Kinetics of Response in Lymphoid Tissues to Antiretroviral Therapy of HIV-1 Infection

See allHide authors and affiliations

Science  09 May 1997:
Vol. 276, Issue 5314, pp. 960-964
DOI: 10.1126/science.276.5314.960


In lymphoid tissue, where human immunodeficiency virus–type 1 (HIV-1) is produced and stored, three-drug treatment with viral protease and reverse transcriptase inhibitors markedly reduced viral burden. This was shown by in situ hybridization and computerized quantitative analysis of serial tonsil biopsies from previously untreated adults. The frequency of productive mononuclear cells (MNCs) initially diminished with a half-life of about 1 day. Surprisingly, the amount of HIV-1 RNA in virus trapped on follicular dendritic cells (FDCs) decreased almost as quickly. After 24 weeks, MNCs with very few copies of HIV-1 RNA per cell were still detectable, as was proviral DNA; however, the amount of FDC-associated virus decreased by ≥3.4 log units. Thus, 6 months of potent therapy controlled active replication and cleared >99.9 percent of virus from the secondary lymphoid tissue reservoir.

In HIV-1 infection, the measurement of viral load in plasma is a useful guide to prognosis and to the efficacy of antiretroviral therapy (1). Ultimately, however, the impact of treatment can only be assessed completely in the lymphoid tissue (LT) reservoirs, where most of the virus is produced by CD4+ T lymphocytes, macrophages, and other lymphoid MNCs and is stored in immune complexes on the surfaces of FDCs. In the asymptomatic stage of infection, the pool of virus on FDCs is at least an order of magnitude greater than that in MNCs (2). In turn, both LT viral compartments exceed by orders of magnitude the quantity of free and cell-associated virus circulating in the bloodstream. In reports published to date, the LT viral pools are little affected by monotherapy with nucleoside analog drugs that inhibit reverse transcriptase (RTIs) (2, 3) and are only moderately reduced by therapy with two or three RTIs (4, 5).

We investigated the effect of treatment with a more potent antiretroviral drug combination on viral burden in serial tonsil biopsies (6), which have been shown to be representative of the secondary LT in HIV-1 infection (2). We used a technique that combined in situ hybridization with computer-assisted quantitative image analysis to measure and differentiate HIV-1 RNA in virus-producing MNCs and in virion-antibody immune complexes deposited on FDCs (2). Biopsies were obtained from participants with visible tonsils in an LT substudy of a clinical treatment trial combining a potent HIV-1 protease inhibitor, ritonavir, and two RTIs, zidovudine and lamivudine. All participants in the parent study received ritonavir throughout, whereas both RTIs were initiated either upon enrollment (immediate group) or 3 weeks later (delayed group) to test the hypothesis that initial reductions in viral replication would forestall the development of drug resistance against RTIs. Thirty-four previously untreated HIV-1–seropositive adults with absolute CD4+ cell counts of ≥50 cells per cubic millimeter and plasma HIV-1 RNA levels of ≥30,000 copies per milliliter were randomized in this open label trial (7). After 6 months of treatment, plasma HIV-1 RNA levels decreased at least 2.9 log units in both groups from a pretreatment median load of 5.3 log units (8).

Participants underwent tonsil biopsy 2 weeks before and 2 days, 22 days, and 24 weeks after treatment commenced. Because we could only obtain a few tonsil biopsies from any one individual, we chose time points to detect anticipated early rapid changes in the MNC compartment and more gradual changes in the FDC-associated pool. We based our selection of time points on published dynamic models of HIV-1 infection that predict a rapid turnover of productively infected MNCs (9,10). We chose the later time points on the basis of previous work describing the protracted retention of conventional virus-containing antigen-antibody complexes by FDCs (11), the apparent stability of the FDC-HIV association (12), and the enormity of the FDC-associated HIV-1 reservoir (2). From these previous observations, we expected a much slower turnover of FDC-complexed virus.

