https://orcid.org/0000-0002-8324-3694
Abstract
Emerging variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) possess mutations that prevent antibody therapeutics from maintaining antiviral binding and neutralizing efficacy. Monoclonal antibodies (mAbs) shown to neutralize Wuhan-Hu-1 SARS-CoV-2 (ancestral) strain have reduced potency against newer variants. Plasma-derived polyclonal hyperimmune drugs have improved neutralization breadth compared with mAbs, but lower titers against SARS-CoV-2 require higher dosages for treatment. We previously developed a highly diverse, recombinant polyclonal antibody therapeutic anti-SARS-CoV-2 immunoglobulin hyperimmune (rCIG). rCIG was compared with plasma-derived or mAb standards and showed improved neutralization of SARS-CoV-2 across World Health Organization variants; however, its potency was reduced against some variants relative to ancestral, particularly omicron. Omicron-specific antibody sequences were enriched from yeast expressing rCIG-scFv and exhibited increased binding and neutralization to omicron BA.2 while maintaining ancestral strain binding and neutralization. Polyclonal antibody libraries such as rCIG can be utilized to develop antibody therapeutics against present and future SARS-CoV-2 threats.
With high rates of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) circulating worldwide, novel and divergent transmitted strains are constantly emerging. Therapeutics for postexposure treatment and prophylactic protection are critical for individuals who are infected or at risk of becoming infected [1–4]. Although the current vaccines continue to provide protection against severe disease for all SARS-CoV-2 strains thus far, infection and transmission still occur in the vaccinated population [5, 6]. Globally, infection rates remain high, putting susceptible immune-naive and immune compromised individuals at risk of increased coronavirus disease-related morbidity and mortality. Vigilant identification of individuals belonging to at-risk populations is crucial so that efficacious interventions can be administered when required [7]. Therefore, viral surveillance needs to be widely implemented to enable the early identification of emerging variants and inform antibody therapeutic effectiveness in advance of community spread [8–10]. For epidemiologic surveillance, implementing tools like pseudoviral neutralization assays can characterize current immune responses or treatment modalities and their effectiveness against current circulating strains. Such methodologies will provide early indicators of treatment success or viral resistance. However, the lag time between the detection of a novel circulating variants and the development of a neutralization assay for that specific variant can pose some delays. Altogether, if implemented properly and efficiently, this information can provide effective guidance for treating early infection and preventing serious disease.
Within the current treatment landscape, there are limited antiviral and antibody drugs authorized for use to mitigate disease or prevent infection. Remdesivir and nirmatrelvir-ritonavir (Paxlovid) have become common and highly effective first-line treatment modalities but can only be provided to treat postexposure cases and require many doses to be effective. Remdesivir has reduced the risk of death in studies [11], and although Paxlovid has been shown to be effective in suppressing viral replication, its use has been hampered by viral rebound after treatment course completion [12]. An advantage of antibody drugs compared with small molecule antivirals is their ability to be administered in a single dose and provide protection over a long prophylactic and protective half-life where antivirals require repeated dosing and are contraindicated for prophylactic treatment. Some of these monoclonal antibody (mAb) treatment modalities include bamlanivimab, casirivimab, imdevimab, sotrovimab, cilgavimab, tixagevimab, and etesevimab. When single mAb usage is determined to be unsuccessful at neutralizing viruses [13–15], combinations determined to complement each other by binding different regions of the receptor binding domain (RBD) were incorporated as therapeutic options for treating patients (etesevimab/bamlanivimab, imdevimab/casirivimab, and cilgavimab/tixagevimab) [16–18]. Soon after omicron became prevalent, in November of 2021, these combination drugs began to lose efficacy [19–21]. Overall, this demonstrates that mAbs used as a monotherapy, or even as a 2-mAb cocktail, are susceptible to significant viral immune evasion, and a true polyclonal strategy may be required to provide the level of coverage that would thoroughly obfuscate immune evasion. The selective pressure put on SARS-CoV-2 by increased population immune memory has revealed that antibodies with epitopes in mutational hot spots are at risk for losing binding, leading to the investigation of the N-terminal domain (NTD) as an alternative therapeutic target [22–24]. At the time of writing this manuscript, November 2022, the only remaining US Food and Drug Administration-approved antibody therapeutic to still be effective to treat the extant dominating strain (BA.4/5) is bebtelovimab [20, 25]. However, the omicron subvariants XBB (a recombination of BA.2.10.1 and BA.2.75 sublineages) and BQ1 (a sublineage of BA.5) are already raising concern, because early testing reveals increased infectivity and reduced antibody sensitivity [26]. This is a reminder that the lifespan of any mAb's therapeutic efficacy seen during this pandemic has been noticeably short lived.
