With very few exceptions, antibody therapeutics approved to date are isotype G immunoglobulins (IgG). IgGs are protein macromolecules secreted in the blood of most vertebrates2 by differentiated plasma B cells that have a high affinity and specificity for their respective antigen. The IgG molecules can then be purified from human or animal plasma to produce polyclonal immune globulin products. These types of products, such as diphtheria antitoxin3, represent some of the first products to be licensed in the United States. In over a century of development, polyclonal products have undergone tremendous advances in the manufacturing process and characterization of safety and efficacy attributes. In the last few decades, antibody therapeutic development has shifted toward the development of monoclonal antibodies - IgG molecules that are produced in vitro after mature B cells have been isolated, immortalized, and cultured.
On the other hand, antibody-dependent enhancement (ADE) of viral infection or disease can also occur, as has been documented in humans for dengue virus10. ADE can arise after natural infection, vaccination, or passive transfer of antibody therapies. It is widely thought that ADE occurs when antibodies of insufficient avidity or concentration are unable to neutralize the virus, but can facilitate the uptake of the virus-antibody complex by FcγR-bearing cells such as monocytes, dendritic cells, or macrophages11, resulting in increased viral production, enhanced immune activation (e.g., cytokine production), and more severe disease12. In addition to flaviviruses13, 14, ADE has been observed for mAbs against influenza virus, HIV-1, and EBOV in cell culture, but not typically when tested in animal models or in clinical trials, with a few exceptions11, 13. The risk of ADE can be reduced by engineering substitutions into the Fc region that disrupt FcγR binding, although these substitutions may also disrupt Fc effector functions that could contribute to clinical efficacy15, 16. Thus, when selecting antibodies best suited for use as an antiviral product, it is critical to optimize binding both to the antigen and FcγRs. For mAbs, IgG isotypes can be selected, Fc glycosylation patterns can be modified, or Fc regions can be engineered with substitutions that enhance or diminish select Fc effector functions. Although ADE in cell culture and animal studies has been observed with antiviral specific polyclonal immune globulins (IG)17, clinical ADE has not been reported for any FDA-approved specific IG products.
During pharmaceutical development, mAb domains often undergo extensive biochemical engineering to optimize the properties of the antibody. For example, the CDRs can be grafted onto the framework regions of V domains from other mAbs and still retain their antigen binding properties in the context of a known protein fold. The Fc region can also be modified to alter pharmacokinetic properties and effector functions. On the other hand, although not subjected to Fc engineering, depending on the antigen or donor population, specific antiviral polyclonal IGs can be “enriched” for a particular isotype18, subclass, or glycosylation signature, leading to different Fc effector functions compared to other polyclonal IG products. For example, IgG1 and IgG4 are the most prevalent subclasses following measles infection or vaccination, with significant differences in titers in infected versus vaccinated individuals19. In addition, anti-SARS-CoV-2 antibodies from convalescent donors can have distinct glycosylation patterns depending on disease severity20. We will discuss some of the methods currently used to design, produce, and characterize antibody products, highlighting the differences between polyclonal and monoclonal antibody therapies.
Production and Characterization of Antibody Therapies
Specific Polyclonal Antibody Therapies
Specific polyclonal immune globulins (SpIG) are purified from pooled animal or human plasma. The first products were developed in 1898 and were comprised of little more than serum from horses vaccinated with live viruses, bacterial toxins, or snake venom. In 1903, diphtheria antitoxin made from vaccinated horses became the first licensed product in the United States. Research during World War II stimulated a major breakthrough in purification of IGs and other proteins from human plasma. IG purification methods are based on sequential alcohol precipitations, each with specific conditions of pH, ionic strength, temperature, protein concentration, and alcohol concentration21, 22.
For some products, purely chromatographic methods or caprylate precipitation methods have partially or completely supplanted alcohol precipitation. These changes are often driven by the need to increase yield of IgG (thus increasing availability)23. Nevertheless, alcohol-based fractionation remains as the backbone of early steps in production of IG products and is often combined with subsequent caprylate or polyethylene glycol precipitations. Modern IG products are further purified using column chromatography to remove unwanted plasma proteins. In addition, a minimum of two orthogonal, robust, dedicated viral clearance steps are performed, which often include solvent-detergent treatment and nanofiltration, as well as other virucidal (caprylate, heat treatment, low pH) and partitioning (chromatography, precipitations, depth filtration) steps. All viral clearance steps must be validated and found to be robust using scaled-down models of the manufacturing process and actual manufacturing intermediates spiked with virus as starting material. It should be emphasized that modern IG purification is highly complex with multiple steps, each of which must be controlled to result in a safe and intact product. Every manufacturing method is unique with respect to purification details and methodology (such as mixing speeds, equipment used, precipitation times, buffer types and concentrations, centrifugation vs. precipitation), and equipment. Thus, each product is also unique with respect to levels and types of plasma protein impurities and IG stability.
