Panel Discussion Summary
One of the most popular live discussions across the week explored antibody affinity maturation and developability improvements. Silvia Hnatova revisits the top themes from the session, featuring:
Dr Stephen Parmley, Vice President, Molecular Biology & Protein Science at AnaptysBio (Chair)
Dr Laura Walker, Director at Adimab, LLC
Dr Andre Frenzel, Founder and CSO at Yumab
Dr Danielle Dicara, Principal Scientific Researcher, Antibody Engineering at Genentech
Dr Alon Wellner, Postdoctoral Researcher at UC Irvine
Dr Stephen Parmley chaired a very practical session on antibody affinity maturation and developability improvements. He opened the session by highlighting the importance of improving antibody libraries free of liabilities.
The first discussion focused on the sequence of steps needed to correct antibody liabilities.
Dr Parmley asked the panel “if an antibody needs to be further optimized and has liabilities, would you initiate diversification for maturation before or after fixing the antibodies”? There were different opinions from panel members, depending on the models they employ, ultimately concluding that the answer would be dependent on the specific antibody.
Dr Laura Walker said that she would first try to identify where the liability comes from. Dr Andre Frenzel listed several approached available for liability correction, including bioinformatic and biomedical approaches.
Dr Danielle Dicara added that the liable residues do not always have to be removed if they do not impact folding, although this may be dependent on the circumstances. Dr Alon Wellner added that in his system they cannot control mutations as the combinations are random and that liabilities can also emerge during affinity maturation.
Dr Parmley concluded that if the liability remains to be fixed at the end of the process, it may impact the affinity maturation that may be context-dependent. He prefers to remove liabilities as early as possible in the process, which usually takes about two weeks using his platform.
Dr Wellner asked the panel whether a liability residue is always a liability or whether it is context-dependent. Dr Walker responded that this needs to be tested depending on a specific antibody, for example, if it does not impact binding it may not be a liability. Dr Parmley and Dr Frenzel agreed that some of the liabilities arise from manufacturing and may not be found in vivo.
Dr Parmley addressed a question to Dr Walker about liability in germline residues and whether they can go away. She responded that from her practice, the developability profiles of antibodies from B cells is broadly good, apart from naïve B cells (see her talk).
She pointed out that thermostability may decline as an effect of mutations, that may accumulate during affinity maturation. Dr Parmley pointed out that ultimately, the main liability will be the specificity of the antibodies.
Related to Dr Wellner’s talk, Dr Dicara asked about Tm threshold as a function of somatic mutations. Dr Walker explained that below a certain point, B cells antigen uptake is compromised.
There was a discussion about how and why thermostability would be affected and how this could relate to gene expression in B cells.
The panel answered an interesting question from the audience about whether there is a list of sequence liabilities that is well-known. Dr Parmley and Dr Dicara explained that although some are well-described, some can be identified through stress-testing.
Next, the panel discussed possible dead ends during antibody optimization. Dr Wellner replied to Dr Parmley’s concerns about mutation rates in his model being too high. He said that based on his model, the mutation rate is not yet as high as he would like it to be, leading to dead ends. However, his only dead end so far was that he was unable to mature the antibody any further.
Dr Dicara voiced her concerns about stopping mutations in a continuous mutation system to identify and characterize them. Dr Wellner responded that there are points where he can isolate plasmids from specific clones, admitting that there may be some noise present.
The panel answered specific questions from the audience.
There was a discussion about improvements in antibody affinity by using insertions and deletions. Dr Parmley stated that although this is infrequent in his practice, the main improvement he has seen was in germline antibodies pulled from naïve cells. He said that he may not have a complete picture because he only sees ‘what was selected for’, saying that insertions are mainly seen in CDRs (duplications of e.g. 8 residues).
Dr Frenzel answered a question from the audience about whether he has seen ‘epitope walking’ with affinity maturation. In Dr Frenzel’s experience, epitope drift may not be fully avoided. Dr Walker said that in her experience she only saw this once, that a true shift would happen.
There was an interesting question from the audience about how immunogenicity could be avoided, that was debated by the panel. Dr Parmley answered that there are a number of ways to reduce this, mainly good biophysical properties of the antibody and avoidance of T cell epitopes with in silico tools.
It was agreed that immunogenicity may not be spotted until the antibody is tried in humans. This was followed by a technical discussion on salt conditions during antibody optimization.
Dr Walker answered a question from the audience asking her opinion on the most successful methods of affinity maturation. Her opinion was that pre-made libraries targetting multiple residues at the same time would be the best method. Dr Wellner added that in his system the most mutations are achieved in CDR1 and CDR2 regions.
Dr Parmley pointed out that COVID-19 convalescent antibodies do not have much activity in H3 regions, suggesting affinity maturation could be possible in these regions.
Dr Parmely highlighted that posttranslational modifications are of main concern in clinical development, and tryptophan residues are the main focus in affinity maturation. Dr Parmley asked Dr Walker about developability properties of in vitro vs in vivo antibodies. Dr Parmley cautioned that right filters are crucial when using antibodies from different kinds of display, e.g. phage.
The discussion was closed with an interesting question from the audience on “how closely do current antibodies follow developability rules?”, questioning the current practices and how successful they are in the antibody selection process.
Dr Walker pointed out that phage antibodies currently in the clinic did not follow developability rules employed now. She suggested that it would be “interesting to repeat the same analysis in 20 years”, to see how successful current developability practices will prove to be.