Recombinant polyclonal SARS-CoV-2 immunoglobulin (rCIG) is a B cell-derived, high-titer antibody therapeutic containing 12 500 unique antibody sequences that target the ancestral SARS-CoV-2 RBD [27, 28]. The urgency for therapeutic interventions during the prevaccine phase of the pandemic drove the rapid 10-month development timeline of rCIG, which was enriched using the ancestral RBD antigen, from B cell collection to good manufacturing practices (GMP)-manufactured drug candidate [28]. It was derived from 16 high potency SARS-CoV-2 convalescent donors collected in April 2020, from which B cells were processed in a microfluidic-droplet-based molecular antibody library generation platform (Surge), which maintains native heavy and light chain pairing and antigen specificity. The antibody sequences that bind SARS-CoV-2 ancestral strain RBD were selected for inclusion in the original rCIG polyclonal drug candidate. Using en masse cloning for single-site-directed transfection in a Chinese hamster ovary (CHO) cell line, a master cell bank was created, and current GMP (cGMP) drug was manufactured for clinical studies using the same methods for production and purification of monoclonal antibodies [28]. We later generated a second version of rCIG, termed rCIG-HY (high yield), that utilized a higher producing cell line to generate a higher antibody production yield. The rCIG-HY contains the same antibody sequences as rCIG and at a similar clonal frequency as rCIG. Here, rCIG was characterized for its ability to block angiotensin-converting enzyme 2 (ACE2) binding and pseudoviral neutralization for a wide range of World Health Organization (WHO) variants of concern. To demonstrate the flexibility of this recombinant hyperimmune platform and to create polyclonal antibody libraries against new SARS-CoV-2 variants of concern, we created an omicron-specific recombinant polyclonal library by enriching for omicron (BA.2) antibody sequences contained in the original rCIG library and investigated its binding and neutralization capabilities. We show that antibodies originating from convalescent donors infected with the ancestral SARS-CoV-2 strain provide a foundationally diverse antibody repertoire capable of recognizing past and potentially future SARS-CoV-2 variants, forming a new, broadly neutralizing therapeutic.
METHODS
Antibody Discovery and Manufacturing
The rCIG was generated, as previously described, from 16 convalescent donors sampled 26 ± 14 days from SARS-CoV-2 symptom onset [27]. Local ethical regulations were followed, and informed consent was obtained for all human sample collection. Antibody production runs of rCIG, rCIG-HY, rCIG-HY BA.2(+), and rCIG-HY BA.2(−) from CHO cells averaged 10 days, after which the harvested cell culture fluid was filtered and purified as previously described [28]. The rCIG-HY was generated in the same manner as rCIG from the same donor libraries, except it was cloned into CHO cells with the translational enhancer 2G UNic (Proteonic, Leiden, the Netherlands). Bamlanivimab, casirivimab, and imdevimab were cloned and produced from publicly available sequences.
Generation of BA.2-Specific Antibody Surface Display Yeast Libraries
Although the rCIG antibody library was made by enriching 8 individual yeast surface display libraries for antiancestral strain RBD antibody sequences [27], to identify and enrich omicron BA.2 binding or nonbinding sequences, the 8 previously ancestral-RBD-enriched single-chain variable fragment (rCIG-scFv) yeast surface display libraries were pooled and panned. The rCIG-scFv BA.2(+) and rCIG-scFv BA.2(−) libraries were made by simultaneously enriching and depleting the yeast surface display rCIG-scFv library through 2 rounds of 1200-nM staining with the omicron BA.2 RBD protein (SPD-C522g, AcroBiosystems) (Supplementary Figure 1). The signal for positive and negative binding was determined by staining the yeast antibody library of an unrelated campaign with the BA.2 antigen as a control to set thresholds for gating on the antigen-binding population with FACSMelody software (BD Biosciences).