Each tonsil biopsy was cut in half. One portion was fixed and embedded in paraffin; the other portion was flash-frozen for later extraction and assay of viral nucleic acids. Paraffin blocks were then sectioned and hybridized in situ to a 35S-labeled RNA probe complementary to >90% of the sequences in HIV-1 RNA. After autoradiography, the hybridization signal overlying FDCs or MNCs was quantitated by computer-assisted image analysis (2).

We obtained sequential tissue samples suitable for evaluation from 10 individuals (13). At baseline, this cohort had a mean of 1.5 × 108 copies of viral RNA per gram on FDCs (Table 1) (14), equal to the mean concentration previously measured at an earlier stage of infection (2). By contrast, the mean frequency of MNCs with >20 copies of HIV-1 RNA per cell was 3.1 × 105 cells per gram (Table 1). In these individuals with more advanced HIV-1 infection, the MNC pool thus proved to be larger than in earlier HIV disease stages (2). The result, however, is consistent with the higher mean baseline plasma HIV-1 RNA levels that imply greater rates of virus production.

Table 1

Effect of combination antiretroviral therapy on HIV stored on FDCs and on productively infected MNCs (those with >20 copies of HIV RNA per cell) in LT. Tonsil biopsies were obtained from 10 HIV-infected adults within 14 days before initiating treatment with ritonavir, zidovudine, and lamivudine, and at the designated times thereafter. One-half of the biopsy was fixed in Streck’s tissue fixative for 24 hours, stored for 3 to 10 days in 80% ethanol, and then embedded in paraffin. Sections (8 μm) were cut, pretreated, and hybridized in situ to a 35S-labeled HIV-specific antisense probe, as described (16). After hybridization, the sections were washed, coated with nuclear track emulsion, and, after 1 day of autoradiographic exposure, developed and stained. The number of copies of HIV RNA per gram on FDCs in baseline, day 2, day 22, and week 24 biopsies was calculated from the number of silver grains determined by quantitative image analysis contained within the cumulative areas of 10 or more sections, as described (2). Productively infected MNCs are easily identified in tissue sections with 1-day exposures. Asterisks denote week 24 biopsies containing detectable MNCs with <10 copies of HIV RNA per cell after 10-day exposures. Time points at which the available tonsil biopsy was inadequate for analysis (usually because the tissue was almost entirely epithelial) are indicated by –. See (18) for a description of how the limits of detection were determined.

View this table:

After 2 days of treatment, we observed a rapid drop from baseline in the frequency of MNCs with the highest intracellular concentration of viral RNA, those with >75 copies of HIV-1 RNA per cell (Fig.1). These MNCs (the right tail of the frequency distribution in Fig. 1A) disappeared quickly with treatment, which affirms our previous proposal that this population consists of cells in the late stages of the viral life cycle in infections that were initiated asynchronously in vivo (2). This subpopulation of infected cells with the highest intracellular concentration of HIV-1 turns over rapidly and contributes most of the daily viral production.

Figure 1

Representative frequency distribution of viral RNA copies per MNC in LT before (baseline) and 2 and 22 days after treatment. Two days after initiating treatment, most of the cells with >75 copies of HIV RNA per MNC at baseline (vertical arrow) have been eliminated from the distribution. These cells are infected cells at later stages of the viral life cycle (represented by a large, spreading horizontal arrow). The frequency distribution of the number of copies of viral RNA per cell was determined by in situ hybridization and quantitative image analysis for ≥100 cells at each time point for each biopsy.

From the data summarized in Table 1 and displayed in Fig.2A, we estimate a steady-state half-life oft 1/2 = 0.9 days for the population of MNCs in the later stages of a productive infection. This half-life corresponds to a turnover rate of 1.2 × 105 cells g 1 day 1, or ∼8 × 107 cells in a 70-kg person (assuming LT is 1% of total body weight) (14). This directly measured rate of infected MNC turnover is about four times the rate we estimated previously for individuals in an earlier stage of HIV-1 infection (2). However, the faster rate of turnover implies a greater production of virus and is consistent with the higher plasma HIV-1 RNA levels (8).

