Executive Summary
The conference was largely focused on COVID-19 antibodies, elaborating on the talks from the Antibody Engineering & Therapeutics Virtual Event held in summer 2020. The focus of this virtual event was on the structure of neutralization antibodies and relevance for vaccine design.
Prof Pamela Bjorkman from Caltech opened the topic with a keynote presentation on The Structural Basis of Neutralization by Antibodies Against Viral Fusion Proteins. She highlighted that the main strategy to prevent COVID-19 infection is via stopping COVID-19 entry into human cells, which would prevent the subsequent multiplication of the virus. As COVID-19 infects human cells by binding virus’ spike protein to the ACE2 receptor in human cells, neutralizing antibodies are best suited to stop this binding.
Prof Bjorkman, in collaboration with the Rockefeller University, found that patients who responded to COVID-19 infection (convalescent donors) produced monoclonal antibodies. Using electron microscopy, the consortium asked whether they could determine the structural correlates of antibody-mediated neutralization of coronavirus infection. Prof Bjorkman was also interested in whether there are multiple epitopes targeted in recovered individuals or a single predominant epitope. To examine this, Prof Bjorkman’s group developed a new technique called negative stain polyclonal epitope mapping (published by Bianchi et al., 2018; Nogal et al., 2020). Using this technique, they were able to determine the binding of single predominant epitopes in polyclonal antibodies from patients named COV21 and COV57.
Prof Bjorkman presented her research using single-particle cryo-EM to highlight the structural correlates of neutralization for potent monoclonal antibodies. COVID-19 spike protein structure includes receptor binding domains (RBDs) that can be in two positions. To bind to ACE2, it needs to be in an “up” position. Neutralizing antibodies block the ACE2 binding site and only bind to “up” RBDs. All neutralizing antibodies identified by Prof Bjorkman were found to bind to C105 epitope, supported by previous work from other laboratories. Prof Bjorkman’s group summarized the neutralization mechanisms based on antibody-spike structures in a paper authored by Barnes et al. (2020).
Prof Bjorkman explained in detail the neutralization mechanism of one antibody named C144, that only binds to “down” RBDs. She revealed that C144 bridges between adjacent RBDs, locking the spike into closed conformation, highlighting the conformational changes in play.
The structures of neutralizing antibodies were grouped into most abundant classes to lay the ground for vaccine and therapeutic strategies. Prof Bjorkman highlighted that the prediction between competition/no competition among neutralizing antibodies is complicated. To illustrate her point, she presented an exception to the classification her group developed. BG1-22, a VH3053/VH3-66 antibody with a long CDRH3 binds to the class 1 RBD epitope.
In the second part of her talk, Prof Bjorkman pointed out that even after developing a COVID-19 vaccine, we are at risk of zootic transmission from coronaviruses from other hosts. She presented her work on a pan-coronavirus vaccine. Using a previously published system, immunogens can be attached to nanoparticles (Brune et al., 2016; Escolano/Gristic et al., 2019). Prof Bjorkman combined the SpyTag particle system with COVID-19 antigens to make four types of RBD-nanoparticles: homotypic SARS-2 and three mosaics (combining SARS-2 particles with other particles). For this, they used RBDs from 8 sarbecovirus spike proteins for making nanoparticles, including RBDs from viruses with spillover potential. They immunized and boosted mice with RBD-nanoparticles of different types. SARS-2 spike ELISA assay results showed that antibodies raised against RBD monomers can bind to RBDs on S trimer.
Prof Bjorkman expressed her surprise that homotypic and mosaic sera led to the same degree of response in immunized mice. She said that “multimerization of RBDs on nanoparticles enhances immunogenicity compared with soluble antigen”. Mosaic sera bound and neutralized bat strains better than homotypic sera, but “neutralization of matched and mismatched strains was only observed after priming”.
Prof Bjorkman pointed out at that protection against a mismatched strain can be induced via mosaic nanoparticles, implying neutralization. Prof Bjorkman’s group then repeated the ELISA’s using antibodies from plasma of people who overcame SARS-CoV-2, that did not bind well to coronaviruses of other types. She concluded that cross-neutralization response SARS-CoV-2 seems unlikely, by saying that “Co-display of SARS-2 RBD with other RBDs shows no disadvantages compared with homotypic SARS-2 nanoparticles for eliciting neutralizing Abs against SARS-Cov-2“. In summary, having had COVID-19 wouldn’t protect against the next pandemic.
