Dataset Information

Co-optimization of therapeutic antibody affinity and specificity using machine learning models that generalize to novel mutational space.

ABSTRACT: Therapeutic antibody development requires selection and engineering of molecules with high affinity and other drug-like biophysical properties. Co-optimization of multiple antibody properties remains a difficult and time-consuming process that impedes drug development. Here we evaluate the use of machine learning to simplify antibody co-optimization for a clinical-stage antibody (emibetuzumab) that displays high levels of both on-target (antigen) and off-target (non-specific) binding. We mutate sites in the antibody complementarity-determining regions, sort the antibody libraries for high and low levels of affinity and non-specific binding, and deep sequence the enriched libraries. Interestingly, machine learning models trained on datasets with binary labels enable predictions of continuous metrics that are strongly correlated with antibody affinity and non-specific binding. These models illustrate strong tradeoffs between these two properties, as increases in affinity along the co-optimal (Pareto) frontier require progressive reductions in specificity. Notably, models trained with deep learning features enable prediction of novel antibody mutations that co-optimize affinity and specificity beyond what is possible for the original antibody library. These findings demonstrate the power of machine learning models to greatly expand the exploration of novel antibody sequence space and accelerate the development of highly potent, drug-like antibodies.

SUBMITTER: Makowski EK

PROVIDER: S-EPMC9249733 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Multispecific antibody-like molecules have the potential to advance the standard-of-care in many human diseases. The design of therapeutic molecules in this class, however, has proven to be difficult and, despite significant successes in preclinical research, only one trivalent antibody, catumaxomab, has demonstrated clinical utility. The challenge originates from the complexity of the design space where multiple parameters such as affinity, avidity, effector functions, and pharmaceutical properties need to be engineered in concurrent fashion to achieve the desired therapeutic efficacy. Here, we present a rapid prototyping approach that allows us to successfully optimize these parameters within one campaign cycle that includes modular design, yeast display of structure focused antibody libraries and high throughput biophysical profiling. We delineate this approach by presenting a design case study of MM-141, a tetravalent bispecific antibody targeting two compensatory signaling growth factor receptors: insulin-like growth factor 1 receptor (IGF-1R) and v-erb-b2 erythroblastic leukemia viral oncogene homolog 3 (ErbB3). A MM-141 proof-of-concept (POC) parent molecule did not meet initial design criteria due to modest bioactivity and poor stability properties. Using a combination of yeast display, structured-guided antibody design and library-scale thermal challenge assay, we discovered a diverse set of stable and active anti-IGF-1R and anti-ErbB3 single-chain variable fragments (scFvs). These optimized modules were reformatted to create a diverse set of full-length tetravalent bispecific antibodies. These re-engineered molecules achieved complete blockade of growth factor induced pro-survival signaling, were stable in serum, and had adequate activity and pharmaceutical properties for clinical development. We believe this approach can be readily applied to the optimization of other classes of bispecific or even multispecific antibody-like molecules.

Dataset Information

Co-optimization of therapeutic antibody affinity and specificity using machine learning models that generalize to novel mutational space.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets