HIV dynamics linked to memory CD4+ T cell homeostasis

Abstract

The dynamics of latent HIV is linked to infection and clearance of resting memory CD4+ T cells. Infection also resides within activated, non-dividing memory cells and can be impacted by antigen-driven and homeostatic proliferation despite suppressive antiretroviral therapy (ART). We investigated whether plasma viral level (pVL) and HIV DNA dynamics could be explained by HIV's impact on memory CD4+ T cell homeostasis. Median total, 2-LTR and integrated HIV DNA levels per μL of peripheral blood, for 8 primary (PHI) and 8 chronic HIV infected (CHI) individuals enrolled on a raltegravir (RAL) based regimen, exhibited greatest changes over the 1st year of ART. Dynamics slowed over the following 2 years so that total HIV DNA levels were equivalent to reported values for individuals after 10 years of ART. The mathematical model reproduced the multiphasic dynamics of pVL, and levels of total, 2-LTR and integrated HIV DNA in both PHI and CHI over 3 years of ART. Under these simulations, residual viremia originated from reactivated latently infected cells where most of these cells arose from clonal expansion within the resting phenotype. Since virion production from clonally expanded cells will not be affected by antiretroviral drugs, simulations of ART intensification had little impact on pVL. HIV DNA decay over the first year of ART followed the loss of activated memory cells (120 day half-life) while the 5.9 year half-life of total HIV DNA after this point mirrored the slower decay of resting memory cells. Simulations had difficulty reproducing the fast early HIV DNA dynamics, including 2-LTR levels peaking at week 12, and the later slow loss of total and 2-LTR HIV DNA, suggesting some ongoing infection. In summary, our modelling indicates that much of the dynamical behavior of HIV can be explained by its impact on memory CD4+ T cell homeostasis.

Fig 1. Median HIV DNA per mm³ changes over time, relative to commencement of ART, for PHI (circles) and CHI (diamonds).

Note the break in the time axis and the regression lines between days 1000 and 3000. Regression lines for each HIV DNA component over the first year are shown (solid lines in the same color as the markers), as well as the corresponding half-lives. Assay values of total HIV DNA/ mm³ are shown as black markers, 2-LTR HIV DNA/ mm³ as green markers, and integrated HIV DNA/ mm³ as red markers. The two lowest 2-LTR markers are due to some of the individual values being below the assay limit of detection as well as missing CD4+ T cell counts at these later time points. Only considering the 2-LTR HIV DNA values from year 1 for CHI (but omitting the low final value) and for the year 10 time point gives a half-life of decay of 3.4 years, considerably slower than estimates of memory cell division half-life of 22 weeks but consistent with the longer division half-life of naïve cells (3.5 years) [26]. Total HIV DNA from year 1 for CHI exhibited a 5.9 year half-life. Integrated HIV DNA increased for the CHI group over the first year leading to a negative half-life.

Fig 2. Uninfected Homeostasis: Mechanisms controlling resting non-dividing R and dividing R⁺, activated non-dividing A and dividing A⁺ memory CD4+ T cells numbers.

Dividing cells R⁺, A⁺ arise at rates λ_R, λ_A and can go through several rounds of proliferation giving rise to θ_R and θ_A copies, which can then revert to non-dividing states at rates ρ_R, ρ_A. Activated non-dividing cells arise from activated naïve cells at rate s_A and from activation of resting, non-dividing cells R at rate β. Activated non-dividing cells revert to a resting state at rate α. Resting and activated cells die at rates μ_R, μ_A. HIV DNA infection of memory CD4+ T cells: The model duplicates the uninfected network for each of the cells containing linear, episomal and integrated HIV DNA. Modifications to the uninfected network to describe dynamics of HIV are highlighted in black. Infection of the uninfected subsets leads initially to cells with linear HIV DNA (subscript L), then either episomal (subscript C) or integrated HIV DNA (subscript I). All parameters are the same across each of these networks. The additional parameters relate to the loss of linear and episomal HIV DNA within cells at rates μ_L, μ_C respectively, progression of linear at rate ν_A, ν_R in activated and resting cells respectively which convert to episomal HIV DNA with probability ϕ or integrated HIV DNA with probability 1- ϕ, a higher death rate of productively infected cells (A_I⁺) at rate μ_AI, and the rate of production of virions from these cells N per day. We also assume dividing resting cells with integrated HIV DNA can produce virions at rate N_R and die at rate μ_RI.

Fig 3

Data (means/medians over the individuals within the PHI and CHI groups, S2 Table) and model simulations for: A) and C) PHI, B) and D) CHI. Vertical lines at data points denote ranges within each group. For A) and B) total memory CD4+ T cells/mm³, linear HIV DNA/mm³, and 2-LTR HIV DNA/mm³ are shown as black lines while components within each cell phenotype are shown as colored lines. Cells/mm³ are described on a linear scale while all other panels are shown with a logarithmic y-scale. The colored lines in each of the cellular and HIV DNA panels represent values in the resting/activated, dividing/nondividing subsets as depicted by the legend in the linear HIV DNA panel. The colored lines in the integrated HIV DNA panels depict the replication competent subsets. The phenotypes are resting (R), resting and dividing (R⁺), activated (A), and activated and dividing (A⁺). In the Total HIV DNA panels each of the resting phenotypes are combined (Rtot) as well as each activated phenotype (Atot). Panels C) and D) show mean log₁₀ HIV RNA copies/ml for PHI and CHI over the first 200 days—dashed lines denote the infectious components of pVL.

Fig 4

pVL simulations with ART (A and C), and treatment intensification after 3 years of ART (B and D). Panels A and B describe contributions to pVL for ART commenced at PHI, while C and D describe pVL commenced at CHI. pVL: total (black), arising from cells directly infected by virus (magenta), and from infected cells that were produced via homeostatic proliferation (through processes within the R⁺ component) from cells with integrated HIV DNA (green). In the treatment intensification panels B) and D), time has been reset to commencement of 12 weeks ART intensification (an integrase inhibitor with efficacy 99.9% is assumed added to a nonRAL regimen which had the same overall efficacy as in Fig 3 simulations), followed by reversion to the original ART regimen at day 84. E), F), G) Rebound of pVL after stopping 6 years of ART. Here we track the origin of the resulting viremia and cellular infection both in terms of whether the infected cell arose through clonal expansion or not, and whether the cells are in the activated dividing or resting dividing infected phenotype. E) Percentage of productively infected cells (A _I⁺) of all virus producing cells (A _I⁺ + R _I⁺) separated into clonal (green) or direct infection (magenta) origin. Prior to interruption almost all cells producing virus are productively infected if they originally arose from direct infection, whereas if the cells arose from clonal expansion they are mostly in the resting dividing subset (R _I⁺). Soon after interruption the majority of cells are productively infected regardless of their original infection type. F) Source of rebounding viremia by original infection type–this also tracks subsequent generations of new infection relative to these subsets. G) Expansion of virus producing cells (A _I⁺, R _I⁺) relative to their source of infection–as depicted in panel E) infection is mostly in the productive infected subset (A _I⁺) regardless of the original source of infection.