Computational optimization of antibody humanness and stability by systematic energy-based ranking

Area of Science:

• Biotechnology
• Immunology
• Computational Biology

Background:

• Conventional antibody humanization grafts animal complementarity-determining regions (CDRs) onto human frameworks.
• This process often reduces antibody stability and antigen-binding affinity, necessitating extensive optimization.

Purpose of the Study:

• To develop a computational method for systematic humanization of animal antibodies.
• To overcome the limitations of reduced stability and affinity in traditional antibody humanization techniques.

Main Methods:

• Developed CUMAb (computational human antibody design), a method utilizing Rosetta atomistic simulations.
• Systematically grafted animal CDRs onto thousands of human framework regions, selecting designs based on energy and structural integrity.
• Applied the method to humanize five distinct antibodies.

Main Results:

• CUMAb-designed humanized antibodies retained similar antigen-binding affinities compared to parental animal antibodies.
• Several designs demonstrated improved antibody stability.
• Non-homologous human frameworks were frequently selected over high-homology frameworks.
• Functionally equivalent humanized antibodies were achieved with numerous mutations and diverse human frameworks.

Conclusions:

• The CUMAb method provides an effective computational strategy for antibody humanization.
• This approach successfully preserves or enhances antibody function and stability.
• CUMAb facilitates the design of humanized antibodies using non-homologous frameworks and diverse mutations.