Specific polyclonal IG is used as the overarching term for all polyclonal preparations that are enriched for certain antiviral, antibacterial, or antitoxin antibodies. Antibody enrichment for human antibodies is achieved by either immunizing donors, or screening and selecting high-titer plasma from routine donations (as for Cytogam24) or convalescent donors (as for early versions of SARS-CoV-2 IG investigational products25, 26). “Hyperimmune” polyclonal antibodies are derived from animal or human donors who have been immunized intentionally for the purpose of obtaining high titer plasma (e.g., rabies, anthrax, and tetanus immune globulins). Nevertheless, convalescent plasma is often inaccurately referred to as “hyperimmune,” even though donors were not immunized. Under FDA-approved plasma center collection protocols, and after investigational safety studies are completed, hyperimmune plasma can be collected from consenting immunized donors.
For purposes of final product testing, a validated bioassay demonstrating neutralization in cell culture or in animals, is ideally performed for SpIG products. In special cases, adequate cell culture or animal models are not available at the time of licensure. In this situation, a binding assay has usually been selected and validated for product release contingent on discussions with FDA. Likewise, national or international IgG standards may be lacking. In these instances, an internal IgG standard is developed by the manufacturer.
Antiviral SpIG products licensed in the United States are shown in
Table 1. A number of SpIGs (human, bovine, or equine plasma-derived) are under investigation for other viral infections, including SARS-CoV-2 (at least 12 studies in clinicaltrials.gov, e.g., including an International Network for Strategic Initiatives in Global HIV Trials study, NCT04910269) and influenza (NCT04850898). SpIGs from animal sources are produced by hyperimmunizing donor animals. Advantages of large animal donors (horses, sheep, or cattle) include the ability to immunize more frequently (which increases the yield and avidity of specific antibodies), to use experimental vaccinations, and to safely collect larger volumes of plasma. A major disadvantage includes potential allergic reactions in patients due to animal proteins, including the active ingredient. Animal-derived antibodies are often treated with pepsin or trypsin, to remove the Fc portion and reduce immunogenicity. These fragments lack effector functions that could be important for antibody activity, depending upon the virus. An interesting strategy has been developed using transchromosomic cattle that produce full-length human IgG antibodies. The cattle are knocked out for bovine antibody heavy and lambda light chains but contain an artificial chromosome encoding the respective human IgG chains. Chimeric antibodies consisting of human IgG heavy chains and bovine kappa light chains are removed during manufacturing
42, thus the resulting IG product manufactured from these bovines contain only human IgGs, thus lowering the risk of immunogenicity. These transchromosomic bovines have been successfully hyperimmunized
43.
Treatment timing and dosing for SpIG.
Treatment timing relative to infection depends on demonstrable efficacy of the product for pre- or post-exposure prophylaxis. Pre- and post-exposure prophylaxis can be effective (if adequately dosed) largely because viral burdens are relatively low. Even if an infection has been initiated, post-exposure prophylaxis attenuates disease severity of measles, HAV, and varicella zoster30, 36. When vaccines are given concomitantly with specific IG, such as for rabies, passive immunization provides a defensive “bridge” that acts immediately to neutralize the virus until vaccine responses arise. It is important that the dose of rabies IG (RIG) is not so high that it suppresses the vaccine response. In such contexts, both a minimum and maximum potency should be defined to assure optimal function of both RIG and the vaccine. Pharmacokinetic studies performed in healthy immunocompetent human subjects are used to define the dose of SpIG that is needed to avoid suppression of vaccine responses yet still be able to provide protection until vaccine responses are sufficiently developed.
Treatment of symptomatic viral disease with SpIG is much more challenging and often ineffective. In these cases, the viral burden may exceed the capacity of the IG, viruses may be relatively inaccessible within infected cells, and cellular immune responses may also be suppressed by the virus44. Notable lack of efficacy by specific IG for treatment of symptomatic infections such as rabies, influenza, HAV, HBV, measles, and varicella have been observed. The time windows for effective post-exposure prophylaxis of each infection have been established based on such failures. Treatment with CMVIG and HBVIG(IV) can prevent symptomatic disease in transplanted patients but are not curative. Vaccinia Immune Globulin is used to treat severe complications (eczema vaccinatum and progressive vaccinia) caused by live vaccinia virus vaccine (ACAM2000), which is used to prevent smallpox. Recently licensed replication-deficient vaccinia virus (Jynneos) generates an immune response but is thought to be incapable of causing eczema vaccinatum or progressive vaccinia. Both vaccines are indicated for prevention of smallpox. Jynneos is also licensed for prevention of monkeypox45.
Monoclonal Antibodies
To date, the FDA has approved four mAb therapies to prevent or treat viral diseases (
Table 2): palivizumab for prevention of RSV in preterm infants and infants with other specific conditions, ibalizumab for treatment of HIV-1 in patients failing their current anti-retroviral regimen, and two products for treatment of Ebola virus disease caused by
Zaire ebolavirus. One of these products, Inmazeb, consists of three mAbs that target non-overlapping epitopes on EBOV glycoprotein and represents the first co-formulated mAb cocktail approved by the FDA
46.
