Lack of evidence for prolonged double-long terminal repeat episomal HIV DNA stability in vivo.

To the Editor: Low-level HIV replication seems to persist during long-term effective highly active antiretroviral therapy (HAART). The sites of this replication are unclear, and the resulting viremia would be too low to detect with conventional procedures. It has been proposed that latently infected resting CD4 T cells release the virus on activation or that viral replication persists in anatomic sanctuaries poorly accessible to antiretroviral drugs. Double-long terminal repeat (LTR) episomal HIV DNA arises when reverse-transcribed viral DNA fails to integrate the host genome and is circularized. It has been suggested that these circular DNA forms might serve as a marker of recent cell infection on the basis of their observed lability in vitro and in vivo. Other in vitro studies have suggested that these DNA circles are highly stable and may persist indefinitely should the infected cells survive and remain in the compartments being sampled, however. This implies that HIV double-LTR DNA circles found in patients on long-term effective HAART could originate mostly from the preHAART period of high-level replication. In an attempt to settle this issue, we compared sequences of the hypervariable gp120 V3 loop in plasma HIV RNA at the outset of first-line HAARTwith those of peripheral blood mononuclear cell (PBMC) episomal HIV DNA during long-term effective HAART (Fig. 1). We selected 3 patients who were highly adherent to HAART and who had plasma HIV-1 RNA <20 copies/mL (Amplicor Ultrasensible; Roche Diagnostics, Meylan, France) for 2 to 5 years without a single detected viral blip. Episomal DNA was isolated from PBMC genomic DNA (QIAprep Spin Mini prep Kit; Qiagen, Hilden, Germany), and polymerase chain reaction (PCR) assay was performed with a primer pair targeting the 2-LTR circle junctions (5#-CAGATCTGGTCT AACCAGAGA-3# and 5#-GTAACTAG AGATCCCTCAGAC-3#). The corresponding PCR products were cloned and sequenced. The sequences of the 2LTR circle junction were determined in each patient and then used to design 1 or 2 specific primers, which were pooled for use. Episomal DNA was then PCR amplified with the 2-LTR junction primer(s) and the Env13 primer (5#CCACTCTATTTTGTGCATCAGA-3#) located in the Env gene. The sequence amplified included the C2 to V4 regions of gp120. Nested PCR was applied to the resulting purified PCR product by using a pair of primers located in the V3 loop of gp120 (B1: 5#-ACACATGGA ATTAGGCCAGT-3# and B2: 5#-CTGC CACATGTTTATAATTTG-3#). RNA was extracted from frozen plasma obtained 3 to 4 weeks before the outset of first-line HAART (Amplicor HCV Lys.V2; Roche Diagnostics); it was then reverse transcribed, and the V3 loop was amplified as described previously for episomal DNA with primers B1 and B2 (Titan One Step; Roche Diagnostics). The corresponding PCR products were purified, cloned, sequenced, and subjected to phylogenic analysis as described elsewhere. As shown in Figure 1, the sequences of double-LTR HIV DNA isolated during long-term effective HAART showed limited heterogeneity, arguing against a long-term archiving process, which requires high stability, as has been shown for the latent resting CD4 T-cell reservoir. Diversity of episomal DNA sequences was lower compared with virus sequences from pre-HAART plasma, as shown by the mean genetic distance analysis (patient A: 0.23% ± 0.1 vs. 0.9% ± 0.5, patient B: 2.3% ± 1.6 vs. 5% ± 1.5, and patient C: 3.02% ± 1.6 vs. 4.3% ± 1.2, respectively). The preHAART plasma virus sequences and the double-LTR episomal sequences obtained during long-term effective HAART showed low heterogeneity in patient A, who started HAART early at a CD4 count of 550 cells/mm (compared with 300 and 221 cells/mm in patients B and C, respectively). It is also noteworthy that patient A took quadrupledrug therapy, whereas patients B and C took triple-drug therapy (see Fig. 1). More importantly, in the 3 patients, the episomal sequences were clearly distinct from the pre-HAART plasma virus sequences, indicating that episomal HIV DNA found during long-term optimal HAART does not result from archiving of pre-HAART replicating viruses. The studies suggesting that episomal HIV DNA was highly stable were based on quantitative monitoring of episomal HIV DNA after cell infection in vitro, and this may not accurately reflect conditions prevailing in vivo. Some authors have also measured HIV episomal DNA in patients’ cells before and after HAART. HAART was found to have little impact on the level of double-LTR HIV DNA, but this could also be explained by continual replenishment attributable to residual virus replication. Here, by examining HIV sequences in vivo, we found that double-LTR episomal DNA sequences during long-term effective HAARTwere distinct from pre-HAART sequences. Env sequences of PBMC provirus have been reported to evolve despite effective HAART, a phenomenon the authors attributed to residual virus replication. This evolution might be better represented by HIV double-LTR circles than by total proviral DNA. Indeed, the latter includes integrated viruses from the latent reservoir, which we have shown to undergo long-term archiving. Together, our data argue against the longterm stability of HIV double-LTR circles. These viral forms therefore should remain under consideration as potential markers of residual virus replication during HAART.