Persistence of episomal HIV-1 infection intermediates in patients on highly active anti-retroviral therapy

After HIV-1 infection of the cell, HIV-1 cDNA synthesis is initiated in the cytoplasm and proceeds at the same time as viral cDNA translocation to the host cell nucleus (Fig. 1a). Within the nucleus, some linear full-length cDNA genomes circularize to form episomes containing either one long-terminal repeat (1-LTR circle) or two long-terminal repeats (2-LTR circle) (Fig. 1a). As HIV-1 lacks necessary factors for the maintenance of episomal replication, these episomes may be labile and thus, indicative of recent infection events. To specifically detect 2-LTR circles, we used a PCR approach⁷ in which 5′ and 3′ primers direct the amplification of viral sequences across the junction that is formed after end-to-end 5′ and 3′ LTR ligation (Fig. 1b). To facilitate the detection of rare viral episomes in a background of a large proportion of uninfected peripheral blood mononuclear cells (PBMCs), we adapted a plasmid DNA isolation procedure that allowed for the selective isolation of viral episomes from genomic cellular DNA (Fig. 1c). Viral cDNA amplified from patient PBMC DNA by 2-LTR circle-specific primers represented genuine 2-LTR circle junctions (Fig. 1d).

We initially assessed the stability of 2-LTR circle forms of viral DNA in acutely infected cells in vitro. We infected CD4⁺ MT-4 T cells and Jurkat-CCR5 cells with the X4 variant HIV-1_LAI and the R5 variant HIV-1_ADA, respectively. We allowed synthesis of viral cDNA to proceed for 24 hours and then restricted further rounds of virus infection and cDNA synthesis by the addition of reverse transcriptase inhibitors (Fig. 1). Within 24–48 hours after the addition of the reverse transcriptase inhibitors, 2-LTR circle numbers decreased by more than 90% in both HIV-1_LAI-and HIV-1_ADA-infected cells (Fig. 2a). The copy numbers of other viral DNA forms identified by the internal LTR primers (mostly linear and integrated viral genomes) remained relatively constant over the same interval (Fig. 2a). Thus, 2-LTR circles seem to be labile intermediates in the virus lifecycle.

We next evaluated whether 2-LTR circles were labile in vivo. We obtained PBMC samples from four HIV-1-infected individuals who, after adjustment of their anti-retroviral regimens to more potent combinations, showed steady decreases in plasma viral RNA levels. There were considerable decreases in 2-LTR circle copy numbers over the interval in which there was a rapid decrease in levels of plasma viral RNA (Fig. 2b). In contrast, when samples were analyzed in parallel with internal LTR primers (Fig. 1b), HIV-1 cDNA levels fluctuated by no more than 300% (Fig. 2b). These results indicate that 2-LTR circles are labile, both in vitro and in vivo, relative to integrated viral genomes.

We next determined whether 2-LTR HIV-1 episomes were present in 63 patients who, throughout treatment with highly active anti-retroviral therapy (HAART), had undetectable levels of plasma viral RNA for sustained periods of time (Tables 1 and 2). Of these 63 patients, 50 (80%) had undetectable levels of plasma viral RNA (assay limit of sensitivity, 400 copies/ml) for 12 months or longer (Table 2). Of these 50 patients, 24 (48%) had undetectable levels of plasma viral RNA for 12 months or more using an assay with sensitivity as low as 50 copies/ml. In 48 of the 63 patients (76%), 2-LTR circles were detected in their PBMCs (Table 2). 2-LTR circle copy numbers ranged from less than 1 copy per 1 x 10⁶ PBMCs to 620 copies per 1 × 10⁶ PBMCs. There did not seem to be any substantial relationship between the frequency of 2-LTR circles in patient PBMCs and the time during which plasma viral RNA was undetectable (Fig. 3). These data indicate that labile replication intermediates are present in a substantial proportion of HIV-1 infected individuals who show sustained suppression of plasma viral RNA while on HAART. 2-LTR circles were not detectable in PBMCs from 15 (24%) patients (Table 2). However, this may not indicate an arrest of ongoing virus replication, because our assay does not account for extrachromosomal linear and 1-LTR circle forms of viral cDNA.