Multiple mAbs are currently in advanced stages of clinical development or have been approved in other countries. Nirsevimab, a half-life extended mAb that targets the RSV fusion (F) protein 47, was recently approved by the European Medicines Agency for the prevention of RSV lower respiratory tract disease in neonates and infants during their first RSV season. In addition, three mAb products targeting the rabies virus glycoprotein have been approved in other countries: two in India (Rabishield, a single mAb, and TwinRab, a cocktail of two mAbs48), and one in China (ormutivimab49 ).
Several mAbs and mAb combinations that target the SARS-CoV-2 spike protein were rapidly developed after the onset of the COVID-19 pandemic and received emergency use authorization (EUA) from the FDA for the pre-exposure prophylaxis, post-exposure prophylaxis, and/or treatment of COVID-19. Although highly effective against early SARS-CoV-2 variants, these products are not currently authorized in the United States due to the emergence and widespread circulation of variants that are resistant to neutralization by these mAbs in cell culture50-56. However, if future variants emerge that are susceptible to these products, their authorization status may change. Refer to the FDA website for updated information on the status of EUAs for mAbs and other COVID-19 therapeutics57.
In addition to the approved and previously authorized mAbs and those directed against SARS-CoV-2, many other mAbs have been or are under development against existing and emerging diseases (see 58-61 for some examples).
Historically, therapeutic mAbs were derived from immunized mice or rats and engineered as chimeric or humanized mAbs to reduce the immunogenicity due the “foreignness” of rodent mAbs in humans. Currently, most mAbs are of human origin, derived from “humanized mice” that express human germline V(D)J region genes, or from phage display libraries generated from human donor lymphocytes. However, many antiviral mAbs are isolated directly from previously infected patients (see58, 62-64). Regardless of the source, many considerations inform the selection and engineering of candidate mAbs.
Fc engineering approaches: Most mAbs developed for viral diseases are selected first for their ability to neutralize virus entry. However, Fc effector functions play a major role in the immune system’s response to infectious diseases8. For mAbs, the contribution of Fc effector functions to disease protection has been demonstrated for several viruses including Ebola virus65, HIV-16,66, 67, influenza68, SARS-COV-269, and Rift Valley fever virus70. However, ADE of infection or disease is a possible negative consequence of FcγR binding71-73. Therefore, depending on what is known about specific viral diseases, different approaches are used to engineer the Fc region of mAbs to either enhance or diminish FcγR binding. Amino acid residues have been identified in the IgG Fc region that contact the complement component C1q, FcγRs, or the neonatal Fc receptor (FcRn), which is responsible for the long half-life of IgG (reviewed in74, 75). Substitutions can be engineered at these residues to alter Fc effector functions or extend the half-life of a mAb, which allows less frequent dosing76.
In addition to Fc engineering, there is a better understanding of specific Fc glycan structures and their association with different effector functions, e.g., afucosylated mAbs have better antibody dependent cellular cytotoxicity (ADCC) compared to highly fucosylated antibodies, and galactosylation is associated with complement dependent cytotoxicity (CDC) and can influence ADCC activity77 . Therefore, cell lines have been engineered to produce mAbs with up to 100% afucosylation to enhance ADCC activity74, 78. The understanding of the relationship between antibody glycan structures and Fc effector functions is ongoing and additional strategies may be developed to further engineer mAb glycan structures. For example, the effect of galactosylation on ADCC activity may depend on the specific linkage of the galactose monosaccharide79. Fc effector functions can be reduced by introducing substitutions at the glycosylation site (N297) in the CH2 domain to prevent the addition of a glycan80, 81, thus providing another glycoengineering approach for antiviral mAbs.
Other approaches for the development of mAbs: Three of the four approved monoclonal antiviral products are single mAbs, but the anti-Ebola virus mAb cocktail of atoltivimab, maftivimab, odesivimab-ebgn was the first fixed dose co-formulated mAb combination product approved by the FDA. Many other mAbs to treat viral diseases and for other indications are used in combination, but only a few to date are co-formulated82. The advantage of antibody cocktails over a single mAb is that they might be less susceptible to escape, depending on the different targeted epitopes. As seen for the anti-SARS-CoV-2 mAb combinations previously authorized for the prophylaxis or treatment of COVID-19, they all target the SARS-CoV-2 receptor binding domain of the SARS-CoV-2 spike protein but have little neutralization activity against current variants. MAbs that target regions outside the receptor binding domain could neutralize virus or mediate Fc effector functions and might be less susceptible to escape. For example, a recent report demonstrated that mAbs targeting the conserved fusion peptide region adjacent to the S2′ cleavage site of the spike protein are broadly neutralizing against betacoronaviruses83.
