Lipid and protein contributions to viral proliferation as studied using MD simulations and advanced MD techniques

Jasmine M. Gardner

doi:10.17918/bzcz-af47

HIV infection is a continuing public health crisis with more than 36 million people living with HIV/AIDS and about 1.8 million people newly infected in 2017. Continuous research efforts have given rise to effective treatments to maintain undetectable viral loads for the duration of treatment, but a cure for AIDS has not been determined. The current thesis examines two independent functional processes within the HIV lifecycle using novel computational approaches to further elucidate the mechanisms of viral replication. Membrane fusion is the first step in viral infection, allowing transport of viral contents to the cytoplasm. Due to its necessity within the viral lifecycle and applicability to other biological processes, the fusion process harbors significant information which may help fight various human ailments. Membrane fusion is not understood in molecular detail, but it is thought to proceed through the stalk-pore mechanism with an intermediate hemifusion diaphragm (HD) structure. The HD is a rearrangement of two independent lipid bilayers into a new structure consisting of a three independent bilayers: the two merging bilayers and the HD bilayer consisting of the distal monolayers of the merging bilayers. The rearrangement introduces a line defect surrounding the HD bilayer where the three monolayers involved in the fusion process intersect called the three-junction. Biological bilayers include many types of lipids and proteins with unique functions. In this thesis, we focus on the functionality of negatively curved lipids and the availability of transmembrane lipid transport mechanisms known as flip-flop to stabilize the three-junction region and, by extension, stabilize the HD intermediate structure. The lipid effects on three-junction stability were studied through the use of a highly coarse-grained lipid model, the Cooke Model, which represents lipids as three sequential beads, one head bead followed by two body beads. Negative intrinsic curvature may be incorporated by reducing the size of the head bead and maintaining the size of the body beads thus producing a conical shape. In the first set of studies, the three-junction line defect was removed from the HD interface in order to better understand the effects of this defect on the larger system. Mixed-membrane bilayers were placed on a honey-comb lattice in a NP_[xx]P_[yy]L_zT environment to allow bilayers to relax but maintain tension and structure of the three-junction. The study was performed with varying degrees of negatively curved lipids at increasing compositions to sweep the available membrane curvature landscape. The results showed that a specific range of membrane curvature was required to stabilize the three-junction without destabilizing the membrane. Without adequate negative membrane curvature, the three-junction ruptured to produce a free-edge and opposing flat bilayer. Both the free-edge and three-junction structures harbor a line tension and the stable structure occurs due to a competition between the free energy of these two systems at a certain membrane curvature. The negatively curved lipids sorted according to the curvature of the three-junction region with depletion occurring when a positively curved free-edge exists and enrichment in the region when the negatively-curved intact three-junction exists. An equation was derived from simple thermodynamic arguments that accurately described the relationship between membrane curvature and three-junction stability reinforcing the dependence. The results of the three-junction study steered further studies into the stability of the three-junction surrounding the HD bilayer. In-situ HD bilayers of 20 nm in radius surrounded by a tensionless bilayer were simulated using the Cooke Model and varying compositions and degrees of negatively curved lipids. The stability of the HD was found to directly correlate to the composition of negatively curved lipids in the system and the stability of the three-junction. The two independent forms of the three-junction structure, intact and ruptured free-edge, were visualized in the HD systems with similar reliance on membrane curvature as found in the three-junction studies. The HD systems relax to three distinct end results. HD's with the least negative membrane curvature underwent a fission event and ended in two defect free bilayers with no indication of the previous HD or three-junction. These systems included ruptured three-junctions with interchanged free-edge and flat bilayer states with almost no intact three-junction as indicated by visual inspection and the depletion of negatively curved lipids in the region. HD's with moderate negative curvature dissipated the entire HD bilayer but formed a fusion pore connecting the two opposing chambers of the merging bilayers. The rate of dissipation when a fusion pore formed was much slower than when a double bilayer was formed and dissipation occurred through an intact three-junction surrounding the HD bilayer as indicated by significant sorting towards the region during dissipation. HD's with high-moderate membrane curvature relaxed to a stable, long-lived HD structure. When accounting for the intermediate state of the three-junction during dissipation, the HD end-states