Variant of Concern Library Cross Reactivity Measurements With Flow Cytometry
Yeast libraries were stained with the SARS-CoV-2 RBD of the ancestral (SPD-C52H3, AcroBiosystems), delta (SPD-C5226, AcroBiosystems), omicron BA.1 (SPD-C522j, AcroBiosystems), or omicron BA.2 (SPD-C522g, AcroBiosystems) strains as previously described for omicron-specific enrichment using the FACSMelody. To normalize for antigen differences in background and signal, the binding-population threshold was set to 5% on the negative control for each antigen before it was applied to each of the anti-CoV-2 libraries with said antigen. Flow cytometry data were analyzed using FlowJo (BD Biosciences).
Biolayer Interferometry (BLI) Kinetics, BLI Angiotensin-Converting Enzyme 2 Blocking Assay, and Pseudotype Neutralization
A detailed description of the methods related to the biolayer interferometry (BLI) antibody kinetics and affinity measurements, the BLI ACE2 and SARS-CoV-2 spike antibody blocking assay, and the pseudotype neutralization of WHO variants of concern can be found in the Supplemental Methods.
RESULTS
Recombinant Polyclonal Antibody Therapeutic Antisevere Acute Respiratory Syndrome Coronavirus 2 Immunoglobulin Hyperimmune Pseudoviral Neutralization Is Demonstrated Against All Variants
Using pseudoviral constructs representing currently available WHO variants of concern, neutralization potency was measured for rCIG and compared with a polyclonal hyperimmune standard from National Institute for Biologicals Standards and Control (NIBSC) code 20/130 and a reference mAb SAD-S35 (Figure 1, Supplementary Table 1). The rCIG neutralized all variants tested and had a median half-maximal inhibitory concentration (IC50) of 0.59 µg/mL (interquartile range [IQR], 0.37–1.09 µg/mL). The NIBSC standard also neutralized all variants tested and had a median IC50 of 19.6 µg/mL (IQR, 11.0–34.4 µg/mL). The mAb SAD-S35 was unable to neutralize beta, gamma, and omicron; for the variants that mAb SAD-S35 could neutralize, there was a median IC50 of 0.42 µg/mL (IQR, 0.30–0.57 µg/mL). Although rCIG could neutralize all tested variants, omicron BA.1 and BA.2 showed the greatest reduction of potency (97- and 34-fold reduced from the median, respectively). Although there is an overall reduction of potency, the observed neutralization result indicates that there is a subset of potently neutralizing clones that can be interrogated further.
![Pseudoviral neutralization is demonstrated against all variants for recombinant polyclonal severe acute respiratory syndrome coronavirus 2 (CoV-2) immunoglobulin (rCIG). Curves show the percentage of green fluorescent protein-positive (infected) angiotensin-converting enzyme 2-expressing cells versus concentration of anti-CoV-2 antibodies ([Abs] each Ab concentration was tested as a single replicate). The National Institute for Biologicals Standards and Control (NIBSC) standard was not tested against BA.1 or BA.2, and SAD-S35 was not tested against BA.2, due to insufficient quantities of antibody reagent. Frequency of infected cells was normalized to the no-antibody control for each variant.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jid/228/5/10.1093_infdis_jiad102/1/m_jiad102f1.jpeg?Expires=1747875501&Signature=RFg29kpGEN6L5dVL59svL9BD7XLFIsSjdCVLxzge2iXni5OKIqQSL5O2yGsZQJ~lqs27f-R27zXrcsMClRbQu9oUb0Q9b32rZmKo6ty1QNVCQliz4V5fFN8YiLnSrY09MU8Uiawi8emYdPaSJwD-uw24xqlc9ZoYPuEcFTFaTh2UdqbDgpWVkZn8dw78CRuGfXghKa-EM2bbQfc1bb9NIs6PpURPwBd1V3WAnLLmE7WNi5OZy5rE7AJtY0doLUCNokkd4clWhPPe8WkGm5HZYzuzv2HJdiTb54394iLiyoc5XB3IlW~tbkUGaIWHToDrY-FOraNZbua~EFJ6sVLy6Q__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Pseudoviral neutralization is demonstrated against all variants for recombinant polyclonal severe acute respiratory syndrome coronavirus 2 (CoV-2) immunoglobulin (rCIG). Curves show the percentage of green fluorescent protein-positive (infected) angiotensin-converting enzyme 2-expressing cells versus concentration of anti-CoV-2 antibodies ([Abs] each Ab concentration was tested as a single replicate). The National Institute for Biologicals Standards and Control (NIBSC) standard was not tested against BA.1 or BA.2, and SAD-S35 was not tested against BA.2, due to insufficient quantities of antibody reagent. Frequency of infected cells was normalized to the no-antibody control for each variant.
