Toward generalizable prediction of antibody thermostability using machine learning on sequence and structure features.

Over the last three decades, the appeal for monoclonal antibodies (mAbs) as therapeutics has been steadily increasing as evident with FDA's recent landmark approval of the 100th mAb. Unlike mAbs that bind to single targets, multispecific biologics (msAbs) have garnered particular interest owing to the advantage of engaging distinct targets. One important modular component of msAbs is the single-chain variable fragment (scFv). Despite the exquisite specificity and affinity of these scFv modules, their relatively poor thermostability often hampers their development as a potential therapeutic drug. In recent years, engineering antibody sequences to enhance their stability by mutations has gained considerable momentum. As experimental methods for antibody engineering are time-intensive, laborious and expensive, computational methods serve as a fast and inexpensive alternative to conventional routes. In this work, we show two machine learning approaches - one with pre-trained language models (PTLM) capturing functional effects of sequence variation, and second, a supervised convolutional neural network (CNN) trained with Rosetta energetic features - to better classify thermostable scFv variants from sequence. Both of these models are trained over temperature-specific data (TS50 measurements) derived from multiple libraries of scFv sequences. On out-of-distribution (refers to the fact that the out-of-distribution sequnes are blind to the algorithm) sequences, we show that a sufficiently simple CNN model performs better than general pre-trained language models trained on diverse protein sequences (average Spearman correlation coefficient, ρ, of 0.4 as opposed to 0.15). On the other hand, an antibody-specific language model performs comparatively better than the CNN model on the same task (ρ= 0.52). Further, we demonstrate that for an independent mAb with available thermal melting temperatures for 20 experimentally characterized thermostable mutations, these models trained on TS50 data could identify 18 residue positions and 5 identical amino-acid mutations showing remarkable generalizability. Our results suggest that such models can be broadly applicable for improving the biological characteristics of antibodies. Further, transferring such models for alternative physicochemical properties of scFvs can have potential applications in optimizing large-scale production and delivery of mAbs or bsAbs.

Expression liabilities in a four-chain bispecific molecule.

Multispecific antibodies, often composed of three to five polypeptide chains, have become increasingly relevant in the development of biotherapeutics. These molecules have mechanisms of action that include redirecting T cells to tumors and blocking multiple pathogenic mediators simultaneously. One of the major challenges for asymmetric multispecific antibodies is generating a high proportion of the correctly paired antibody during production. To understand the causes and effects of chain mispairing impurities in a difficult to express multispecific hetero-IgG, we investigated consequences of individual and pairwise chain expression in mammalian transient expression hosts. We found that one of the two light chains (LC) was not secretion competent when transfected individually or cotransfected with the noncognate heavy chain (HC). Overexpression of this secretion impaired LC reduced cell growth while inducing endoplasmic reticulum stress and CCAAT/enhancer-binding protein homologous protein (CHOP) expression. The majority of this LC was observed as monomer with incomplete intrachain disulfide bonds when expressed individually. Russell bodies (RB) were induced when this LC was co-expressed with the cognate HC. Moreover, one HC paired promiscuously with noncognate LC. These results identify the causes for the low product quality observed from stable cell lines expressing this heteroIgG and suggest mitigation strategies to improve overall process productivity of the correctly paired multispecific antibody. The approach described here provides a general strategy for identifying the molecular and cellular liabilities associated with difficult to express multispecific antibodies.