Quiescent T Lymphocytes as an Inducible Virus Reservoir in HIV-1 Infection

The major reservoir for HIV-1 in the peripheral blood compartment of infected individuals is the CD4⁺ T lymphocyte (1, 2). The high percentage of cells (1 to 0.01%) within this reservoir that contain viral DNA (2) is difficult to reconcile with low percentage of infected cells (0.01 to 0.001%) that express viral RNA at levels detectable by in situ hybridization (3). In addition, the gradual depletion of CD4⁺ T lymphocytes during disease progression (4) contrasts with the acute cytocidal nature of HIV-1 infection of permissive T lymphocytes in vitro (5). These features suggest that a small population of infected cells is permissive for virus replication, whereas the majority of host cells harbor HIV-1 in a minimally replicative yet inducible state.

The life cycle of retroviruses can be separated into pre- and postintegration stages. Integration of HIV-1 DNA with the host cell genome, which depends on the activated state of the host cell (6), must occur for a productive virus infection (6, 7). Thus, although HIV-1 retains the capacity to bind and infect quiescent T lymphocytes in vitro (6–8), inefficient reverse transcription (7) and a block to integration of full-length HIV-1 DNA (6) restrict the viral life cycle in these cells to preintegration events.

Current in vitro models for HIV-1 persistence and latency have focused on agents that activate HIV-1 gene expression in cells harboring latent integrated provirus (9). However, initial integration requires T cell activation (6, 7), and, because the number of activated T cells in asymptomatic individuals is low, it has been hypothesized that quiescent T cells form a latent and inducible reservoir for HIV-1 in vivo (6, 7).

To identify molecular events during HIV-1 replication in vivo, we analyzed the arrangement of proviral DNA in single infected cells. This approach allowed us to determine whether the presence of extrachromosomal HIV-1 DNA within patient lymphocytes is due to restricted integration in quiescent cells or to superinfection (10, 11) of permissive cells by HIV-1. By initially determining HIV-1 proviral DNA copy number and the percentage of cells harboring HIV-1 within each infected individual included in this study (Table 1), we could analyze multiple (10 to 40) cell fractions, each containing not more than one HIV-1–infected cell (12). DNA was extracted from multiple replicate cell fractions and separated into high molecular weight DNA (containing integrated provirus) and low molecular weight DNA (containing extrachromosomal HTV-1 DNA forms) by agarose gel electrophoresis, and HIV-1 DNA was identified in high and low molecular weight fractions by polymerase chain reaction (PCR) (12).

Application of our protocol to 8E5 cells (13), which contain one noninfectious provirus per cell and completely lack unintegrated viral DNA, demonstrated that, if there were fewer than eight infected cells in each sample analyzed, mechanical shearing did not lead to separation of detectable amounts of integrated viral DNA in the low molecular weight fraction (Fig. 1A). Because our protocol (12) ensured that there was one infected cell in each fraction, we were able to distinguish integrated and unintegrated forms of HIV-1 DNA. Analysis of DNA isolated from cells infected with an HIV-1 mutant which, because of deletions in the HIV-1 intergrase coding region, is unable to associate with host-cell DNA (6, 14), demonstrated that trapping of low molecular weight extrachromosomal HIV-1 DNA forms in the genomic DNA fraction during gel electrophoresis did not lead to detectable amounts of unintegrated HIV-1 DNA in the high molecular weight genomic fraction (Fig. 1B).

The PCR protocol we used could detect a single HIV-1 provirus. Four HIV-1–infected cells (containing one provirus per cell) were mixed with 2 × 10³ uninfected lymphocytes and distributed into five fractions to be analyzed by PCR. Four of five fractions gave positive amplifications with primers specific for HIV-1 pol, whereas primers to the α-tubulin gene indicated the presence of equivalent amounts of genomic DNA in each fraction (Fig. 1C).

Table 1 summarizes the results of our analysis on the arrangement of HIV-1 DNA within individual infected T lymphocytes from eleven asymptomatic and seven AIDS/ARC (AIDS-related complex) patients. The number of peripheral blood lymphocytes (PBL) harboring HIV-1 genome (Table 1) varied among individuals within asymptomatic and AIDS groups; however, the increased virus DNA load in AIDS patients is in agreement with other studies (2, 15). The HIV-1 pol primers detect both complete and incomplete products of reverse transcription. However, use of primers [long terminal repeat (LTR)–gag] spanning the primer binding site and 5′ LTR-gag junction—synthesized after both template switching events (16)—identified late products in reverse transcription and gave identical results to those with the pol primers (Fig. 2).

In all patients analyzed, the presence of integrated and unintegrated HIV-1 DNA forms within the same fraction was rarely observed. When both forms were detected in the same fraction, this was due to the presence of more than one infected cell in that fraction, and preparation of a second dilution series containing fewer cells in each fraction resolved this. For example, in AIDS patient B12, analysis of replicate cell fractions in aliquots of 250 cells per fraction revealed integrated and unintegrated HIV-1 DNA in 40% of the replicate samples (Table 1). Repeat analyses of the same lymphocyte preparation with 100 or 50 cells per fraction resulted in the presence of both DNA forms in 18.75% and 0% of the replicates, respectively (Table 1). Limiting dilution and discrimination of integrated and extrachromosomal HIV-1 DNA in all patients in this study demonstrated that superinfection did not appear widespread in lymphocytes of individuals infected with HIV-1. This is not surprising because infected quiescent cells are not productive for virions and thus do not provide the conditions for superinfection. On the other hand, superinfection of a permissive (activated) T lymphocyte may be restricted because of receptor interference (10, 11, 17) or may be a rare and rapid event, difficult to detect in vivo. In any event, because of the low numbers of infected cells within patients (≤ 1%), superinfection is statistically unlikely.