Figure 2

Effect of treatment on the frequency of productively infected MNCs and on viral burden associated with FDCs in LT. The frequency of productively infected MNCs (A) and the number of copies of HIV-1 RNA on FDCs (B) were determined as described in Fig. 1 and Table 1 and were plotted against days of treatment for all 10 patients (represented by the various symbols). Over the course of the first 22 days of treatment, changes in the two cellular compartments parallel each other with initially rapid and then slower rates of decrease. (C and D) Mean decreases for the LT study group as a whole (in log units).

In vitro studies of FDC interactions with HIV-1 have shown that virus remains tightly bound to the FDCs for at least several days (12). For this reason, we had not anticipated the surprisingly rapid initial rate of elimination of virus from FDCs that we observed. The loss from FDCs closely followed the rapid decline in the number of productively infected lymphoid MNCs (Fig. 2) and in the amount of plasma HIV-1 RNA (8). From the half-life of FDC HIV-1 RNA (t 1/2 = 1.7 days), we calculate an initial clearance rate of 4.3 × 107 copies g 1 day 1 of viral RNA, or ∼2.1 × 107 virions g 1 day 1 for the FDC pool. We interpret this lability of the FDC pool as evidence of a pretreatment state in which the binding and dissociation of virion-antibody complexes from FDCs is in equilibrium with the production and clearance of virus from the peripheral circulation. When potent therapy reduces the amount of newly produced virus available to bind to FDCs, clearance and dissociation continue apace, and consequently the FDC pool quickly shrinks.

Because the source of virus and the largest pool of virus are at steady state before treatment, we can directly estimate pretreatment virus production in vivo from the initial decrease in productive cells and from the virus lost from the FDC pool (14). From the initial loss of ∼1.2 × 105 productively infected cells g 1 day 1 of LT and the decrease of 2.1 × 107 virions g 1 day 1 in LT, we estimate that each cell produces ∼175 HIV-1 virions. This dynamic measurement of HIV-1 production is in good agreement with our previous indirectly derived estimates (2) and is also consonant with published estimates derived from plasma virus clearance of ∼1010 virions per day (10, 15).

By day 2, more than 75% of the infected MNCs contained only 20 to 60 copies of HIV-1 RNA each, a considerable reduction in the productive burst size. Between day 2 and day 22 of treatment (Fig. 2, C and D), the MNC and FDC pools continued to decay in parallel with one another, but with slower kinetics (t 1/2 ≈ 15 and 14 days, respectively). In addition, we found many MNCs with >20 but <75 copies of HIV-1 RNA in the third week of treatment (Fig. 1). It is possible that these cells with lower concentrations of intracellular RNA are cells that are blocked at an early phase of the viral life cycle and did not progress to later stages after 3 weeks of treatment. However, it is more likely that these cells represent a longer lived subpopulation of cells infected before initiation of treatment. The continued production of smaller amounts of virus by this subpopulation of MNCs would partially compensate for the ongoing clearance from the FDC-associated viral pool and thereby for the parallel and slower rate of decay in the FDC virus pool between days 2 and 22.

Because infected macrophages might fulfill the criteria of longer lived cells with lower levels of viral RNA, we examined day 22 tissue sections for evidence of HIV-1 RNA in macrophages by a double-label technique (16). Sections that had been hybridized in situ to detect HIV-1 RNA were stained immunohistochemically with CD68 monoclonal antibody (mAb) to unambiguously identify macrophages (17). In these experiments, a few macrophages containing HIV-1 RNA were identified, but at day 22 of treatment, the vast majority of the cells in which viral RNA could be detected did not stain with CD68 mAb (Fig. 3).

Figure 3

Macrophages after treatment. Double- label in situ hybridization (16, 17) was used to identify macrophages and HIV RNA in MNCs. In the developed autoradiographs of tonsil tissue obtained 22 days after treatment, most cells with the greatest concentration of HIV RNA are not macrophages. In this image, trans- and epipolarized light make silver grains appear white, indicating HIV RNA in a cell (arrow). The HIV RNA hybridization signal does not colocalize with the dark macrophages (circled area) that have been stained immunohistochemically with CD68 mAb (original magnification, ×400).

Between 3 and 24 weeks of treatment, the number of MNCs with >20 copies of HIV-1 RNA per cell fell to undetectable amounts in all 10 LT substudy participants (mean decrease in cells per gram, >2.3 log units), as did the amount of FDC-associated HIV-1 RNA in most individuals (mean decrease in HIV-1 RNA copies per gram, ≥3.4 log units). To extend the analysis to residual pools of virus and infected cells, we lengthened the autoradiographic exposure time of 1 day (appropriate for quantitating large amounts of HIV-1 RNA) to 10 days. We thus increased the limits of sensitivity to about one HIV-1 RNA copy per cell (18). With this more sensitive assay, we found considerable heterogeneity in the subjects’ treatment response at 6 months. In 6 of 10 individuals, there was a mean residue of 2 × 105 copies of HIV-1 RNA per gram on FDCs (Fig.4, A and B), equivalent to ∼108 virions in a 70-kg individual. With these longer exposures, in five individuals we also observed 103 to 104 MNCs per gram with fewer than 10 copies of HIV-1 RNA per cell (Fig. 4C). At week 24 of treatment in one individual (patient 5, Table 1), we could not detect any trace of residual HIV-1 RNA in either cellular compartment, down to a sensitivity of <300 cells per gram for cells with at least one copy of viral RNA per cell.

Figure 4

Viral burden in LT after 24 weeks of treatment. The sensitivity of detection of infected cells and residual HIV RNA in FDCs was increased by an order of magnitude by lengthening the autoradiographic exposure time. Before treatment (A), a large and diffuse hybridization signal (silver grains) overlies FDCs in a GC. Dense collections of silver grains overlie three infected MNCs. At 24 weeks, signals are greatly reduced but still detectable in some cells and GCs (B and C). The circled area in (B) defines a GC with 150 copies of HIV RNA on FDCs. The arrow in (C) denotes a MNC with <10 copies of HIV RNA. Original magnifications, ×160 (A and B) and ×400 (C).

In 6 of the 10 subjects (patients 1 to 6, Table 1), frozen tonsil tissue specimens were available for study. Nucleic acid extracted from the frozen portion of the week 24 biopsy of patient 5 also lacked detectable HIV-1 RNA, as assessed by a nested reverse transcription polymerase chain reaction (RT-PCR). However, in all six patients for whom frozen biopsy specimens were available, HIV-1 DNA was detectable by nested PCR amplification of proviral DNA extracted from the week 24 frozen biopsy (19).

After 6 months of triple therapy, LT still harbors infected cells, and in some of these cells there is evidence of low levels of viral gene expression. The life-span and function in HIV-1 disease of MNCs in which we can still detect small amounts of viral RNA requires further investigation, but as long as they and latently infected cells (20) live, there will be reason to continue treatment. Despite the persistent infection, the number of copies of HIV-1 RNA per gram cleared from LT was ≥3.4 log units over just 6 months of therapy. If continued at the same rate seen between 3 weeks and 6 months, this extrapolates to elimination of viral RNA within an average of 30 months of triple antiretroviral therapy. Further studies will be necessary to ascertain whether it is possible to completely purge HIV-1 infection from LT, or whether lifelong maintenance therapy will be required after initial “induction” treatment. Nevertheless, we have shown that within 6 months, triple drug therapy eliminates more than 99% of the lymphoid cells actively producing the virus that is responsible for immune depletion (21).


View Abstract

Stay Connected to Science

Navigate This Article