Prof Bjorkman concluded the talk by highlighting that mosaic-RBD nanoparticles are promising vaccine candidates for SARS-COV-2 and potential future emerging zoonotic sarbecoviruses. In her opinion, a mosaic nanoparticle strategy would be the best approach to protect against SARS-COV-2 and potentially emerging sarbecoviruses.
Prof Dennis Burton from Scripps Research Institute elaborated upon the results presented in previous virtual events four months ago in his presentation on Isolation of Potent SARS-CoV-2 Neutralizing Antibodies and Protection from Disease in a Small Animal Model. He first presented the collaborative effort to identify neutralizing antibodies from COVID-19 patients, published in Science (Rogers et al., 2020). In summary, the consortium looked into immune responses to SARS-CoV-2 spike protein and compared it to the original SARS-CoV-1. The plasma from 17 donors was isolated soon after symptom onset, followed by neutralization assays in vitro using 2000 mAbs isolated from 8 donors. 33 mAbs from 3 donors showed notable neutralization. Prof Burton highlighted that vaccines can do much better than natural infection, by targeted design.
Prof Burton’s group took the potent antibodies forward to describe the binding mechanism to SARS-CoV-2. Binding to 3 major epitopes on RBD (RBD-A, RBD-B, RBD-C) and 3 epitopes that are non-RBD (D proteins) was described. Prof Burton explained that the antibodies against RBD proteins proved to be most potent, although some antibodies showed incomplete neutralization. He focused the rest of his talk on the antibodies binding to RBD proteins.
RBD-A-ACE2 binding specific antibodies were most prevalent and most potent. Prof Dennis Burton explained this by competition between neutralizing antibodies to RBD-A epitope with the ACE2 receptor (Yuan et al., 2020): the most potent antibodies showed most competition with the ACE2 receptor. Prof Burton’s consortium concluded that the antibodies are very potent with very few somatic mutations, leading to a concept of ‘super-antibodies’ that occur in a minority of infected individuals as a minor part of the antibody response. RBD-B seemed to be a better target for cross-reactive antibodies, despite displaying a lower potency. Non-RBD S-protein nAbs demonstrated both weak and incomplete neutralization.
Prof Burton’s groups further examined the potential of the potent antibodies from in vitro experiments to validate these in vivo using a Syrian hamster model. The antibodies selected were against RBD-A epitope and non-RBD epitopes. The potent neutralizing antibody CC12.1 against RBD epitope protects hamsters from COVID-19 disease, both weight loss and viral titers in the lungs. As expected, the CC12.23 antibody against S protein does not protect hamsters from COVID-19.
Prof Burton concluded his talk with very interesting results of antibody performance against different COVID-19 variants, including the recently emerged variant from Europe. Selected most potent antibodies, including the antibody CC12.1 against RBD epitope, identified in convalescent plasma displayed protection against most COVID-19 variants. However, some COVID-19 variants escaped protection from monoclonal antibodies. Prof Burton closed his talk listing the SARS-CoV-2 nAbs that were licensed to Merck KGaA and Serum Institute of India to undergo clinical trials. He emphasised the need of providing the antibodies to low and middle-income countries.
Prof Andrew Ward from Scripps Research Institute dived deeper into COVID-19 antibodies and their structure with his keynote presentation on Characterizing Polyclonal Antibody Responses Using Single-Particle Electron Microscopy. He started by introducing a technique developed in his lab: electron microscopy polyclonal epitope mapping (EMPEM; Bianchi et al., 2018). By combining traditional serology with EMPEM, Prof Ward’s group can “visualize diverse polyclonal antibody responses to subunit vaccines and rapidly map epitopes, accelerating vaccine design process”. Prof Ward presented his comprehensive work on HIV epitopes and he closed his talk by presenting his research on antibodies in sera of COVID-19 patients. He concluded that polyclonal Fabs resemble known neutralizing antibodies isolated from COVID-19 convalescent patients.
In parallel to the focus on COVID-19, the virtual event covered a range of talks that discussed antibody engineering strategies and biophysical properties of antibodies.
Dr Paul Carter from Genentech delivered a fascinating keynote presentation on Engineering Bispecific Antibodies as Therapeutics: Utilizing Intrinsic Heavy/Light Chain Pairing Preferences and Mitigating High Viscosity. Bispecific antibodies present an exciting development in the therapeutics field: there are 2 FDA approved antibodies (blinatumomab and emicizumab) and 186 in clinical development.