Monoclonal antibody to the major histocompatibility complex class II human lymphocyte antigen (HLA)–Dr, which is expressed on macrophages and activated T lymphocytes, but not quiescent T lymphocytes (18), was used to deplete patient lymphocyte populations of activated T cells. After depletion of activated T cells, the proportion of cells harboring exclusively unintegrated viral DNA increased from 35 to 80% for patient B28 and from 20 to 100% for patient B30 (Table 1). Depletion of interleukin-2 receptor–positive lymphocytes of asymptomatic patient B24 demonstrated a similar striking predominance of extrachromosomal HIV-1 DNA forms in quiescent lymphocytes (19).

The biological activity of extrachromosomal HIV-1 DNA forms in infected individuals was assessed after T cell activation in vitro. Lymphocytes from patient B4 (HIV-1–positive, asymptomatic) were isolated; half of these were activated with phytohemagglutinin (PHA) and half of these were untreated. After 24 hours, 20 replicate cell fractions were prepared from the quiescent and activated cultures, and each replicate fraction was then analyzed for the presence of integrated and extrachromosomal viral DNA (12). High concentrations (7 μg/ml) of soluble CD4 (sCD4), which has been shown to completely inhibit HIV-1 binding and infection (20), were incorporated in the culture medium to prevent new rounds of infection by virus released from permissively infected (activated) lymphocytes. Of ten infected replicate cell aliquots (400 cells per fraction) from patient B4, six had exclusively extrachromosomal HIV-1 DNA forms (Fig. 2 and Table 1), as detected with the use of HIV-1 pol primers or HIV-1 LTR-gag primers that detect late products in reverse transcription. Two cell aliquots displayed exclusively integrated provirus, and two aliquots had both DNA forms (Fig. 2 and Table 1). After T cell activation in vitro and fractionation, all replicate cell samples containing HIV-1 genome displayed exclusively integrated provirus (Fig. 2 and Table 1). Thus, a large proportion of HIV-1 genome in asymptomatic individuals exists as full-length, extrachromosomal DNA, which retains the ability to integrate upon activation of the host cell.

To provide additional verification for the presence of full-length, extrachromosomal HIV-1 DNA with quiescent PBLs, we isolated total cellular DNA from HLA-Dr–depleted lymphocytes of AIDS patients B28 and B29 and analyzed this DNA with a ligation-mediated PCR protocol (21). The method was adopted so that free blunt 3’ LTR termini could be identified within enriched quiescent T lymphocytes of HIV-1–infected individuals. The blunt 3′ LTR terminus, which comprises full-length (−) strand (complementary to genomic viral RNA) and (+) strand (same polarity as genomic viral’RNA) viral DNA, exists only after completion of reverse transcription (22). The presence of free HIV-1 DNA termini in cells infected with an integration-defective HIV-1 mutant (HIV-1 ΔIN) was evident (Fig. 3C, upper panel), whereas no amplification products, even after two rounds of amplification (Fig. 3C, lower panel), were detectable in DNA isolated from 8E5 cells, which contain one integrated provirus per cell and completely lack extrachromosomal HIV-1 DNA (13). Integrated provirus and circular forms of unintegrated HIV-1 DNA (containing one or two LTRs) (23) are not detected by this approach because these viral DNA forms do not provide free 3’ LTR termini for attachment of the common linker that is required for subsequent PCR amplification (Fig. 3). Application of this ligation-mediated PCR method to DNA from HLA-Dr–depleted lymphocytes from patients B28 and B29 confirmed the presence of complete HIV-1 (−) strand cDNA terminating at nucleotide 9718 (Fig. 3A, step 1), as evidenced by the amplification of a 155–base pair (bp), HIV-1 LTR terminus–specific product (Fig. 3D). Use of HIV-1–specific primers directed to the 5’ LTR-gag junction demonstrated the abundance of HIV-1 DNA in each sample analyzed (Fig. 3B, upper panel). Sequence analysis of ligation-mediated PCR products (two rounds of amplification) by means of the nested LTR R primer (Fig. 3A, step 4) confirmed the amplification of complete 3’ LTR ends containing a GCAGT terminus (19). In a second series of PCR reactions, double-stranded DNA extracted from HLA-Dr–depleted lymphocytes from patients B28 and B29 was ligated directly to the common linker as in Fig. 3A, step 3 (essentially bypassing DNA denaturation and primer extension steps in Fig. 3A, steps 1 and 2). PCR amplification with the nested R primer and longer oligomer of the common linker resulted in amplification of a 155-bp product for DNA from both patients’ lymphocytes and from CD4⁺ cells infected with the HIV-1 integrase mutant, but not for 8E5 cell DNA (Fig. 3C). This evidence points to the presence of full-length (+) strand in extrachromosomal HIV-1 DNA in quiescent T lymphocytes of infected individuals.

Our results demonstrate the existence of a quiescent T cell reservoir where HIV-1 integration is restricted and HIV-1 DNA is harbored in an extrachromosomal state. The demonstration that a large fraction of HIV-1 DNA can exist in an extrachromosomal state in infected individuals is in agreement with other reports (24). However, the results presented in this study attribute the extrachromosomal DNA pool in HIV-1–infected individuals to infection of quiescent T cells, restricted integration, and establishment of a latent and inducible HIV-1 reservoir. Because T cell activation is a prerequisite for viral DNA integration, the contribution of factors affecting T cell activation (mitogenic agents, opportunistic infection) to reactivation of extrachromosomal HIV-1 DNA forms thus becomes apparent.