Evaluation of Antiviral Activity
Prior to being evaluated in clinical studies as antiviral therapies, biological activity and potential mechanisms of action of antibodies are investigated in preclinical studies performed in model systems. In practice, assays to assess antibody activity usually fall into three broad categories: biochemical (e.g., binding) assays, cell culture assays, and animal models. Early during the pharmaceutical development of the antibody therapies, these assays are performed as part of candidate selection and then to characterize the antibody product that is being developed. Multiple such assays can be performed with the goal of understanding different aspects of antibody antiviral activity. Examples include antibody binding affinity, epitope characterization, neutralization activity, and assays to characterize Fc effector functions. Some of these assays will be developed as quality control potency assays to ensure lot-to-lot consistency and stability of the product. Federal regulations define potency as “the specific ability or capacity of the product, as indicated by appropriate laboratory tests or by adequately controlled clinical data… to effect a given result.” (21 CFR. 600.3(s)). Thus, for antiviral antibody therapies, potency assays provide a quantitative measure of the antibody activity linked to its primary mechanism of action. Fit-for-purpose potency assays are often performed prior to initiation of Phase 1 clinical studies, and full validation is completed by the time of a biologics license application (BLA) submission. FDA guidance describing current thinking on the development and validation of such studies for mAbs is available in draft form at the time of writing this article85. Some points to consider when designing preclinical studies to evaluate antiviral activity and assess potency are discussed below.
Types of Potency Assays
SpIGs and mAbs may exert their antiviral effects via one or more potential mechanisms: virus neutralization, ADCC, opsonization and phagocytosis, complement lysis and/or complement dependent cytotoxicity 86-88. There are just a few examples of antibodies potentially acting at virus post-entry steps87, 88 but their role in the overall antiviral humoral immune response is still to be established. If SpIGs or mAbs under clinical development have multiple mechanisms of action, multiple assays are developed and implemented for quality control. In general, the selected potency methods should reflect the product’s proposed mechanism as closely as possible. Potency is usually evaluated by a comparison to an appropriately qualified reference standard and is expressed as a percentage of the reference material value. For SpIGs, international or national standards are often used, e.g., for anti-rabies, anti-hepatitis B virus, or anti-measles IGs. The potency is then expressed in international units or alternative units, as appropriate. As for all quality control release methods, key assays for demonstrating the antiviral mechanism(s) of action should be shown to be suitable for their intended purposes during development and validated by the time of an application for approval. Ideally, potency assays which adequately reflect the proposed mechanism(s) of action should be qualified and implemented before pivotal clinical trials85.
Antibody-antigen binding is a necessary step for both virus neutralization and Fc effector functions. Therefore, binding assays such as an enzyme-linked immunosorbent assay (ELISA) or a surface plasmon resonance (SPR) assay are a logical approach to evaluate drug potency. There is significant experience with these types of assays, and they may be easier to qualify and validate compared to cell-based methods used to assess the antiviral activity of therapeutic antibodies. In general, potency binding assays are developed and used during early stages of product development. However, direct binding assays may not provide a comprehensive assessment of the product’s mechanism of action. For antibodies targeting virus-cellular receptor(s) interactions, inhibitory binding assays (ELISA or SPR) may better reflect their mechanism of action, but even these assays may not fully represent the antibody-mediated suppression of the complex virus-cell fusion process. Furthermore, broadly neutralizing antibodies may target complex, conformation-dependent, non-linear epitopes which can be challenging to reproduce in a binding assay.
In comparison to binding assays, cell-based methods can provide a more comprehensive assessment of antibody-mediated antiviral activity, either via virus neutralization and/or Fc effector functions85.
There is already significant expertise with the development and validation of cell-based ADCC potency assays for a variety of mAbs for the treatment of different neoplasms89 and the qualification/validation of methods to evaluate the ADCC activity of antiviral antibodies follow the same general principles. However, challenges remain regarding the selection and qualification of relevant target and effector cells employed in these assays (discussed later in the manuscript).
Virus neutralization assays can employ authentic (wt) viruses, replication-competent pseudotyped virions, or cell-fusion capable, but replication- incompetent pseudotyped virus-like particles (VLPs). Pseudotyped viruses and VLPs are considered safer alternative methods for studying a growing number of viruses which pose enormous health and socioeconomic risks because of their high pathogenicity (including Ebola, Sudan, Marburg, Hendra and Nipah viruses, SARS, and MERS) or their capacity to cause a widespread pandemic (HIV-1, SARS-CoV-2, certain influenza virus A subtypes). Furthermore, highly pathogenic viruses require biosafety level-3 (BSL-3) or BSL-4 facilities, which have high costs and limited availability, impeding the successful development of new therapeutic modalities.