To determine whether patients with 2-LTR circles also had cells with replication-competent virus, we did high-input viral co-culture assays on PBMCs from nine patients with 2-LTR circles and four patients without 2-LTR circles (Fig. 4). Replication-competent virus could readily be isolated from eight of the nine patients with 2-LTR circles. Virus could not be isolated from patient W1, who had a very low circle copy number. Infectious virus could not be isolated from three patients who lacked 2-LTR circles, even though co-culture was done on 4 × 10⁷–6 × 10⁷ CD8-depleted patient PBMCs. For patient L8, who also lacked 2-LTR circles, only one of three cultures yielded infectious virus (Fig. 4). These results indicate a correlation between the presence of 2-LTR circles and cells with replication-competent virus. Plasma-based viral RNA assays therefore fail to demonstrate the full extent of viral activity in infected individuals who are being treated with HAART.

One important variable with a big effect on the decay characteristics of viral compartments in patients on HAART is the possible contribution of a reservoir of ongoing virus replication. Such a reservoir would be predicted to provide a continuous source of infectious virus, which could sustain the pool of productively and latently infected cells. We have used a new approach to demonstrate ongoing virus replication in a substantial percentage of patients who have sustained undetectable levels of plasma viral RNA. Our observations support recent studies that presented evidence for ongoing virus replication in aviremic patients on the basis of viral sequence evolution^8,9 and the presence of HIV-1-specific transcripts¹⁰.

Several characteristics of the reservoir in which virus replication continues in patients on HAART can be inferred from our analyses. 2-LTR circles are formed only after completion of viral cDNA synthesis and translocation of the viral cDNA to the host cell nucleus^11,12, and both processes occur only after infection of cycling but not resting T cells^13,14 or peripheral blood monocytes¹⁵. These extrachromosomal episomes in the host cell nucleus are distinct from the extrachromosomal cytoplasmic cDNA intermediates that result from the restricted infection of resting CD4⁺ T cells^12-14. CD8⁺ T cells contain proviral DNA in AIDS patients¹⁶. We have made a preliminary analysis of purified cells in two patients with readily detectable 2-LTR circles (patients L7 and L19). 2-LTR circles were detected only in purified CD4⁺ lymphocytes and not in CD4⁻ cells. Furthermore, 2-LTR circles were detected in both CD4⁺/RO⁺ and CD4⁺/RO⁻ subpopulations (data not shown), which is in agreement with studies indicating that both naive and memory T cells are susceptible to HIV-1 infection in vitro^17-19. One question that emerges from our analysis is the nature of the reservoir in which virus replication is ongoing. Although 2-LTR circles were detected in patient PBMCs, it is unlikely that this compartment was the source of the infectious virus. Rather, CD4⁺ T cells may have been infected during their passage through extravascular tissues. At present, we can only speculate on the tissue source of infectious virus. For example, ‘sanctuary sites’ may exist in some or all tissues where virus replication can occur in an environment that is relatively inaccessible to anti-retroviral agents.

Our study has implications for the development of strategies to eradicate virus replication in HIV-1-infected individuals. Although complete elimination of HIV-1 replication may be difficult with current anti-retroviral regimens, our study, along with other published observations^9,20, indicates that there do seem to be examples in which even the most sensitive assays fail to demonstrate ongoing replication in some well-suppressed patients^9,20. It is also likely that, as more potent anti-retroviral therapeutics enter the clinic, ongoing or ‘covert’ virus replication may be arrested in more patients. A better understanding of the nature of the reservoir that sustains virus replication in aviremic patients on HAART may lead to the development of more effective strategies for the arrest of virus replication.