has made it clear that the use of therapeutic antibodies has the most benefit when provided very early in the disease course, with very limited therapeutic value in severely ill patients (Bégin et al., 2021;Lattanzio et al., 2021;Piscoya et al., 2021). The work described here supports those findings, with most treatments showing good efficacy in the hamster model when provided prior to infection, but with diminished efficacy, especially at preventing interstitial pneumonia, when treatment was administered 8 h after virus exposure.
Initial evaluation of candidate therapeutics using a microscopic imaging-based immunofocus assay found the NT 50 against SARS-CoV-2 WA-01 for most individual mAbs, or mAb cocktails, fell within a relatively close range (Fig. 1). Several candidate therapeutics were not as effective at neutralizing the WA-01 isolate of SARS-CoV-2 when compared to mAbs. The reduced inhibition by polyclonal treatments is not surprising given that the composition of active (i.e. neutralizing) versus inactive antibodies is unknown in these treatments, however increasing the dose administered in clinic can increase treatment efficacy. The determined NT 50 values in this assay were consistent with the data provided from manufacturers. Three candidate therapeutics (D, Y and Z) had particularly low NT 50 values (i.e. high inhibitory capacity) suggesting that these candidates could have enhanced therapeutic benefit.
Given the limited utility of laboratory mice as a disease model for early isolates of SARS-CoV-2 and the susceptibility of hamsters to wildtype SARS-CoV-2 infection, the hamster model was chosen for these studies. Hamsters do not develop severe disease following SARS-CoV-2

Fig. 7. In vivo assay overview
Colored boxes correspond to protective effects of therapeutics in study experiments compared with controls.

Fig. 8. Prophylactic (-24h) and post-virusexposure (þ8h) antibody treatment correlations across endpoints
Numerals indicate Spearman rank correlations by endpoint. Comparison of endpoints across prophylactic and post-virus-exposure treatment studies. In vitro live virus neutralization assays did not correlate well with in vivo virology or pathology assays, but correlated well with pseudovirus neutralization serology. Pathology endpoints (including body weight loss AUC and interstitial pneumonia) showed some correlation to virology endpoints including reduction in viral shedding (PCR) and plaque formation using day 3 lung samples (− 24h only). Correlations were less clear overall in the post virus-exposure treatment (+8h) study. Virology endpoints (viral plaque and PCR) were closely correlated to each other in the pre-exposure prophylaxis (− 24h) studies but less correlative in the post-exposure treatment (+8h) studies. Serology endpoints (pseudovirus neutralization) were most clearly correlated to reduction in plaque formation (− 24 and +8h), PCR, and body weight change (− 24h only). Overall, endpoints were more highly correlated in the pre-exposure prophylaxis (− 24h) studies when compared to the post-virus-exposure treatment (+8h) studies.
WA-01 infection but do develop moderate clinical disease that can be subjectively scored and quantitatively assessed by body weight loss or a lack of weight gain (Bednash et al., 2022;Sia et al., 2020). Preliminary evaluation of this model in our hands also demonstrated that sampling of a subset of hamsters at mid-disease course (i.e. day 3 post-exposure) allowed for quantification of viable virus and viral RNA in lung tissue as well as assessment of interstitial pneumonia. By day 10 post-exposure, most hamsters had largely recovered from clinically evident disease.
In the development of this project, it was determined that a single consistent therapeutic dose should be evaluated to provide comparability between candidate therapeutics. However, each manufacturer was also provided the opportunity to select a dose that they felt was more appropriate for their candidate therapeutic(s). Each of the 2 doses was initially evaluated using a prophylactic approach with hamsters treated 24 h prior to virus exposure. These prophylactic studies found that most candidate therapeutics effectively protected hamsters from disease. Some hamsters developed mild interstitial pneumonia at day 3 with virus present, indicating partial protection from infection with the presence of mild disease. The limited disease seen in these hamsters was not surprising given the relatively large concentration of therapeutic antibodies provided and the known persistence of human antibody products in hamster blood following IP delivery (Yadav et al., 2021;Davies et al., 2013). The detection of SARS-CoV-2 specific neutralizing antibody levels in serum from treated hamsters days after passive transfer of a human monoclonal antibody is consistent with previous findings in the hamster model (Brocato et al., 2020b). While it is possible that some endogenous IgM antibody would be present in hamsters at day 7, it is unlikely that substantial quantities of virus-specific antibodies would be present at day 3 post-exposure. Subsequently, it is assumed that the virus specific antibody detected is from the candidate therapeutic administered.