Pseudotyped viruses, also referred to as chimeric viruses90, are typically generated by replacing the gene(s) expressing the surface glycoprotein(s) of a virus with low pathogenicity (e.g., vesicular stomatitis virus) with the gene(s) encoding the envelope (Env) glycoprotein(s) of a BSL-3 or a BSL-4 pathogen (e.g., EBOV), thus creating a replication-competent virus that can be used in a BSL-2 environment90-95. However, in the case of HIV-1, replication competent, infectious pseudotyped viruses were created within the same species by replacing the original envelope gene with the one from a different HIV-1 strain, usually for the purpose of studying virus tropism and/or neutralization susceptibility96-101. In general, in vitro infection with either wt or pseudotyped replication competent viruses involves a self-spreading infection among the target cells unless the time of the assay is shortened by design. Reporter genes are often inserted into the genome of pseudotyped viruses to assess the level of infection90, 94, 95, 101-103 as an alternative to measuring viral proteins, nucleic acids, or cytopathic effects91, 98, 100, 104. Target cells, stably transfected to express a reporter gene(s) when infected can also be used to quantify virus infection and assess the activity of antivirals, including neutralizing antibodies101, 105-109.
In addition to pseudotyped viruses, fusion-competent, but replication-incompetent VLPs can be used to assess virus fusion and entry. VLPs are produced by co-transfecting producer cells (usually 293T cells) with a plasmid(s) encoding the desired virus surface glycoprotein(s) and a plasmid (or plasmids(s)) encoding viral proteins necessary for VLP production. In general, VLP contain either an incomplete or no viral genome which renders them capable of just a single round of virus entry followed by a partial or no virus replication. Currently, VLPs have been successfully generated for both enveloped and non-enveloped viruses 102, 110-120. For the generation of enveloped VLP, retroviral (HIV-1 or murine leukemia virus derived) or rhabdoviral (VSV) based packaging vector systems are commonly used110, 119, although other vectors are also described101, 108, 121. As with the pseudotyped viruses, to facilitate the assessment of VLP cell fusion and entry, VLP are often engineered to include a reporter gene encoding an enzyme or a fluorescent protein (luciferase, alkaline phosphatase, β-galactosidase, green fluorescent protein), where expression reflects the level of infection97, 99, 102, 103, 113, 114, 122.
Alternatively, VLPs can be used to infect stably transfected cell lines containing a reporter gene under the control of a viral regulatory protein123-125. The TZM-bl cell line, stably transfected with the luciferase and β-galactosidase genes under the control of the HIV-1 long terminal repeat promoter, activated by the HIV-1 tat protein, is probably the best-known example of this approach95, 106, 109, 124, 126.
A more elaborate VLP system, based on EBOV minigenomes that encode a reporter gene, has been designed to study almost all aspects of the EBOV life cycle in a BSL-2 environment 112, 127. This system may potentially be used for the development and/or screening of anti-EBOV antibodies, but its applicability for this purpose remains to be demonstrated.
It should be noted that for reporter gene encoding virus particles (wt/pseudotyped viruses or VLPs), the reporter gene expression depends not only on the virion-cell fusion, but also on post-entry events leading to the synthesis of the encoded protein. To study solely the viral cell entry process, replication competent virions or VLPs have been designed to incorporate an enzyme or a fluorescent protein that is expressed in producer cells. This is achieved by utilizing vectors that encode chimeric molecules, consisting of the “reporter protein” fused to a viral protein, which directs the entire molecule into the budding virions112, 128-133. The assays that employ “reporter protein” containing virions can be a valuable tool to study virus entry inhibitors, but currently there are limited data, compared to the reporter gene-based methods, regarding their use for the assessment of virus neutralizing antibodies132.
A broad range of enveloped viruses belonging to different families and including human pathogens, can induce cell-cell fusion between infected cells and neighboring non-infected cells134. This phenomenon served as the basis for the development of assays measuring the level of fusion between virus surface glycoprotein-expressing cells (effector cells) and cells expressing the relevant virus receptor(s) (target cells) via quantitation of giant, multinuclear cells (syncytia) formation, fluorescent dye transfer, or reporter gene expression. The virus envelope-mediated cell-cell fusion assays are a useful tool for assessing virus-cell fusion inhibitors, including neutralizing antibodies, as a rapid surrogate for the virus entry methods. Also, like the pseudotyped and VLP systems, cell-cell fusion assays allow studying in BSL-2 environment of viruses that are otherwise restricted to a higher level of biocontainment (BSL-3 or BSL-4)107, 135-138.
The variety of methods which can be used to evaluate the effects of neutralizing antibodies raise the issue of how these methods compare to each other regarding their sensitivity and ability to predict a correlation between the in vitro and in vivo results. To address this issue, efforts have been made to apply a standardized approach among different labs for the assessment of antibody-mediated virus neutralization124, 139, 140.
