|Home | About | Journals | Submit | Contact Us | Français|
The persistence of human immunodeficiency virus type 1 (HIV-1) in memory CD4+ T cells is a major obstacle to the eradication of the virus with current antiretroviral therapy. Here, we investigated the effect of the activation status of CD4+ T cells on the predominance of R5 and X4 HIV-1 variants in different subsets of CD4+ T cells in ex vivo-infected human lymphoid tissues and peripheral blood mononuclear cells (PBMCs). In these cell systems, we examined the sensitivity of HIV replication to reverse transcriptase inhibitors. We demonstrate that R5 HIV-1 variants preferentially produced productive infection in HLA-DR− CD62L− CD4+ T cells. These cells were mostly in the G1b phase of the cell cycle, divided slowly, and expressed high levels of CCR5. In contrast, X4 HIV-1 variants preferentially produced productive infection in activated HLA-DR+ CD62L+ CD4+ T cells, which expressed high levels of CXCR4. The abilities of the nucleoside reverse transcriptase inhibitors (NRTI) zidovudine and lamivudine to stop HIV-1 replication were 20 times greater in activated T cells than in slowly dividing HLA-DR− CD62L− CD4+ T cells. This result, demonstrated both in a highly physiologically relevant ex vivo lymphoid tissue model and in PBMCs, correlated with higher levels of thymidine kinase mRNA in activated than in slowly dividing HLA-DR− CD62L− CD4+ T cells. The non-NRTI nevirapine was equally efficient in both cell subsets. The lymphoid tissue and PBMC-derived cell systems represent well-defined models which could be used as new tools for the study of the mechanism of resistance to HIV-1 inhibitors in HLA-DR− CD62L− CD4+ T cells.
Human immunodeficiency virus type 1 (HIV-1) infection in vivo is transmitted predominantly by HIV variants that utilize CCR5 (R5 variants), which dominate in early stages of the disease. Later, in many cases, CXCR4-utilizing variants (X4 variants) evolve and are associated with depletion of CD4+ T cells and with a rapid progression to AIDS. Recent studies have clearly established that infection with HIV-1 predominantly occurs in CD4+ memory T cells (8, 14, 27, 29, 47, 51, 57, 67). Memory cells comprise several subsets defined by differential expression of CD45RA, CD45RO, homing receptors, and activation markers (16, 35, 41, 52). Both CD45RO+ CD62L+ central memory CD4+ T cells (TCM) and CD45RO+ CD62L− effector memory CD4+ T cells (TEM) (35, 52) present CXCR4 on their surfaces. Expression of CXCR4 is rapidly upregulated during T-cell activation (7, 36). CD4+ TEM cells express higher levels of CCR5 than CD4+ TCM cells (24, 33, 36, 46, 50, 72). Expression of CCR5 is upregulated more slowly than expression of CXCR4 during in vitro T-cell activation (7, 36).
We have previously shown that CD62L− CD4+ TEM cells in the lymphoid tissue are preferential targets for productive infection with R5 HIV-1 variants (24), whereas CD62L+ CD4+ TCM cells are preferentially productively infected with X4 HIV-1 variants (24). The transient coexpression of both CD45RA and CD45RO molecules in the latter cells suggests that they are in the process of transition from the naïve to the memory phenotype and have been activated recently (2, 24). Here, we investigated the effect of the activation status and proliferation of CD4+ T cells on the predominance of R5 and X4 HIV-1 variants in different subsets of CD4+ T cells in ex vivo-infected human lymphoid tissues and peripheral blood mononuclear cells (PBMCs).
Recent results have shown that CD4+ T cells repeatedly stimulated with influenza virus antigen in a mouse model exhibited extensive downmodulation of CD62L and sustained proliferation activity (31). In order to simulate this process in vitro and to prepare long-term cultures of CD62L− CD4+ Tcells from PBMCs, we periodically activated noninfected CD4+ T lymphocytes with phytohemagglutinin (PHA) in the presence of interleukin-2 (IL-2). A similar method was used in the early days of HIV research to keep persistently infected cultures of CD4+ T cells from PBMCs viable for more than 3 months (30). Indeed, we found that this procedure results in enrichment of the cell culture by CD45RO+ CD62L− CD4+ T cells that produce, after adequate stimulation, the functional markers of TEM cell gamma interferon (IFN-γ) and IL-4.
