Neutralizing antibodies represent an important class of therapeutics that could provide immediate benefit in the treatment of SARS-CoV-2 or as a passive prophylaxis before vaccination. Passive prophylaxis also could be an alternative to vaccination in populations where vaccines have been found to be less efficacious.1
When AbCellera became a participant in DARPA’s P3 (Pandemic Prevention Platform) program in 2018, the aim was to build a pandemic-ready, rapid response, antibody discovery platform capable of developing field-ready medical countermeasures within 60 days of isolation of an unknown viral pathogen. At the time, AbCellera didn’t know that within two years its platform and team would be mobilized to react in record time to a real-life global health scenario, the COVID-19 pandemic.
In every person’s body, billions of antibodies exist, each with unique properties and functions. AbCellera’s full-stack AI-powered antibody discovery platform deeply searches natural immune responses to identify antibodies with desired properties such as virus neutralization, safety, longevity, and manufacturability.
The company’s mission is to make its technology stack available and to empower all antibody-based drug discovery programs. To date, the technology stack has been used in over 100 programs to address a wide range of indications in addition to COVID-19 and other infectious disease.
Putting a plan into action
At the end of February 2020, AbCellera obtained a single blood sample from a recovered COVID-19 patient. Close to six million peripheral blood mononuclear cells (PBMCs) were screened in three days, and thousands of therapeutic antibody potentials eventually narrowed to a final subset of 500 unique antibodies that recognized the interaction between the SARS-CoV-2 spike protein’s receptor binding domain (RBD) and the angiotensin converting enzyme 2 (ACE2) cellular receptor.
These candidates underwent extensive analysis and characterization—approximately 500 data points per molecule. An important aspect of the technology stack is the machine learning/artificial intelligence (ML/AI) tool that allows analysis of large data clouds comprised of approximately 250,000 data points for this data set.
“We were able to visualize the analysis and down selection process using our custom in-house visualization software, Celium,” said Ester Falconer, PhD, chief technology officer at AbCellera, who led the AI-powered antibody discovery platform’s development.
Bryan Jones, PhD, senior research fellow at Eli Lilly and Company, co-led most of the discovery activities that took place at Lilly, in close partnership with AbCellera. His group played a large role analyzing and transitioning the initially discovered set of hundreds of antibodies—through selection, production, and characterization of a narrowed set of antibodies—that ultimately led to the identification of LY-CoV555 (bamlanivimab) for clinical development.
“Due to the rapidly evolving and geographically diverse nature of the SARS-CoV-2 virus, continued scientific innovation remains critical to develop additional treatments,” said Jones, who adds that Lilly remains committed to developing complementary neutralizing antibodies to address potential SARS-CoV-2 variants that undoubtedly will arise. Currently, a next-generation antibody, LY-CoV1404, is in the preclinical pipeline.
Intense pressure and challenges
“Our platform and the extended team were ready,” said Falconer. “But in reality, this was a new virus that the world was racing against. We had no experience with it and did not know what type of immune response it would elicit. To develop an antibody database to mine, we had only one blood sample from one of the first recovered patients in the United States.
“Plus, this was an early immune response. The blood was drawn 20 days after onset of symptoms. It was not clear what we were going to find. We had one shot, it had to work, and it had to be the fastest discovery ever. The pandemic-ready platform and team had to deliver.”
The speed at which things needed to progress was the biggest challenge. “Because of the urgency,” Jones recalled, “we were making decisions based on little data and testing a new virus, while simultaneously trying to coordinate these activities across multiple organizations ranging from Lilly and AbCellera, to the Vaccine Research Center (VRC) at the National Institutes of Health (NIH), to the numerous academic collaborators who were providing critical data.”
All of these efforts played out against a backdrop of everyone trying to learn how to work nearly completely remotely.
Antibodies were characterized deeply to ensure that any data points that could direct the team in the right direction would not be overlooked. The number of data points per antibody highlights the importance of the visualization software Celium and the infrastructure AbCellera developed to generate, aggregate, and process the data. High-stakes decision making was made in real time.