In general, multiple in vitro binding and cell-based methods are used for antibody characterization during product development, with the goal of defining the antibody’s mechanism(s) of action and the critical quality attributes potentially affecting its anti-viral activities, as both the neutralization and Fc -mediated effects of the antibody are being investigated. Assays employing wt infectious viruses are likely to remain an important part of product characterization and serve as a basis for comparison with the alternative virus neutralization methods. However, as mentioned earlier, biosafety concerns may limit the use of authentic BSL3-3 and BSL4-4 pathogens. The use of pseudotyped viruses and/or VLP can offer less restrictive biosafety requirements and may facilitate antibody characterization in several additional ways: a. pseudotyped viruses/VLP can be more readily manipulated allowing faster assessment of potential mutations in the virus surface glycoproteins b. the level of entry may be easier to quantify, and c. panels of pseudotyped VLP, representing a wide range of virus strains that are generated using the same packaging system, can be created and tested for their neutralization susceptibility126, 141-143. Ideally, the pseudotyped viruses or VLP should closely resemble the corresponding authentic viruses, but certain differences may exist regarding shape, glycosylation, and density of the envelope glycoproteins due to the packaging system and/or producer cells used. The incubation times, readout methods and target cells may also be different140, 144. However, similar neutralization sensitivity was demonstrated when replication competent HIV-1 and HIV-1 Env pseudotyped VLP (generated using an HIV-1 derived packaging system) were both produced in 293T cell line and tested on the same target cells99, 106. Also, similar neutralization profiles were observed for spike protein pseudotyped VLP and authentic SARS-CoV-2 in Vero E6 target cells90.
Usually, one or more of the characterization methods are adapted to become a potency assay(s) for the purpose of drug substance and drug product release and stability testing. Development and implementation of an adequate potency assay(s) is a critical quality control measure to ensure that each lot is consistently produced with the potency necessary to achieve clinical efficacy and that such potency is maintained over the shelf life of the product. Data regarding the validation of certain commonly used pseudotyped virus/VLP assays have already been published101, 109. It should be emphasized that adequate qualification of the assays’ critical reagents is an integral part of the validation process. Detailed information regarding the generation, quantitation, and stability of the virus/VLP stocks should be provided. Determination of the ratio of functional (infectious, or fusion-capable) to non-functional virus/virus-like particles may also be important for qualification of the virus stocks 140. Finally, control measures should be in place to ensure consistency in the performance of the target cells in the virus neutralization assays or both the target and effector cells in the virus envelope mediated cell-cell fusion assays 85.
Cells for Potency Assays
For most viruses, there are a wide variety of cell lines that can be used to assess the potency of polyclonal and monoclonal antibodies with antiviral activity. For example, for SARS-CoV-2 alone, cell lines that have been used to assess antibody neutralization activity include Vero/Vero E6 (African green monkey kidney), 293/293T (human embryonic kidney), HeLa (human cervical carcinoma), Huh7 (human hepatocellular carcinoma), Calu-3 (human lung carcinoma), HT1080 (human fibrosarcoma), U2OS (human osteosarcoma), and HOS (human osteosarcoma) cells, often engineered to stably express human angiotensin-converting enzyme 2 (ACE2, the SARS-CoV-2 receptor) and/or transmembrane serine protease 2 (TMPRSS2, a protease involved in SARS-CoV-2 entry)50, 52, 55, 145-150.
When selecting an appropriate cell line for the evaluation of antibodies, a large number of factors can be considered: physiologic relevance, activity to be measured (e.g., neutralization, ADCC, etc.), desired assay throughput and readout (e.g., plaque formation), feasibility of using primary cells in lieu of immortalized cell lines, expression of host factors, tissue and species origins, and potential for amino acid polymorphisms in the receptor or coreceptor that might affect activity. For example, if an animal cell line will be used, species-specific differences in receptor or coreceptor expression levels and/or amino acid sequences may affect antibody activity. Likewise, if immortalized cell lines will be used, the results may not accurately reflect those obtained with primary cells. For example, several research groups have reported that the neutralization activity of some anti-SARS-CoV-2 mAbs is affected by ACE2 expression levels, leading to variable potency (in terms of both the half-maximal effective concentration [EC50] values and maximal percent inhibition) in different cell lines145, 151-153. If primary cells will be used, it may be beneficial to test cells from multiple donors (ideally of different sexes and ethnicities) to assess variability in potency. For viruses that infect multiple cell types (e.g., herpesviruses), either due to expression of the same receptor on multiple cell types or the use of multiple receptors, antiviral activity can be assessed in distinct cell types.
In addition to neutralization assays, cell type is an important factor to consider for other types of antibody assays as well, including assessments of Fc effector functions (e.g., ADCC, ADCP, CDC), ADE, and other types of studies. Characterization of the effector function is a consideration for both monoclonal and polyclonal antibodies. In general, these assays have not been well standardized, and it is often unclear which cell types and Fc effector functions are most likely to be relevant for clinical efficacy. To further complicate matters, Fc effector function assays often involve at least two cell types: target cells expressing the antigen and immune effector cells that respond to the IgG-bound antigen. In most cases, target cells consist of a cell line (e.g., CHO, Jurkat, or 293T) that has been transiently transfected or engineered to constitutively or inducibly express viral antigen(s), such as envelope protein. For example, Fc effector functions were assessed for the FDA-approved anti-EBOV mAbs (Inmazeb and Ebanga) using target cell lines with inducible expression of the EBOV glycoprotein154, 155. Alternatively, virus or viral-like particles may be used as targets in some assays (e.g., for ADCP). For effector cells, common cell types that have been used include Jurkat (immortalized human T) or NK-92 (immortalized human NK-like) cells engineered to stably express specific FcγRs and primary human cell types, such as NK cells, monocytes, monocyte-derived macrophages, or peripheral blood mononuclear cells. Many manufacturers have developed reporter cell lines (often Jurkat-based) to quantify FcγR activation as a surrogate for ADCC, and some research has been performed to compare these assays to classical ADCC assays156, 157. Fc effector function assays can be performed with cells that express different FcγRs and FcγR variants, e.g., the FcγRIIIa F158 and V158 variants, which have distinct binding affinities for IgG1 and IgG3158.