Kinetics Comparison of Severe Acute Respiratory Syndrome Coronavirus 2-Specific Antibodies
Affinity measurements of rCIG, rCIG-HY, and other clinically relevant antibodies were carried out to compare the potency of rCIG to other already proven treatments albeit ones that have since lost their efficacy against BA.2. When comparing the binding kinetics of rCIG-HY to the monoclonals bamlanivimab, casirivimab, and imdevimab against ancestral strain RBD, the affinity of both rCIG libraries and clinical mAbs were similar, suggesting that rCIG might have comparable potency (Figure 2A, Table 1). It should be emphasized that rCIG is polyclonal and as such the apparent affinity will be a weighted average of the thousands of sequences that it contains. Of note, the previously reported KD of these mAbs (provided in Table 1) were measured using a different platform (surface plasmon resonance) and different protocols, which may explain why these results are slightly different than those published by other sources.
Kinetics and angiotensin-converting enzyme 2 (ACE2) blocking of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-specific antibodies. (A) Biolayer interferometry (BLI) kinetic fit of association (5 minutes) and dissociation (10 minutes) signals for recombinant polyclonal SARS-CoV-2 immunoglobulin high yield (rCIG-HY), casirivimab, bamlanivimab, and imdevimab loaded onto antihuman Fc probes in response to a titration of ancestral receptor binding domain (RBD). A 1:1 kinetic model was fit to the data for on-rate, off-rate and KD calculations. (B) The BLI blocking assay measuring the percentage of receptor ligand binding when interrupted by anti-RBD antibody compared with a no-antibody condition. Reported percentages are the S1-His and human ACE2 (hACE2) binding signal lost in response-specific antibody concentrations. S1-His loading time and concentration were optimized for Max1/2 signal, and hACE2 binding was optimized for saturation and plateau in the no-antibody condition. conc, concentration.
Kinetics fit Model Binding Ancestral RBD
| Sample | Published KD, nM | K-Gator KD, nM | kon, 1/Ms | koff, 1/s | R2 |
|---|---|---|---|---|---|
| rCIG-HY | … | 0.60 | 2.4 × 105 | 1.5 × 10−4 | 0.99 |
| Bamlanivimab | 0.071 [28] 0.8737 [29] | 1.63 | 3.4 × 105 | 5.6 × 10−4 | 0.99 |
| Casirivimab | 0.046 [29] | 0.88 | 7.7 × 105 | 6.8 × 10−4 | 0.98 |
| Imdevimab | 0.047 [29] | 0.81 | 6.1 × 105 | 5.0 × 10−4 | 0.98 |
| Sample | Published KD, nM | K-Gator KD, nM | kon, 1/Ms | koff, 1/s | R2 |
|---|---|---|---|---|---|
| rCIG-HY | … | 0.60 | 2.4 × 105 | 1.5 × 10−4 | 0.99 |
| Bamlanivimab | 0.071 [28] 0.8737 [29] | 1.63 | 3.4 × 105 | 5.6 × 10−4 | 0.99 |
| Casirivimab | 0.046 [29] | 0.88 | 7.7 × 105 | 6.8 × 10−4 | 0.98 |
| Imdevimab | 0.047 [29] | 0.81 | 6.1 × 105 | 5.0 × 10−4 | 0.98 |
Abbreviations: HY, high yield; RBD, receptor binding domain; rCIG, recombinant polyclonal antibody therapeutic antisevere acute respiratory syndrome coronavirus 2 immunoglobulin hyperimmune.