Dr Carter pointed out that bispecific antibodies have several advantages compared to traditional antibodies: their design exploits the modular architecture of antibodies, aiming to match the desired mechanism of action and clinical application, and they have long serum half-life.
Although the current manufacturing process for bispecific antibodies is very successful and robust, it is resource-intensive, expensive and inconvenient for manufacturing.
Using current methods, light and heavy chains are expressed separately in two separate cells and then assembled in vitro into bispecific IgGs. Genentech improved and simplified the manufacturing process: bispecific IgGs were re-designed for 1-cell production. Dr Carter presented work on v11 IgG as an example of this design (Dillon et al., 2017). He showed that 1-cell and 2-cell bispecifics have similar in vitro potency and pharmacokinetics. Most significantly, 1-cell production of bispecific IgGs at large scale is successful. One such bispecific IgG has reached the clinic.
Building upon his work on 1-cell bispecific IgGs, Dr Carter presented results of cognate light chain/heavy chain (LC/HC) pairing preference. To evaluate the LC/HC preference, his group quantified bispecific IgG yield with orbitrap LC-M after transient expression of IgG pairs in a single cell (Joshi et al., 2019). It was found that cognate LC/HC pairing preference is a common occurrence and can be strongly influenced by residues in CDR L3 and CDR H3. Such knowledge may be advantageous to reduce risks occasionally encountered with engineered proteins and the number of Fab mutations needed for efficient production of bispecific IgG where there is strong intrinsic pairing preference.
In the second part of his talk, Dr Carter focused on the viscosity of bispecific antibodies. Due to their clinical potential, it is necessary to consider the delivery method. Currently, subcutaneous delivery forms one-third of all delivery methods. Because it is highly convenient, it improves patient compliance with the drug. The cons of some antibodies are their high viscosity, limiting delivery to small injection volumes and increasing injection site pain.
Viscosity properties of bispecific antibodies are unknown. Although subcutaneous delivery and low viscosity would be highly desirable, Genentech found that bispecific antibodies have an unusually high viscosity. Dr Carter suggested several strategies to resolve the high viscosity including the addition of excipients to antibody formulation (NaCL or Arginine-HCl), co-formulation with hyaluronidase or re-engineering of the antibody.
Genentech found that the introduction of specific mutations into a bispecific antibody reduces viscosity to its parent IgG. Dr Carter illustrated his point with the same strategy applied to monospecific antibodies, using anti-GCGR IgG antibody as an example. He demonstrated that mutations of single aromatic residues can help to reduce the viscosity of both monospecific and bispecific antibodies.
Related to Genentech’s fascinating results on the engineering of bispecific antibodies, Dr Greg Lazar (Genentech) delivered a presentation on Engineered Antibody Platforms for Receptor Agonism. There are several mechanisms on how to improve the agonist potential of antibodies.
Dr Lazar illustrated his point with an example: TNFRSF antibodies are typically poor agonists on their own and none of the current TNFRSF agonist antibodies has yet passed clinical trials to become therapeutics. The emphasis of the talk was on Fc-mediated cross-linking and our inability to control this.
Dr Lazar proposed several engineering design principles that could help to overcome this challenge, exploring receptor signalling without relying on Fc receptor engagement. He concluded the talk by highlighting that FcγR expression is variable and not under our clinical control. Instead, engineering of the antibody may enable better control over the biological and pharmacological profile of the agonists.
Dr Laura Walker from Adimab delivered a unique presentation on Biophysical Properties of Human B-cell derived Antibodies. In her publication (Jain et al, 2020), she divided the biophysical properties of 400 human B cell-derived mAbs into five main groups: stability, hydrophobicity, long-term aggregation propensity, ELISA plate binding and cross interaction propensity.
The focus of her research was to address whether B cell subsets have different biophysical properties. She found a few correlations between the biological and biophysical properties of antibodies:
Dr Walker closed the talk with an interesting observation that naïve B cell-derived mAbs show higher thermostability compared to memory B cell-derived mAbs. Somatic mutations led to decreased thermostability and the location of destabilizing mutations was found to be antibody-dependent. In general, Kappa LCs and VH4 germline antibodies were found to be more thermostable than other antibodies. Dr Walker concluded that human B cell-derived antibodies show favourable polyreactivity, hydrophobicity and thermal stability properties. Her work laid important ground on principles to control polyreactivity, hydrophobicity and thermal stability of antibodies.