The selection of cell lines for assessment of ADE is also an important issue, particularly for viruses in which ADE is known to be a significant issue (e.g., Dengue and Zika viruses). For these viruses, ADE has been assessed using K562 (FcγRIIa+ human erythroleukemia), Raji (human B lymphoblastoid), U937 (human myeloid leukemia), THP-1 (human monocytic leukemia), and primary monocytes or macrophages159-162. In many cases, mAbs are engineered to have Fc substitutions that are expected to enhance, diminish, or abrogate Fc effector functions and/or ADE. In these cases, cell culture studies can be performed to verify that the substitutions have the intended effects. It may be beneficial to test multiple versions of a mAb in parallel to identify one with optimal properties, i.e., versions with an unmodified Fc region, different Fc substitutions, or distinct Fc glycosylation patterns. Lastly, cell lines for other types of studies should also be carefully selected, such as for studies of antibody resistance mechanisms, cell-cell transmission, and cell-cell fusion or syncytium formation.
Resistance
As part of development, it is critical to characterize the resistance mechanisms and pathways of antiviral antibodies in preclinical (and later clinical) studies. There are two different mechanisms by which viral resistance can develop. Naturally occurring resistance arises as the virus naturally evolves, as strains with increased fitness (e.g., due to enhanced infectivity, replication, or immune escape) become dominant in the human population. This type of resistance is unrelated to treatment but may lead to treatment failure or non-response. In contrast, treatment-emergent resistance happens in response to the specific antibody treatment, with treatment providing the selective pressure for the emergence of the resistant variant. The potential for antiviral antibody therapies to be affected by either form of resistance should be assessed in preclinical studies. These studies are often highly valuable for informing the dose and dosing interval, optimization of treatment regimens (e.g., mAb monotherapy vs. mAb combinations vs. mAbs+other antivirals), interpretation of clinical resistance data, identification of patients infected with susceptible viral variants, likelihood of cross-resistance with other mAbs (important for rescue/salvage therapy), and genomic surveillance efforts. Approaches for resistance characterization vary widely depending on the virus, antibody, and antibody target. For example, approaches for SARS-CoV-2 have included testing the effects of single amino acid substitutions in the spike protein (S) on antibody binding in biochemical assays, screening large libraries of S proteins with substitutions using a yeast display system, determining the effects of S substitutions on antibody neutralization in cell culture using pseudotyped VLPs, and performing resistance selection in cell culture and animal models with replication-competent chimeric (i.e., VSV-spike) and authentic viruses147, 163-166. For mAbs that target host proteins, resistance could potentially arise from genetic polymorphisms that alter the expression or sequence of the host protein, rather than through changes in the virus. In these cases, differences in host genetics (e.g., single-nucleotide polymorphisms) can be assessed by bioinformatics and, if necessary, biochemical or cell culture studies.
Some studies performed to assess resistance, particularly those involving the selection or characterization of replication-competent authentic or recombinant viruses, can raise significant ethics and biosecurity concerns. These studies are conducted only under appropriate biocontainment and in strict accordance with all applicable institutional, local, regional, and national guidelines and regulations. In some cases, it may be possible to adequately characterize antibody resistance using biochemical or cell culture assays that involve only the viral antigen or a replication-defective virus. In other cases, it may be possible to select for antibody resistance using a replication-competent virus that poses less risk, such as a replication-competent chimeric virus, a related animal virus that expresses the same epitope but cannot infect human cells, or a deliberately attenuated version of a human virus. One caveat of these approaches is that the substitutions observed in these viruses may not accurately predict the resistance substitutions observed with authentic virus. When an authentic virus must be used due to unavailability of other approaches or inadequate resistance information from such approaches, risk can be mitigated by using an authentic virus that is susceptible to current vaccines and antivirals and to which vaccine-induced or natural immunity is already widespread.
Animal Studies
Animal studies can be a powerful tool for assessing the safety and efficacy of antiviral antibody therapies prior to commencing clinical trials. To be useful, an animal model should recapitulate as closely as possible the critical aspects of human disease, including susceptibility to the pathogen, route of infection, viral tropism, severity of outcomes, pathophysiology of systemic and end organ disease as applicable, and host responses. Although non-human primates, being evolutionarily closest to humans, are most likely to fulfill these criteria, many other mammalian species have been successfully used as models of viral disease, including for assessing antibody therapies. Examples include mice for West Nile virus, ferrets for influenza, cotton rats for RSV, and hamsters for hantavirus and SARS-CoV-2167.