Kinetics fit Model Binding Ancestral RBD
| Sample | Published KD, nM | K-Gator KD, nM | kon, 1/Ms | koff, 1/s | R2 |
|---|---|---|---|---|---|
| rCIG-HY | … | 0.60 | 2.4 × 105 | 1.5 × 10−4 | 0.99 |
| Bamlanivimab | 0.071 [28] 0.8737 [29] | 1.63 | 3.4 × 105 | 5.6 × 10−4 | 0.99 |
| Casirivimab | 0.046 [29] | 0.88 | 7.7 × 105 | 6.8 × 10−4 | 0.98 |
| Imdevimab | 0.047 [29] | 0.81 | 6.1 × 105 | 5.0 × 10−4 | 0.98 |
| Sample | Published KD, nM | K-Gator KD, nM | kon, 1/Ms | koff, 1/s | R2 |
|---|---|---|---|---|---|
| rCIG-HY | … | 0.60 | 2.4 × 105 | 1.5 × 10−4 | 0.99 |
| Bamlanivimab | 0.071 [28] 0.8737 [29] | 1.63 | 3.4 × 105 | 5.6 × 10−4 | 0.99 |
| Casirivimab | 0.046 [29] | 0.88 | 7.7 × 105 | 6.8 × 10−4 | 0.98 |
| Imdevimab | 0.047 [29] | 0.81 | 6.1 × 105 | 5.0 × 10−4 | 0.98 |
Abbreviations: HY, high yield; RBD, receptor binding domain; rCIG, recombinant polyclonal antibody therapeutic antisevere acute respiratory syndrome coronavirus 2 immunoglobulin hyperimmune.
Biolayer Interferometry Angiotensin-Converting Enzyme 2 and Severe Acute Respiratory Syndrome Coronavirus 2 Spike Blocking With Ancestral and an Omicron Subvariant
A BLI blocking assay was utilized to determine the ability of rCIG and the other antibodies to block the interaction of ancestral or omicron BA.1 spike (S1) variants to the human ACE2 (hACE2) receptor. The rCIG and rCIG-HY had the same level of activity in this blocking assay for the ancestral S1 CoV-2 antigen (Figure 2B, Supplementary Figure 2A). Each of the 3 mAbs outperformed rCIG and SAD-S35 for the blocking of the ancestral strain to hACE2. However, when the BA.1 omicron S1 variant was used, rCIG outperformed SAD-S35 and bamlanivimab antibodies, rCIG was equivalent to imdevimab, and rCIG was less potent than casirivimab.
Variant-Specific Antibody Enrichment Yields Antibodies Capable of Binding to New and Old Receptor Binding Domain Variants
We next analyzed the ability of scFv-expressed polyclonal libraries to bind different RBD variants using yeast surface display. The original rCIG scFv library was created by enriching 16 SARS-CoV-2 convalescent donor repertoires for their ability to bind to ancestral RBD using yeast display [27]. Consistent with the results above, rCIG-scFv was confirmed following antigen staining and flow cytometry to have high antiancestral (84.7%) and anti-delta (55.1%) RBD binding; however, rCIG-scFv showed a significant decrease in binding to omicron variants BA.1 and BA.2, with only 14.2% and 19.0% of scFv showing binding, respectively (Figure 3). Of note, due to noise in the assay, 100% separation of positive and negative events is typically not observed, and thus binding values of approximately 80%–90% are considered the maximum attainable when the negative control is gated at approximately 5% binding.
