Decoding epitope immunodominance in HIV Env using cryoEM and machine learning.

Area of Science:

• Structural biology
• Immunology
• Vaccine design

Background:

• Viral surface glycoproteins like HIV Env have many antibody epitopes, but immune responses often target only a few (immunodominance).
• Understanding the molecular basis of immunodominance is crucial for designing effective vaccines but remains incomplete.

Purpose of the Study:

• To characterize the structural determinants of epitope immunodominance in HIV Env.
• To develop a predictive model for identifying immunodominant regions on Env.

Main Methods:

• Integrated high-resolution cryo-electron microscopy (cryoEM) epitope mapping across HIV clades.
• Analyzed epitope topology, accessibility, and physicochemical properties.
• Developed a machine-learning model (Antigen Surface Immunodominance - ASI model) to predict immunodominance.

Main Results:

• Resolved over 70 new structures, creating a library of >100 Env-antibody complexes.
• The ASI model accurately predicted immunodominant regions and identified key structural features driving immune bias.
• Redesigned immunogens based on model predictions successfully redirected antibody responses to subdominant epitopes.

Conclusions:

• Structural insights combined with machine learning can uncover complex patterns governing immune responses.
• The ASI model provides a framework for predicting immunodominance and designing targeted vaccine antigens.
• This approach holds promise for developing more effective HIV vaccines by focusing immune responses.