☆ 4.8 Article

Intrinsic physicochemical profile of marketed antibody-based biotherapeutics

PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA (2021)

Journal

PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA

Volume 118, Issue 37, Pages -

Publisher

NATL ACAD SCIENCES

DOI: 10.1073/pnas.2020577118

Keywords

biologics; antibody; developability; computation; drug discovery

Funding

Boehringer Ingelheim Phar-maceutical, Inc.

Ask authors/readers for more resources

Protocol

Community support

Reagent

Community support

Automated Summary New
Abstract

By analyzing the physicochemical descriptors of Fv regions in marketed antibody-based biotherapeutics, this study derived a set of descriptors that can help guide hit selection, lead identification, and optimization of biotherapeutic candidates. The research found that 33% of CST antibodies deviate in physicochemical properties from currently marketed biotherapeutics, providing valuable information for risk assessment and collaborations.

Feeding biopharma pipelines with biotherapeutic candidates that possess desirable developability profiles can help improve the productivity of biologic drug discovery and development. Here, we have derived an in silico profile by analyzing computed physicochemical descriptors for the variable regions (Fv) found in 77 marketed antibody-based biotherapeutics. Fv regions of these biotherapeutics demonstrate significant diversities in their germlines, complementarity determining region loop lengths, hydrophobicity, and charge distributions. Furthermore, an analysis of 24 physicochemical descriptors, calculated using homology-based molecular models, has yielded five nonredundant descriptors whose distributions represent stability, isoelectric point, and molecular surface characteristics of their Fv regions. Fv regions of candidates from our internal discovery campaigns, human next-generation sequencing repertoires, and those in clinicalstages (CST) were assessed for similarity with the physicochemical profile derived here. The Fv regions in 33% of CST antibodies show physicochemical properties that are dissimilar to currently marketed biotherapeutics. In comparison, physicochemical characteristics of -29% of the Fv regions in human antibodies and -27% of our internal hits deviated significantly from those of marketed biotherapeutics. The early availability of this information can help guide hit selection, lead identification, and optimization of biotherapeutic candidates. Insights from this work can also help support portfolio risk assessment, in-licensing, and biopharma collaborations.

Authors

I am an author on this paper

Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.8

Not enough ratings

Secondary Ratings

Novelty

Significance

Scientific rigor

Rate this paper

Deep learning in preclinical antibody drug discovery and development

Yuwei Zhou, Ziru Huang, Wenzhen Li, Jinyi Wei, Qianhu Jiang, Wei Yang, Jian Huang

Summary: Antibody drugs have become essential in biotherapeutics and have benefited patients with various diseases. However, their development process is time-consuming, costly, and risky. To accelerate development, reduce costs, and increase success rates, artificial intelligence, particularly deep learning methods, are extensively used in all stages of preclinical antibody drug development. This review systematically summarizes the use of deep learning in antibody drug discovery and development, including antibody encodings, deep learning architectures, and models. We also critically discuss the challenges, opportunities, current applications, and future directions of deep learning in antibody drug development.

METHODS (2023)