Variant-specific antibody binding to new and old variants as measured by flow cytometry. (A) Flow cytometry pseudo-color density dot plots of negative control, recombinant polyclonal severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) immunoglobulin (rCIG) parent, rCIG BA.2-enriched (+) and rCIG BA.2-depleted (−) antibody libraries stained with SARS-CoV-2 receptor binding domain (RBD) from ancestral, delta, BA.1, and BA.2 strains. Gates set on the x-axis based on the single-chain variable fragment (scFv) c-myc expression reporter tag to be above the background signal of the secondary detection antibody. The y-axis gating was set at approximately 5% of the expressing population in the negative control strain for each antigen stain and then applied to the other libraries to control for differences in antigen fluorescence and background. The percentages shown are based on the antibody-expressing populations. (B) Bar graph depicting the frequency of binding antibody sequences in each library stained ancestral, delta, BA.1, or BA.2. Binding frequencies are normalized, per antigen, by subtracting the background percentages obtained by staining a non-anti-SARS-CoV-2 antibody library. APC, antigen-presenting cell; FITC, fluorescein isothiocyanate.
Although a significant portion of the rCIG antibodies do not bind to omicron variants, there must be potent binders/neutralizers present based on the pseudotype neutralization and BLI blocking results presented above. Thus, we attempted to enrich for the omicron-specific antibody sequences present in rCIG by sorting for omicron BA.2 binding sequences using rCIG-scFv as the input library, creating rCIG-scFv BA.2(+), and in parallel depleting for BA.2 binding sequences, creating rCIG-scFv BA.2(−) (the sorting strategy is illustrated in Supplementary Figure 1). As expected, rCIG-scFv BA.2(+) had a much higher clonal frequency of binders than rCIG-scFv to both BA.1 (54.6%) and BA.2 (67.7%) (Figure 3). It is interesting to note that rCIG-scFv BA.2(+) maintained strong binding activity to the ancestral stain (91.0%) and showed improved binding to the delta strain (79.7%) compared with the original rCIG-scFv library. In contrast, rCIG-scFv BA.2(−) lost all ability to bind to BA.1 and BA.2 (<2% each) but maintained binding to the ancestral (80.3%) and delta (47.5%) variants at similar levels as rCIG-scFv. Given that the ancestral-based vaccine continues to provide some level of protection against omicron variants despite the numerous significant amino acid changes, and that the combination of changes in omicron substrains collectively provide improved fitness and maintenance of ACE2 binding due to epistatic interactions of all mutations [29], the ancestral and omicron neutralizing antibodies bind conserved epitopes and maintain neutralizing efficacy against an array of variants.
Angiotensin-Converting Enzyme 2 Blocking and Pseudoviral Neutralization With BA.2 Enriched Antibody Libraries
To determine whether the enrichment of rCIG for omicron BA.2 would yield a recombinant polyclonal library with improved ACE2 blocking and neutralization of the BA.2 variant, the rCIG-scFv BA.2(+) and rCIG-scFv BA.2(−) libraries were expressed as full-length polyclonal libraries using the high-yield CHO cell line system rCIG-HY BA.2(+/−). The rCIG BA.2-HY(+/−) libraries were analyzed with the BLI blocking assay with the BA.2 omicron subvariant along with the previous antibody panel. The rCIG-HY BA.2(+) outperformed all other rCIG libraries as well as 3 of the 4 monoclonal antibodies (Figure 4A, Supplementary Figure 2B).
Angiotensin-converting enzyme 2 (ACE2) blocking and pseudoviral neutralization of BA.2-enriched libraries. (A) Biolayer interferometry blocking assay measuring the percentage of receptor ligand binding when interrupted by anti-receptor binding domain (RBD) antibody (Ab) compared with a no-antibody condition. Reported percentages are the S1-His and human ACE2 (hACE2) binding signal lost in response-specific antibody concentrations. S1-His loading time and concentration were optimized for Max1/2 signal, and hACE2 binding was optimized for saturation and plateau in the no-antibody condition. (B) Curves show the percentage of GFP-positive (infected) ACE2-expressing cells versus concentration of anti-coronavirus-2 antibodies (all samples were tested in duplicate, with mean ± standard deviation shown). Frequency of infected cells was normalized to the no-antibody control for each variant. conc, concentration; rCIG-HY, severe acute respiratory syndrome coronavirus 2 immunoglobulin high yield.
