Although HIV can be treated by HAART, long-lived reservoirs prevent eradication. Thus, HIV can persist as latent provirus in cells that do not produce replicating virus. While laborious methods exist to investigate long- lived cells that may harbor HIV, little is known about how frequently HIV becomes latent, and how latency periods may affect HIV diversity, maintenance and adaptation potential. Our hypothesis is based on the fact that HIV accumulates mutations while replicating, but not when latent. The result will be that some virus lineages will have evolved less than others. We can reconstruct the entire evolutionary history of all sampled lineages using phylogenetic inference methods and estimate the different evolutionary rates in that history.
The specific aims of this proposal are 1) To develop a phylogenetic framework to identify and quantify real latency in clinical data, and 2) To develop a latency model for prediction and simulation of the genetic repertoire. We will investigate 1) in silico simulated data to assess model limitations, 2) reanalyze meta-data from many published studies where plasma and reservoir sequences are available to test our method on real data as well as critically reexamine existing data, and 3) analyze new data collected from serially sampled patients before and during treatment both in plasma (when retrievable) and reservoirs using ultra-deep and limited dilution sequencing. One strength of our phylogenetic approach is that it does not depend on sequencing of cells that have been sorted by markers or specialized sampling of potential reservoir tissues. Thus, in addition to investigating reservoir virus sequences, we will also be able to measure the effect of latency in the plasma virus population if it was reactivated at any point in the patient's infection history. e show that a preliminary method, which avoids statistical issues stemming from common evolution, is able to detect latency in simulated data. Compartmental modeling systems, like ODE models of viral dynamics, typically cannot include mutation processes and simulate sequence evolution. Although other systems, like agent-based models can do this in principle, it has not been done in this context. Thus, we will develop two new frameworks i) detection/quantification of real latency in clinical sequence data using a novel molecular clock phylogenetic approach, and ii) simulation and prediction of latent reservoirs in different scenarios using a novel dynamic model that includes sequence evolution. Integrating the efforts from aims 1 and 2 will allow us to create a bioinformatic tool that can take HIV sequence data from a patient, detect and quantify effective latency, then feed those results to the latency model to project how that specific HIV population would evolve under different treatment scenarios.
This proposed project will develop a novel computational detection-quantification-prediction method to investigate HIV latency. We envision that such a tool will be useful to scientists and clinicians investigating latency and similar systems. The results will provide us with better understanding of latent HIV reservoirs and the clinical application may guide future treatment protocols.