Recovery of Replication-Competent HIV Despite Prolonged Suppression of Plasma Viremia

See allHide authors and affiliations

Science  14 Nov 1997:
Vol. 278, Issue 5341, pp. 1291-1295
DOI: 10.1126/science.278.5341.1291


In evaluating current combination drug regimens for treatment of human immunodeficiency virus (HIV) disease, it is important to determine the existence of viral reservoirs. After depletion of CD8 cells from the peripheral blood mononuclear cells (PBMCs) of both patients and normal donors, activation of patient CD4 lymphocytes with immobilized antibodies to CD3 and CD28 enabled the isolation of virus from PBMCs of six patients despite the suppression of their plasma HIV RNA to fewer than 50 copies per milliliter for up to 2 years. Partial sequencing of HIV pol revealed no new drug resistance mutations or discernible evolution, providing evidence for viral latency rather than drug failure.

Treatment with potent antiretroviral regimens can produce sustained suppression of HIV-1 replication, with reduction of HIV RNA in infected individuals to below the limits of detection in blood for 2 years or more (1). Although viral DNA remains detectable in PBMCs as well as in lymphoid tissues, several groups have reported that virus cannot be recovered from PBMCs or lymphoid tissue samples from these patients by standard coculture techniques (2, 3). The inability to isolate infectious HIV from these patients has led to speculation that the residual proviral DNA represents defective viral genomes incapable of replication and that HIV may be at the point of eradication (3-5). This possibility is supported by studies of patients treated early after infection who showed reductions in virus-specific antibody titers after prolonged therapy with potent antiretroviral regimens, suggesting elimination of viral antigen production (3, 6). Moreover, defective genomes are found in a high proportion of proviral DNA from PBMCs of chronically infected patients (without viral suppression) (7).

Estimates of 2 to 3 years have been made for a theoretical time to eradication of HIV infection (4), but Perelson et al. stated that such projections assumed no third phase of decay or long-lived, latently infected cell population (4). Recently, Chun et al. have shown, in treated patients with incomplete viral suppression, that integrated proviral DNA resides in resting memory CD4 cells (CD45RO+, HLA DR) and that such cells are capable of producing replication-competent HIV upon appropriate activation (8).

Here, we assessed the replication capacity of HIV-1 provirus persisting in the PBMCs of patients with complete and sustained suppression of plasma viremia by attempting recovery of virus with both standard and enhanced coculture conditions. The six patients examined were enrolled in the San Diego cohort of the Merck 035 study (1). All patients received the triple drug regimen of zidovudine (AZT), lamivudine (3TC), and indinavir (IDV) and were selected for this study on the basis of reductions of HIV RNA in blood to <50 copies/ml for 52 to 92 weeks (mean, 74 weeks) (9). At the time of blood collection for our experiments, plasma samples were analyzed with the Roche Ultrasensitive HIV RNA assay (10). In one case, the HIV RNA was at the limit of detection of 50 copies/ml despite having been undetectable during the preceding 12 months, whereas in all other cases it was undetectable (arbitrary sensitivity of 50 copies/ml) (Table 1). Residual viral RNA and proviral DNA load in lymph node and blood had been assessed in patients A, B, and C after 1 year of treatment (2). In each case, lymph node viral RNA concentrations had been 4 log copies per gram of lymph node tissue lower than in treated patients with detectable plasma virus. However, proviral DNA had remained detectable in all PBMC and lymph node samples, even though virus isolation had been unsuccessful by standard coculture protocols at the 1-year time point. Furthermore, sequencing of the protease gene and the first 242 codons of the reverse transcriptase (RT) from both sources of residual RNA and DNA at the 1-year time point had demonstrated little or no change in viral sequences from study entry, which suggested that replication in vivo was suppressed below the extent necessary for discernible evolutionary changes (11).

Table 1

HIV RNA concentrations in plasma from study patients. Samples were first assayed with the Amplicor assay (Roche). For samples with values of <400 copies/ml, the Ultrasensitive assay (Roche) was used with a limit of detection of 50 copies/ml. Assay results corresponding to virus isolation time points are shown in bold (n.d., not done).