Pseudotype neutralization assays confirmed that full-length rCIG-HY BA.2(+) showed an improved ability to neutralize omicron BA.2 versus rCIG-HY, while still maintaining neutralization of the ancestral strain and omicron BA.1 (Figure 4B, Table 2). Consistent with the yeast binding results, rCIG-HY BA.2(−) was unable to fully neutralize BA.2 and was less potent against omicron BA.1, but it was still highly potent against the ancestral antigen. These data affirm the consensus that omicron strains have significant changes in important RBD epitopes leading to the reduced activity of potent mAbs against earlier variants.
Pseudoviral Neutralization Results of rCIG Libraries
| Recombinant Polyclonal Library | IC50 (µg/mL) | ||
|---|---|---|---|
| Ancestral RBD | Omicron BA.1 RBD | Omicron BA.2 RBD | |
| rCIG-HY | 0.535 | 40.9 | 32.5 |
| rCIG-HY BA.2(+) | 2.55 | 31.6 | 5.32 |
| rCIG-HY BA.2(−) | 0.208 | No neutralization | No neutralization |
| Recombinant Polyclonal Library | IC50 (µg/mL) | ||
|---|---|---|---|
| Ancestral RBD | Omicron BA.1 RBD | Omicron BA.2 RBD | |
| rCIG-HY | 0.535 | 40.9 | 32.5 |
| rCIG-HY BA.2(+) | 2.55 | 31.6 | 5.32 |
| rCIG-HY BA.2(−) | 0.208 | No neutralization | No neutralization |
Abbreviations: HY, high yield; RBD, receptor binding domain; rCIG, recombinant polyclonal antibody therapeutic anti-severe acute respiratory syndrome coronavirus 2 immunoglobulin hyperimmune.
Pseudoviral Neutralization Results of rCIG Libraries
| Recombinant Polyclonal Library | IC50 (µg/mL) | ||
|---|---|---|---|
| Ancestral RBD | Omicron BA.1 RBD | Omicron BA.2 RBD | |
| rCIG-HY | 0.535 | 40.9 | 32.5 |
| rCIG-HY BA.2(+) | 2.55 | 31.6 | 5.32 |
| rCIG-HY BA.2(−) | 0.208 | No neutralization | No neutralization |
| Recombinant Polyclonal Library | IC50 (µg/mL) | ||
|---|---|---|---|
| Ancestral RBD | Omicron BA.1 RBD | Omicron BA.2 RBD | |
| rCIG-HY | 0.535 | 40.9 | 32.5 |
| rCIG-HY BA.2(+) | 2.55 | 31.6 | 5.32 |
| rCIG-HY BA.2(−) | 0.208 | No neutralization | No neutralization |
Abbreviations: HY, high yield; RBD, receptor binding domain; rCIG, recombinant polyclonal antibody therapeutic anti-severe acute respiratory syndrome coronavirus 2 immunoglobulin hyperimmune.
Overall, these data demonstrate the ability of a recombinant polyclonal hyperimmune to be comparatively robust to losing blocking activity, particularly when cold and hot spot epitopes are considered in the final enrichment strategy. Although only a few amino acid changes can completely abrogate a monoclonal antibody cocktail, a polyclonal may have favorable resistance to viral immune escape over time by targeting many epitopes simultaneously.
DISCUSSION
As SARS-CoV-2 evolves in humans to gain greater fitness through adaptations that lead to enhanced transmissibility and immune escape, so too must our drug strategies continue to evolve. At a minimum, what is required of a neutralizing antibody drug to protect against a constantly changing biological target is a resistance to loss of potency. One straightforward way to reduce the rate of disease burden is to maximize the vaccination rate of the population. This will not prevent the spread of SARS-CoV-2 through infection, but it will provide protection against the disease length and severity. Although this strategy will work for most individuals, there will always be a population of patients that are at elevated risk for severe disease due to pre-existing conditions and/or a limited immune memory response from vaccination. For these patients, protective antibody therapeutics can be a life-changing and life-extending prophylactic in addition to being a critical postinfection treatment modality.