While prophylactic treatment identified near uniform protective effect in the hamster model, post-exposure treatment was more useful at delineating the efficacy of individual candidate therapeutics. Nearly all post-exposure treatments, and two therapeutic treatments (AC and AF), reduced viable virus load in the lung, but not viral replication as measured by RT-qPCR for viral genome. The viral RNA detected was likely intracellular, in infected cells within the tissue, as any shed virus would be bound by circulating antibodies. Two candidate therapeutics (M and Y) showed a high degree of protective efficacy in most parameters measured (Fig. 7) with the exception of interstitial pneumonia in hamsters tested at both day 3 and day 7 post-infection. Two other candidate therapeutics (AQ and S) reduced the impact of interstitial pneumonia at both days 3 and 7, but only at the highest dose tested. Although the impact of post-exposure treatment was markedly decreased relative to prophylactic treatment, these data do suggest that some therapeutic candidates may have a wider window for clinical efficacy than others. One challenge with this model, in relationship to clinical treatment approaches, is the delivery route. In the clinic, therapeutic antibodies are delivered IV resulting in a nearly immediate systemic distribution (Gupta et al., 2016). In the hamster model, estimates for C max are days following IP administration. Following a − 24 h prophylactic IP treatment, C max is estimated to occur by day 1-3 post-virus-exposure while following a +8 h post-exposure treatment, C max will be reached 2-3 days post-exposure (Yadav et al., 2021;Davies et al., 2013).
We discovered that most of the mAbs tested had a similar ability to block virus infection when tested for neutralization in a cell culture assay with WT virus. In addition, in vivo studies showed that nearly all candidate therapeutics tested were effective at reducing the viral burden in the lungs, as determined by plaque titration, PCR, and ISH. They also improved the severity of interstitial pneumonia relative to mock-treated hamsters when administered prophylactically (− 24 h), an approach that ensured a reasonable quantity of candidate therapeutic antibody was in circulation at the time of virus exposure. However, most of the candidate therapeutics tested were less effective when given 8 h after virus exposure. This therapeutic approach and IP dosing delayed the appearance of the candidate therapeutic in the circulation and allowed virus to effectively seed target tissues. These data support clinical findings demonstrating limited efficacy of therapeutic antibodies when given to hospitalized patients with severe COVID-19, but they do suggest that antibody treatment with any of the candidate therapeutics tested could be effective if provided prophylactically or to patients with mild or moderate COVID-19 that are also at a high risk for hospitalization or death. Furthermore, it is possible that the combination of supportive care in a clinical setting, but not modeled in hamsters, combined with therapeutic antibody would result in improved clinical outcome.
This study is unique in that it directly compared a panel of candidate therapeutic antibodies from different manufacturers in the preclinical stages or following emergency use authorization for the treatment of COVID-19. The design of this project was not to support or refute the use of certain candidate therapeutics but to systematically evaluate a broad range of candidates, to challenge the validity of the hamster model in predicting clinical outcomes, and to provide a baseline of data for an early lineage virus for comparison to new variants that might be difficult to assess clinically. These data clearly demonstrate that most candidate therapeutics tested could have had value as a prophylactic treatment or possibly for treatment shortly after a known exposure. However, it is similarly clear that most of these candidate therapeutics were unlikely to be useful for the treatment of advanced or severe disease. Furthermore, this study demonstrates the value of the hamster model for testing candidate therapeutics designed to treat COVID-19, as the disease course in this model is very consistent and can be quantified by several different measures including virus specific PCR, live virus titration, and semiquantitative histopathological scoring of lung tissue lesions by a qualified veterinary pathologist. This small animal model facilitates the use of appropriate sample sizes for statistical power and does not require the expense and infrastructure necessary to support nonhuman primate studies. Standardized assays may be a quick screening tool to prioritize compounds in need of USG funding for accelerated development in a time-constrained, pandemic setting.
When considering the challenges of pandemic preparedness, it is worth noting that both the IRF and USAMRIID have the capacity to conduct animal studies in the highest level of containment, ABSL-4. The combined bandwidth achieved when these two research facilities join forces and harmonize methods is substantial and could be a model for responding to future pandemics, especially if they involve agents requiring ABSL-4.
These in vitro and in vivo models of SARS-CoV-2 offer insights into clinical therapeutic efficacy that appertains to preventing infection or limiting viral replication, shedding, and spread; increasing the likelihood that a candidate therapeutic will reduce SARS-CoV-2-associated morbidity and mortality in humans. This dataset provides a baseline to compare forthcoming clinical data, with the goal of validation of the animal model as a tool to predict clinical outcomes. This established method for efficacy testing will be applied to emerging SARS-CoV-2 variants of concern.
Funding This project has been funded in whole or in part with federal funds from the National Institute of Allergy and Infectious Diseases ( The animal study protocol at the IRF was reviewed and approved by the NIH NIAID DCR IRF-Frederick Institutional Animal Care and Use Committee (IACUC) and the animal study at USAMRIID was reviewed and approved by the USAMRIID IACUC in compliance with all applicable federal regulations governing the protection of animals and research. The facilities where this research was conducted adhere to principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 2011.