Improved prediction of HIV‐1 coreceptor usage with sequence information from the second hypervariable loop of gp120

Area of Science:

• Virology
• Molecular Biology
• Computational Biology

Background:

• Human immunodeficiency virus type 1 (HIV-1) entry relies on CD4 and coreceptors, primarily dictated by the gp120 V3 loop.
• Current coreceptor usage prediction relies solely on V3 genotyping, potentially missing crucial information.
• Mutations outside the V3 loop are known to influence HIV-1 coreceptor tropism.

Purpose of the Study:

• To investigate the impact of the V2 loop on HIV-1 coreceptor usage prediction accuracy.
• To determine if incorporating V2 loop features improves upon V3-only prediction models.

Main Methods:

• Analysis of sequence differences and position-independent features in V2 and V3 loops.
• Training and cross-validation of prediction models using support vector machines on clonal and clinical data.
• Statistical comparison of prediction model performance using V2, V3, or both loops.

Main Results:

• Statistically significant differences in V2 loop mutations and features between R5 and X4 HIV-1 viruses were identified.
• Models incorporating both V2 and V3 loop features demonstrated significantly higher predictive accuracy (AUC) than V3-only or V2-only models.
• Improved prediction accuracy was consistent across both in vitro clonal data and ex vivo clinical data sets.

Conclusions:

• The V2 loop provides valuable information for enhancing HIV-1 coreceptor usage prediction.
• Integrating V2 loop data significantly improves the accuracy of predicting viral tropism.
• This approach offers a more robust method for predicting coreceptor usage in both research and clinical settings.