The discovery efforts, which have been detailed in Science Translational Medicine, included the rapid identification and characterization of the potent anti-spike neutralizing antibody, LYCoV555, derived from PBMCs isolated from a patient after recovery from COVID-19.1
It takes a village
“With the right team and technology, what seems impossible is achievable. With Lilly, we broke the mold for the fastest discovery through IND for any drug, challenging the current drug development process,” Falconer said. “It does not need to take years and years. Nearly six million cells were screened in three days, antibody genes were sequenced in an additional two days, and antibodies were generated and tested a week later—an incredible feat.”
“There really was not a tremendous difference in the workflow of discovery, clinical development, and manufacturing, except that everyone was singularly focused,” Jones added. “But a key difference was partnership with the FDA to identify innovative approaches to get treatments to patients as quickly as possible. We learned that a large and committed group of people concentrated on a sole objective can do amazing things.”
Pandemics are unpredictable
Bamlanivimab proceeded from sample to Emergency Use Authorization (EUA) in just over eight months, a groundbreaking achievement in the development of antibody therapeutics. In November 2020, the FDA granted an EUA for bamlanivimab alone, and subsequently granted an EUA in February 2021 for bamlanivimab together with etesevimab (LY-CoV016), for the treatment of mild to moderate COVID-19 in adults and pediatric patients (12 years of age and older weighing at least 40 kg) with positive results of direct SARS-CoV-2 viral testing, and who are at high risk for progressing to severe COVID-19 and/or hospitalization.
Due to the sustained increase of SARS-CoV-2 variants that are resistant to bamlanivimab alone, Lilly requested and received in April 2021 a revocation of the EUA for bamlanivimab used alone. The EUA remains in effect for the combined use of bamlanivimab and etesevimab.
The Carterra LSA facilitated the rapid kinetic characterization of the selected recombinantly expressed antibodies to the SARS-CoV-2 spike protein and the RBD.
Spike protein–dependent viral entry is initiated by upward movement of the RBD at the apex of the protein, allowing access to bind the ACE2 cellular receptor. Upon receptor engagement, coordinated proteolytic cleavage and shedding of the S1 subunit occur, and conformational rearrangement of the S2 subunit leads to viral fusion with the cell and transfer of genetic material.1
“The LSA allowed for an extensive epitope analysis to be performed on the antibody panel which demonstrated broad epitope coverage, distinct classification into known binder categories such as S1 and S2, and the determination of neutralization of ACE2 binding,” said Dan Bedinger, PhD, applications scientist team lead at Carterra.
These assessments were key elements in the ranking and selection of the subset of leads for further characterization.
The ability to rapidly complete these analyses in a parallel, unattended fashion with minute amounts of recombinant antigen—a scarce and valuable resource—made it possible to characterize the full candidate panel in an extremely accelerated development timeline, 90 days from initiation to first in human.
According to Bedinger, no other bioanalytical platform could have provided this rich picture of epitope binning classification, a picture that delivered a deep understanding of the interaction between sequence diversity and epitope recognition.
Coronavirus Immunotherapy Consortium
As variants of SARS-CoV-2 emerge, Carterra is collaborating with the Coronavirus Immunotherapy Consortium (CoVIC), a Gates Foundation–sponsored program. CoVIC has collected nearly 300 antibodies from a wide variety of sources that target the SARS-CoV-2 spike protein.
In a highly interdisciplinary effort across many institutions, these antibodies are being evaluated for a variety of properties including viral neutralization and escape, effector function, epitope recognition, and binding specificity. The LSA is being utilized to characterize the relative binding of each antibody to a variety of spike mutants and to create a comprehensive epitope binning profile.
“Data show that these antibodies can be clustered into epitope communities which map to various regions or faces of the spike protein when analyzed by cryo-EM,” said Bedinger. “Clones within these communities often share properties like neutralization and binding sensitivities to certain types of mutations.”
With a high-throughput approach, new antibodies can be rapidly binned into relevant communities and compared against the existing antibody population to understand their novelty and to predict their behavior.
1. Jones BE, Brown-Augsburger PL, Corbett KS, et al. The neutralizing antibody, LY-CoV555, protects against SARS-CoV-2 infection in nonhuman primates. Sci. Transl. Med. 2021; 13(593): eabf1906. DOI: 10.1126/scitranslmed.abf1906.