View this table:

In our study, the majority of CD8 T cells were removed from both the patient and healthy donor cell preparations before culture (12) to avoid the induction of soluble factors and chemokines suppressive to viral replication and alloreactive cytolytic T lymphocytes targeting dividing CD4 cells (13). The chosen activation conditions were based on observations by Spina et al. that immobilized antibody to CD3 and CD28 strongly activated CD4 lymphocytes, permitting more extensive HIV replication in CD45RO+ memory cells than did the standard activation stimuli of phytohemagglutinin (PHA) and recombinant interleukin-2 (rIL-2) (14). The patient PBMCs were so activated; however, other investigators have reported selective down-modulation of the CCR5 viral coreceptor on lymphocytes after stimulation with anti-CD3 and anti-CD28, but not with PHA mitogen (15). To provide permissive CD4 cells for spread and growth of both syncytium-inducing and non–syncytium-inducing HIV isolates generated from the activated patient cells, we prestimulated the healthy donor cells with PHA for 1 to 3 days before their addition to the coculture. Recombinant IL-2 was then added to the coculture and replenished at 3- to 4-day intervals to maintain optimal lymphocyte proliferation. In some replicate cultures, monocytes removed by fibronectin adhesion before CD8 cell depletion were added back to patient CD4 cells, or patient CD4 cells were pretreated with recombinant tumor necrosis factor–α (rTNF-α), to overcome potential barriers to viral transcription that might be attributable to abnormalities in the TAT-Tar axis (16).

Virus was recovered from all six patients by means of these enhanced coculture conditions, whereas standard coculturing techniques (17) (in which coculturing is done with PHA- and IL-2–stimulated donor PBMCs) resulted in virus isolation from only one patient (patient M, Table 2). In one case (patient B) virus was recovered on two separate occasions 2 months apart, with the enhanced conditions only. Viral RNA concentrations in culture supernatants during the first 21 days of coculture from patient B were measured after varying several components of the isolation conditions. CD8 depletion from both the patient and healthy control donor PBMCs was critical for successful virus isolation from these patients (18), whereas the addition of soluble TNF-α (R&D Systems, Minneapolis) or autologous macrophages may have accelerated virus recovery from some of the cultures (19). In those experiments where HIV RNA concentrations rose in culture supernatants (typically by day 7 to 14), HIV p24 antigen became detectable between 11 and 21 days (range of 36 to 10,630 pg/ml). With the exception of patient M, p24 antigen was not detectable in the standard cocultures carried out to 28 days. In patients A, B, L, and M, culture supernatants were passaged in donor PBMCs stimulated with PHA and IL-2 to produce high-titer viral stocks (p24 of 240,000 to 530,000 pg/ml). Similar passaging and expansion of viral isolates from the other two patients is in progress.

Table 2

Result of cultures by standard techniques or with the use of enhanced conditions, as described in the text. Positive cultures all had rising HIV RNA concentrations and positive p24 concentrations. Serial HIV RNA concentrations in culture supernatants are from a single representative condition out of the six enhanced conditions used. HIV RNA on supernatants from the standard cultures were all negative, with the exception of patient M (18). For all patients: PHA stimulated CD8-depleted HC and anti-CD3/anti-CD28–stimulated CD8-depleted patient cells. Patient A, + TNF; patient B1, + macrophages; patient B2, + macrophages and TNF; patient C, no addition; patient K, + macrophages; patient L, + macrophages and TNF; patient M, + macrophages. Passage codes: +, viral cultures have been successfully passaged to obtain high-titer stocks; ∗, experiments are still ongoing. B1 and B2 are the first and second culture from patient B's cells.

