Across the week there were a number of fascinating panel discussions. Here we explore some of the key themes that came from a discussion on antibody discovery and developability featuring:
Dr Andrew Bradbury, Chief Scientific Officer at Specifica
Tushar Jain, Principal Scientist at Adimab
Dr Lucy Ahmed, Postdoctoral Fellow at Boehringer Ingelheim Pharmaceuticals
Tileli Amimeur, Senior Data Scientist at Just-Evotec Biologics
Dr John McCafferty, CSO at Iontas
Dr Jordan Dimitrov, Scientist at the Department of Immunology at INSERM - Centre de Recherche des Cordeliers
This very technical discussion focused on the engineering aspects of COVID-19 antibodies, supported by opinions of the leaders in the field. Dr Bradbury opened the discussion by asking participants about their approaches to antibody design and the role of AI. Tileli Amimeur explained that the 'whole AI approach is entirely sequence-based', explaining the role of AI in antibody engineering saying that 'we need a lot of data for the approach to work'. She summarized her approach as depending on the generation of the data, to be
able to answer the antibody design questions better. Tushar Jain explained the challenges of publishing huge datasets of sequences in a public domain. Tileli Amimeur elaborated his point adding that the public datasets usually lack negative examples, that 'would enable us to learn'. There was a technical discussion sparked by the audience questions on general antibody design in silico, with Tushar Jain concluding that a parallel design approach testing affinity and specificity would be very useful. Dr Dimitrov
disagreed saying that antibody design is a 'case-by-case' problem, saying that antibody cannot be designed based on a single parameter only. Dr Bradbury asked the panel whether any blood proteins could modify antibody function. Dr Dimitrov replied that there are many reasons for hemolysis and that they are trying to understand the process. Dr Bradbury asked if there are any examples of changes of antibody specificity following hem exposure, with Dr Dimitrov explaining that hem exposure does not affect antibody binding.
The panel discussion was followed by an intriguing exchange between Dr Bradbury and Dr McCafferty, who discussed their experience in COVID-19 antibody design and their vision of future COVID-19 treatments. They agreed that it is important to keep as many as possible antibodies in the clinical trials, to have the most chances to beat the virus. Dr McCafferty highlighted that once a vaccinated patient is COVID-19 positive, the antibodies will be needed because the vaccine will no longer be useful. Dr McCafferty and Dr Bradbury agreed on the ideal future vision for COVID-19 therapy. If a patient would test positive for COVID-19 test, a dose of neutralizing antibody could be administered immediately. Through his research, Dr McCafferty highlighted that by using mRNA from convalescent plasma donors, they obtained a library of potent neutralizing antibodies within two months. Unfortunately, his research was thrown into unknown charters because the that UK funding was withdrawn at short notice, due to the existence of cheaper antibodies being repurposed for COVID-19. Dr McCafferty is hopeful that a new partner could help to bring the neutralizing antibodies into the clinic.
Dr Bradbury and Dr McCafferty then discussed the success of different neutralizing antibodies. This prompted a technical discussion about motif design in the 3-53 gene that encodes the immunoglobulin heavy-chain variable region, that was previously shown to target the spike protein of COVID-19.
Dr McCafferty outlined that the problem with selecting higher affinity antibodies might generate 'stickier' antibodies. He said that through his research, he learned that some paratopic residues have a higher affinity. Dr Bradbury elaborated on the point explaining that the 'the stability is increased by virtue of interaction, preventing the antibody from falling apart'. They concluded that 'we don't want to have high-affinity antibodies that are difficult to use'.
