Superinfection provides insights into HIV-1 reservoir dynamics

From the Overbaugh/Lehman lab, Human Biology Division.

Despite the success of antiretroviral therapy (ART) to treat HIV infection, a cure that eliminates all of the virus from the patient’s body remains elusive. While ART can suppress viral replication to undetectable levels, it does not eliminate the HIV reservoir, a subset of long-lived resting T-cells that carry a stably integrated copy of the viral genome. Upon ART interruption, these latent cells can reactivate and resume virus production. Therefore, achieving an HIV cure requires controlling or eliminating the reservoir. Essential to develop strategies to eliminate the HIV reservoir is the characterization of the reservoir dynamics (i.e. seeding timing and turnover rate) to target these long-lived HIV infected cells during their establishment. It is well-documented that the HIV reservoir is established early during infection; however, there is debate about reservoir composition and turnover rate as the infection progresses in non-suppressed individuals.

Multiple studies have addressed this question by following HIV variants that emerge after ART interruption. Due to the suppressive nature of ART, the HIV reservoir remains “frozen in time” after therapy is initiated. Thus, the rebound viruses reflect the genetic composition of the reservoir before ART initiation. These studies suggest two possibilities: the reservoir is seeded primarily during early infection or continuously as cells turnover throughout untreated infection. To address this question, the Overbaugh/Lehman lab (Human Biology Division) retrospectively studied a cohort of Kenyan women that were found to have been sequentially infected with two genetically distinct HIV variants in two different occasions, an event known as “superinfection”.


Phylogenetic tree showing genetic distance between HIV sequences from longitudinal samples obtained prior ART initiation. Black arrows indicate HIV DNA sequences most related to RNA from a certain time point.
Phylogeny of HIV sequences obtained from longitudinal samples shows reservoir sequences are most related to time points close to antiretroviral therapy initiation. Image provided by Dr. Dara Lehman

In contrast to other studies that track the evolution of a single HIV variant to establish seeding dynamics, superinfection provides a “higher resolution” into reservoir turnover and seeding by following two distinct viral strains that have determined temporal differences. The group recently published their findings in PloS Pathogens. Dr. Dara Lehman, a senior staff scientist leading her own NIH-funded projects within the Overbaugh lab and the corresponding author of the study, summarized her findings in the context of the field: “early work by other groups had shown that a reservoir of HIV-infected cells that persist during long term treatment is established very early in infection and suggested it may predominately be seeded during acute infection when HIV levels are highest. However, our group has now shown, based on detailed sequencing in women that have been “superinfected” with 2 genetically distinct viruses at different times during infection, that the majority of HIV sequences instead enter the reservoir near the time of initiation of antiretroviral therapy.”

The superinfection cohort had 6 cases that remained untreated for >5 years and subsequently started ART for >6 months (note: this cohort was initiated before ART was available in this setting). The researchers used next-generation sequencing to obtain sequences from viruses in plasma samples taken at 0-2 months post-infection, every two years until ART initiation, and a one-time PBMC sample taken after >0.9 years of suppressive ART. The authors then classified the sequences into three categories: lineages that established initial infection (initial), superinfection lineages (superinfecting), and lineages that emerged as both viruses combined (recombinant variants). The authors found that prior to ART initiation, the relative proportions of lineages in plasma samples varied over time and that, for most cases, all three lineages are present in the reservoir of individuals under suppressive ART. The authors concluded that, because the reservoir was composed of initial and superinfecting variants, the reservoir is not only seeded during initial infection but continuously seeded during untreated infection.

The researchers then sought to determine which reservoir seeding pattern could explain the observed composition of the reservoir by comparing the observed data to a “cumulative” model (viruses enter the reservoir continuously throughout infection in proportion to their abundance in plasma) and “last time point” model (the reservoir reflects the composition of the viruses circulating just prior ART initiation). They found that the observed data fit the “last time point” model better suggesting a seeding pattern corresponding to a high turnover rate of the reservoir during untreated infection.  Dr. Daniel Reeves, an Associate in VIDD who co-authored the study, explained the implications of this remarkable finding: “before meeting with Dara and seeing her data, I, like many others in the field, would have thought the reservoir was made really early after infection and then drifted along with minor additions over time. We all thought that reservoir cells were the same before and after ART. What she and her group have shown, and what I was lucky to get to help model, was that this is not true. The reservoir actually is predominantly created around the time of initiation of ART. This is an unfortunate paradox because our best weapon against active HIV seems to help its latent form.”

To estimate the seeding time of each reservoir sequence, the group used pairwise analysis of individual sequences from viruses from the untreated infection and the reservoir. The authors found that in a phylogenetic tree, the majority of reservoir sequences branch more closely to sequences from viruses circulating just before the start of ART, suggesting that the reservoir sequences are more closely related to viral sequences at the last time point before ART. When the investigators combined the data from all individuals, they discovered that the maximum genetic distance between those closely related sequences is within two years.

Finally, to formally test if the composition of the reservoir reflects the dynamics of viral lineages overtime or the proportion of viral lineages just before ART, the authors created a model for reservoir establishment based on a scenario of 10 years of untreated infection before ART. In this model, the sequences enter the reservoir in proportion to typical viral load patterns (i.e. peak viremia followed by a stable setpoint viral load). The authors observed that, compared to the observed data, this model overestimated the portion of the reservoir created during viremia. To test how different decay rates would impact the reservoir composition in this model, the authors applied two reservoir half-life estimates from previous studies. Interestingly, even with a shorter half-life, the model underestimates the sequences that enter the reservoir during the final two years before ART. The authors estimate the reservoir half-life to 25 months –shorter than the two previously reported measurements. Dr. Lehman highlighted the implications of her findings: “these data support studies from other groups - and together suggest that HIV-infected cells turn over rapidly during untreated infection, and that suppressive treatment creates an environment that favors the persistence of these infected cells. Thus, novel strategies to target the HIV reservoir should focus on the time just prior to ART initiation.”

This work was supported by grants from the National Institutes of Health.

Pankau MD, Reeves DB, Harkins E, Ronen K, Jaoko W, et al. (2020) Dynamics of HIV DNA reservoir seeding in a cohort of superinfected Kenyan women. PLOS Pathogens

Fred Hutch/UW Cancer Consortium member Dr. Julie Overbaugh participated in this research.