View this table:

Partial sequencing of HIV pol (20) was performed on virus from culture supernatants collected before peak p24 concentrations from all six patients, as well as in one patient after in vitro passage of the virus. Sequences obtained directly from baseline patient plasma collected 2 years earlier showed high sequence homology to the virus recovered in coculture (Fig.1) (20). The differences (mean 9/1000 base positions) between sequences from study baseline and after 2 years of treatment are minimal, in contrast to mean differences of 17/1000 base positions found in incompletely suppressed patients on triple therapy participating in the same study after only 8 to 12 months (11). For each patient, the sequences from the two time points resembled each other and did not resemble any laboratory strain; this finding excluded laboratory contamination to explain virus recovery, and relatively little viral evolution over the 2-year period was evident from analysis of the sequence data.

Figure 1

Inferred phylogenetic tree demonstrating the relation of sequences at baseline, of posttreatment virus isolates, and of laboratory strains of HIV-1. Letter designations correspond to sequences from patients A to C and K to M. Suffix codes: 0, sequences obtained directly from baseline plasma; 1, sequences from the first coculture supernatant; 2, sequences from the second coculture; and 3, sequences from the passaged, high-titer virus stock. LAI, HXB2, and NL4-3 refer to sequences of the respective laboratory strains of HIV-1. The tree was inferred by a maximum likelihood method (20). Branch lengths are scaled to genetic distance, with a distance (×100) of 1.0 corresponding approximately to a 1% difference in nucleotide sequences (with weighting of substitutions at individual positions and a transition/transversion ratio of 1.4). These laboratory reference strains were chosen to rule out contamination as an explanation for the virus isolations. When trees were inferred by use of maximum parsimony analysis (with bootstrap analysis of 100 data sets), the results showed nearly identical branch orders and bootstrap values of 99 and 100 at internal nodes separating sequences of each patient from all others (18).

No new substitutions associated with drug resistance (21) were identified in any of the virus recovered after 2 years (Table 3). In patient B, the Lys70 → Arg mutation, associated with AZT resistance, was absent in virus recovered from the first coculture experiment but was present in virus from a second isolation; this finding suggested the existence of mixtures of genotypes in the PBMCs of this patient, with stochastic events in coculture favoring the outgrowth of one or the other genotype. Plasma HIV RNA concentrations have remained undetectable (<50 copies/ml for all patients) 1 to 4 months after the recovery of virus (Table 1). In aggregate, these observations demonstrate that the recovery of virus in these patients was not attributable to drug failure, but rather to the continued persistence of long-lived and latently infected T cell populations.

Table 3

Lack of emergence of drug resistance mutations in six patients treated with up to 2 years of triple therapy and undetectable plasma RNA. In five of six patients, substitutions in position 63 of the protease corresponding to known natural polymorphisms (24) were seen at baseline and in virus isolates. In patient K, a natural polymorphism was found in position 71 at baseline and after 2 years. Box denotes data indicating a residue where the sequence after 2 years of treatment was wild type in one culture despite a resistance mutation at baseline. (–) indicates identity with consensus clade B sequence. Abbreviations for amino acids: A, Ala; D, Asp; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; T, Thr; V, Val; and Y, Tyr.

View this table:

Patient M had low but measurable viral RNA in plasma 1 year earlier (301 copies/ml) and at the time of these experiments (50 copies/ml). There was no recognized nonadherence to medications. In this case alone, virus was also recovered with standard coculture conditions. This sporadic appearance of small but detectable amounts of HIV RNA in plasma may indicate a less complete suppression of virus replication than was the case for the other five patients. However, even in this patient, the lack of acquisition of resistance mutations and the modest plasma viremia argue against the development of drug resistance as a reason for these differences (Table 3).

Whether the in vitro conditions used to recover virus in these studies translates to a comparable capacity for reactivation of viral infection (in the absence of suppressive antiviral therapy) in the complexity of the in vivo environment is not known. Nevertheless, these data suggest that it would be premature to discontinue suppressive therapy in such patients within this time frame. At the same time, the absence of discernible viral evolution, including the lack of emergence of new drug resistance substitutions, argues for the durability of the antiviral drug regimen used and provides encouragement that antiretroviral suppression can be sustained by quenching the replication of the viral progeny of these latently infected CD4 lymphocytes should they become activated in vivo.


Stay Connected to Science

Navigate This Article