The role of animal studies in the development of antiviral therapies can be illustrated by the recent COVID-19 pandemic. For SARS-CoV-2, initial cell culture studies focused on identifying pathways underlying viral entry, tropism, molecular pathways of disease processes, and mechanisms for neutralization in cell lines and organ-like systems, with the ultimate goal being the discovery of effective preventive and therapeutic strategies, including antibody therapies. Although critical for identifying and measuring neutralization activity of antiviral antibodies, cell culture studies such as these cannot account for
in vivo activity or their distribution in the mucosal or lung tissue - often the point of entry and viral replication for SARS-CoV-2. Studies in animal models, including hamsters, transgenic mice, ferrets, and non-human primates validated the findings from cell culture studies and demonstrated the potential for efficacy, including of anti-SARS-CoV-2 antibodies
168-175. Similar studies were used to support EUA packages for mAbs that received such authorizations (
https://www.fda.gov/drugs/coronavirus-covid-19-drugs/cder-scientific-review-documents-supporting-emergency-use-authorizations-drug-and-biological). These studies demonstrated that monoclonal and polyclonal antibodies have the potential to protect against disease when used as prophylaxis and to improve outcomes when used as a therapeutic, thus providing preliminary data to support their investigation in clinical trials. They also demonstrated that ADE was not observed.
A critical question for advancing any therapeutic modality to clinical trials is related to the starting dose, which should be both safe and potentially efficacious. Although the definite proof of safe and effective dose will come from well-designed and adequately controlled clinical trials, a scientific, data driven rationale should inform the dose(s) chosen to advance to the clinic. Data supporting safety are obtained from nonclinical toxicology, safety pharmacology and other pertinent studies, as outlined in the appropriate FDA guidance documents176. Deriving a potentially efficacious dose, on the other hand, is not as standardized, and multiple approaches can be applied based on the availability of cell culture and animal models, new or existing pharmacokinetic (PK) and pharmacodynamic (PD) data in both animals and humans, physiologically based pharmacokinetic (PBPK) models, and biomarkers that correlate with efficacy. To continue with the example of the recent pandemic, the data submitted by sponsors in EUA packages sheds light on some of the methods used57. For all these antibodies, a necessary component and often the first step in efficacious dose estimating was identification of the antibody concentration providing 90% of maximal effect (EC90 value), then target clinical exposures that exceed this level, taking into account antibody distribution to the respiratory tract. Animal data can assist in deriving such concentrations, even through direct measurements or through methods to relate serum to tissue concentrations such as antibody biodistribution coefficient or allometric scaling177. It should be noted that anatomical and physiological characteristics differ, depending on the species and the degree of phylogenetic similarity with humans. Thus, biodistribution of the antibody to the site of action (such as lung or gastro-intestinal tract) may be quite different. This may also be further influenced by variability in disease presentation and pathological processes, for example.
Other considerations that influence the translatability of efficacy data from animal studies include differences in expression patterns of Fc receptors on effector cells and the affinity of the human antibodies to the animal orthologs178, 179. In addition, Fc modifications intended to alter FcRn or FcγR binding may not have the same effects in animals and humans. For example, M252Y/S254T/T256E, referred to as “YTE” variant, has a half-life four times longer than unmodified IgG1 in humans but a rapid clearance in rodents57, 180. To overcome these limitations, transgenic and humanized mouse models that incorporate human FcγR genes have been developed181, 182.
In certain infectious disease settings, specifically when clinical trials are not feasible or ethical, adequate and well-controlled animal studies can be used to provide substantial evidence for efficacy through a pathway known as approval under the Animal Rule. Detailed advice on considerations when developing a therapy under this pathway are described in the pertinent FDA guidance183. The guidance outlines many clinical and nonclinical aspects of such programs, including considerations for choosing the animal models, such as the challenge agent, susceptibility to disease, mechanisms of virulence, pathophysiology and its comparison with human disease (natural history studies), trigger for intervention, mechanism of action for the treatment, dose, and the necessary studies that can be used to derive a dose and dosing regimen in humans (such as PK and/or PD studies in animals and humans). It should be noted that seeking regulatory approval under the Animal Rule is not a way to circumvent performing clinical studies or to simplify the approval process for a therapeutic. Safety and PK studies in humans are still needed, whereas performing studies to adequately characterize at least two animal models, often undertaken under high containment conditions in BSL-4 facilities, adds significant complexity to this process. This is reflected by the number of approvals: a total of sixteen drugs are approved under this rule but none of them are antiviral antibodies, although there are four antibodies targeting bacterial infections or toxins (please refer to the FDA database for up-to-date information, see FDA webpage184).