Recently, all vaccine boosters in the United States have become an ancestral and omicron BA.1 bivalent (mRNA-1273.214). However, clinical trials that compare the original ancestral vaccine mRNA-1273 and mRNA-1273.214 show only a meager statistical significance in the bivalent third booster versus an ancestral third boost, which has little clinical relevance [30]. Previously, studies have described the phenomenon of “original antigen sin”, demonstrating that immune responses primed by early strains later lead memory B cells to recognize shared epitopes of future strains boosted by subsequent infection or vaccination [31]. The rCIG gives strong support for how vaccination and boosting with the ancestral strain alone can still be protective against all known variants of SARS-CoV-2 because it contains many neutralizing clones to delta, omicron BA.1, and omicron BA.2. Our yeast display binding data show that within the antibody repertoire of ancestral strain infected donors collected in April 2020, a proportion of these clones can neutralize the omicron BA.2 strain. These antibodies that provide a more durable neutralizing activity appear to bind conserved epitopes that are less likely to change, possibly because these regions are necessary for the viral lifecycle. Although we have focused here on enrichment of anti-RBD antibodies, the complete antibody repertoires of infected individuals were captured in the parent libraries. A future option to improve rCIG's breadth of targets might involve selection of antibodies to other highly conserved protective epitopes, such as those found in the NTD [23, 24]. Now that these other therapeutic targets have been validated, enrichment and inclusion of these sequences would likely further expand the robustness of cross-variant protection. In future studies, we would also want to confirm the neutralizing ability of these libraries using live SARS-CoV-2 virus, which were previously performed with the ancestral strain, and found to correlate well with the pseudoviral neutralization assay [27].
Novel variants by their very nature represent the unknown and as such, any strategy to plan for them will have inherent risk. This is why we believe it is necessary to advocate for a strategy that consists of a thousands-diverse SARS-CoV-2-specific antibody library like rCIG with a refined selection for antibodies that bind potentially conserved epitopes. Because SARS-CoV-2 will likely never be eradicated and will continue to evolve indefinitely, rCIG not only demonstrates maintained neutralizing potential, but it also remains a source to discover antibody sequences that will bind emerging variants and can be used to create polyclonal antibody libraries against future strains. The Surge antibody discovery platform reduces the reliance on traditional drug discovery pipelines to repeat the lengthy and costly steps of searching for antigen-specific cells, 1 clone at a time. This platform could also be used to inform vaccine development strategies by demonstrating that neutralizing antibodies are present and ready to target new viral variants after vaccination. Finally, using this platform to iteratively enrich a foundationally diverse repertoire like rCIG against divergent variants such as omicron BA.2, the development of a broadly neutralizing polyclonal antibody therapeutic can be quickly achieved.
CONCLUSIONS
The rapidly changing SARS-CoV-2 has led to inevitable development of both drug resistance and an increase in public awareness of the need for a new class of therapeutic, like the one we describe. A clear limitation of our study is that the strains contained herein are no longer extant because they have been replaced by new strains that have emerged even as this body of work was being carried out. Future strains, and their evolved resistance, are themselves evidence that a new class of robust and broadly neutralizing therapeutics derived using the methods presented here targeting mutational cold spots through iterative variant enrichment is needed [32, 33].
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Acknowledgments. We thank the staff at MedPharmics (Metairie, LA) and Access Biologicals (Vista, CA) who were instrumental in acquiring convalescent coronavirus disease 2019 (COVID-19) donor samples under institutional review board-approved protocols. We thank Jennifer Keller (GigaGen, Inc., San Carlos, CA) who helped to organize collection of COVID-19 samples and Carter Keller (GigaGen, Inc., San Carlos, CA) who provided advice and support. We thank the staff at Integral Molecular for their diligent work in creating new pseudoviruses for every new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant. We thank Dr. Sharon Schendel and CoVIC (https://covic.lji.org/) for collecting, coordinating, and providing the information about the landscape of recombinant antibodies targeting SARS-CoV-2. GigaGen paid for the Open Access publication charges for this article.
Financial support. All work was completed with funding from GigaGen and Grifols.
References
Author notes
N. P. W. and A. R. N. contributed equally to this work.
Potential conflicts of interest. All coauthors are past or current employees of GigaGen, receive salary, and may have stock ownership. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
