Date of Award

12-18-2014

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Biology

First Advisor

Irene T. Weber

Second Advisor

Robert W. Harrison

Third Advisor

John E. Houghton

Abstract

HIV-1 protease (PR) is an effective target protein for drugs in anti-retroviral therapy (ART). Using PR inhibitors (PIs) in clinical therapy successfully reduces mortality of HIV infected patients. However, drug resistant variants are selected in AIDS patients because of the fast evolution of the viral genome. Structural, kinetic and MD simulations of PR variants with or without substrate or PIs were used to better understand the molecular basis of drug resistance. Information obtained from these extensive studies will benefit the design of more effective inhibitor in ART.

Amprenavir (APV) inhibition of PRWT, and single mutants of PRV32I, PRI50V, PRI54M, PRI54V, PRI84V and PRL90M were studied and X-ray crystal structures of PR variants complexes with APV were solved at resolutions of 1.02-1.85 Å to identify structural alterations. Crystal structures of PRWT, PRV32I and PRI47V were solved at resolutions of 1.20-1.40 Å. Reaction intermediates were captured in the substrate binding cavity, which represent three consecutive steps in the catalytic reaction of HIV PR. HIV-1 PR20 variant is a multi-drug resistant variant from a clinical isolate and it is of utility to investigate the mechanisms of resistance. The crystal structures of PR20 with inactivating mutation D25N have been determined at 1.45-1.75 Å resolution, and three distinct flap conformations, open, twisted and tucked, were observed. These studies help understand molecular basis of drug resistance and provide clues for design of inhibitors to combat multi-drug resistant PR.

The evaluation of electrostatic force in MD simulations is the computationally intensive work, which is of order theta(N2) with integration of all atom pairs. AMMP invokes Amortized FMM in summation of electrostatic force, which reduced work load to theta(N). A hybrid, CPU and GPU, parallel implementation of Amortized FMM was developed and improves the elapsed time of MD simulation 20 fold faster than CPU based parallelization.

Share

COinS