represent two distinct systems: non-fusogenic systems ending in the double bilayer and proceeding through the free-edge three-junction state and fusogenic systems ending in the fusion pore or stable HD and proceeding through the intact three-junction state. When lipid movement is studied in these two structures, evidence of two competing dissipation mechanisms arose during HD relaxation. In order for the HD to decrease in area, lipids must move from the distal monolayers to the proximal monolayers, a process typically reserved for transmembrane flip-flop movement. The mechanism for the fusogenic, intact three-junction systems proceeds through the flip-flop mechanism as discovered through rates dependent on membrane curvature similar to the flip-flop rates of Cookie lipids and inspection of the structure. For non-fusogenic systems, the dissipation was mostly independent of the composition of negatively curved lipids similar to the lateral diffusive movement of Cooke lipids. The free-edge system produces a flat bilayer connecting the distal and proximal monolayers from the opposing bilayers and thus allows for HD relaxation through a diffusion-based mechanism. Addressing a second important aspect of the viral lifecycle, the thesis continues to study the effect of large scale conformational changes on the binding of a ligand fragment to HIV-1 protease (PR). HIV-1 protease (PR) is a functional protein within the HIV virus. When HIV buds from an infected cell, it does so in the immature state. This state holds the integral proteins as one connected poly-protein. PR is the enzyme which cleaves the poly-protein into smaller functional proteins. Due to it's necessity for viral maturation, PR is a commonly targeted protein for drug development and has been extensively studied using experimental approaches and MD simulations. The current work uses an all-atom, explicit-water MD simulations using the PDB entry 1F7A, a closed-form PR bound with a ligand representative of the GAG poly-protein cleavage site to study the conformational changes present during ligand unbinding. The system was studied through long MD simulations in an NPT ensemble which revealed the closed and bound form of protease is long-lived. Ligand unbinding was studied using Temperature Accelerated MD (TAMD) simulations, a technique which increases the likelihood of rare events through a high-temperature biasing of a slow, auxiliary collective variable (CV) that is adiabatically separated from the atomic system. The slow moving variables in the PR binding process were determined to be (1) flap opening and (2) ligand diffusion and binding. Two unique CV sets are studied to determine the most energetically costly step of flap opening, flap unbinding or conformational opening of the flaps. TAMD simulations revealed (1) the most energetically costly mechanism of flap opening is the unbinding of the flap tips, (2) flaps must open for ligand unbinding to occur, (3) the system does not show preference towards wide flap opening, (4) the system does not show a preference towards threaded or lateral ligand movement, and (5) the system does not show preference towards asymmetric or symmetric flap opening. A hydrophobic pocket distal to the binding pocket is discovered which appears to have significant effects on the preparation of the protein for ligand binding. The distal pocket is found to be related to the engagement of a previously-noted elbow region and remains open in unbound proteins but appears closed in proteins prepared for ligand binding. String method in collective variables (SMCV) was used to study the free energy barriers associated with the conformational change to the wide open flap state and preparing the pocket for ligand entry. SMCV is an advanced MD technique that uses multiple replica simulations to minimize the free energy pathway of an event and uncover the most probable mechanistic pathway. In this case, PR binding mechanism is characterized using trajectories from representative TAMD simulations through two protein movements believed to be crucial for accessibility of the native ligand to the binding site: wide flap opening and distal pocket closing. A correlation is uncovered between releasing contact between the crucial ILE-50 residues at the flap turn and the distal pocket closing. The distal hydrophobic pocket appears to stabilize the wide open conformation by recovering the energy necessary to disengage the flap tips. A crucial residue in the distal pocket, PHE99, is mutated to TYR, a similar but hydrophobic residue. When TAMD is performed on the mutant, the elbow region is resistant to engage and distal pockets remain open. The ligand appears to be unhappy in the binding pocket, quickly exiting. Despite the qualitative change in behavior, when SMCV is performed on the mutant, the resulting free energy profile is very similar to WT. These findings indicate that the energetic costs associated with opening the flaps is much larger than that harbored by the hydrophobic interactions in the distal pockets or the ILE-50 residues.

Lipid and protein contributions to viral proliferation as studied using MD simulations and advanced MD techniques

Files and links (1)

Abstract

Metrics

Details

Lipid and protein contributions to viral proliferation as studied using MD simulations and advanced MD techniques

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media