Whereas quiescent (G0) T lymphocytes in tissue culture are completely refractory to HIV-1 replication, several mechanisms related to progression from the G0 phase to the G1 phase of the cell cycle can render T cells susceptible to HIV infection without substantially changing their phenotype (14, 15, 53, 56, 59-61, 63, 65, 68). We found that R5 HIV-1 variants preferentially produced productive infection in slowly dividing CD25− HLA-DR− CD4+ T cells, which were mostly in the G1b phase of the cell cycle, expressed high levels of CCR5, and were mostly CD62L−. In contrast, X4 HIV-1 variants preferentially produced productive infection in activated CD25+ HLA-DR+ CD4+ T cells, which expressed high levels of CXCR4 and were mostly CD62L+. We addressed the nature of the sensibility of HIV replication in these cell subsets to reverse transcriptase inhibitors. The nucleoside reverse transcriptase inhibitors (NRTI) azidothymidine (AZT) and lamivudine (3TC) were 20 times more potent in stopping HIV-1 replication in activated than in slowly dividing HLA-DR− (G1b) CD4+ T cells. Higher resistance of HIV-1 replication to AZT in slowly dividing CD25− HLA-DR− CD4+ T cells than in activated CD4+ T cells correlated with lower levels of thymidine kinase, an enzyme necessary for metabolic transformation of AZT to AZT triphosphate, in cells of the former type. The nonnucleoside reverse transcriptase inhibitor (NNRTI) nevirapine was equally efficient in both cell subtypes. The relative resistance of replication of R5 HIV-1 variants in slowly dividing CD62L− CD4+ T cells to NRTI was demonstrated both in a highly physiologically relevant ex vivo lymphoid tissue model and in PBMCs. These cell systems represent well-defined models which could be used as new tools for the study of the mechanism of resistance to HIV-1 inhibitors in HLA-DR− CD62L− CD4+ T cells, as well as for testing new drugs able to inhibit persistent virus replication.
PBMCs of healthy donors were separated on Ficoll-Hypaque gradients. Aliquots of 1 × 108 PBMCs depleted of monocytes by adherence to a plastic culture flask were activated with PHA-P (Difco, Franklin Lakes, NJ) at 2 μg/ml in RPMI 1640 supplemented with 200 U/ml recombinant IL-2 (Chiron), 15% fetal calf serum, and antibiotics for 3 days. CD4+ T cells were purified by depletion of CD8+ T cells with the anti-CD8 antibody at saturating concentration with magnetic beads coated with goat anti-mouse antibody (Miltenyi Biotech, Bergisch Gladbach, Germany), and the positively labeled cells were removed as described previously (23). CD4+ T cells were then infected with HIV-1 NL4-3 (1), with HIV-1 49.5 (containing the V3 region of HIV-1 Bal in the HIV-1 NL4-3 backbone ), or with HIV-1 AD8 (62) at a multiplicity of infection (MOI) of 0.001 or 0.01 tissue culture infective doses (TCID) per cell. Alternatively, CD4+ T cells were transduced with HIV-1-derived vector (HDV) (kindly provided by D. R. Littman, Skirball Institute of Biomolecular Medicine, NY), prepared, and used as described by Unutmaz et al. (63). Peripheral blood lymphocytes were cultured at a concentration of 106 cells/ml in the presence of IL-2 and were fed and analyzed every 3 or 4 days for the expression of activation and memory markers and for production of intracellular p24gag.
Ex vivo-infected human lymphoid tissue supports productive infection by HIV-1 virus without exogenous activation. Tonsillar tissue blocks placed on top of collagen sponge gels were infected with the X4 virus HIV-1 NL4-3 or the R5 virus HIV-1 AD8 as described elsewhere (20, 21). In a typical experiment, 3 to 5 μl of clarified virus-containing medium, approximately 150 50% TCID per block, was applied to the top of each tissue block. Flow cytometry analysis was performed on cells mechanically isolated from the control and ex vivo-infected blocks of human lymphoid tissue 12 days postinfection. Lymphocytes were identified according to their light-scattering properties and then analyzed for the expression of lymphocyte-specific memory and activation markers and virus-specific intracellular p24gag (24).
The antibodies anti-CD3-peridinin chlorophyll protein (PerCP), anti-CD4-PerCP, anti-CD4-phycoerythrin (PE)-cyanine 7 (Cy7), anti-CD62L-fluorescein isothiocyanate (FITC), anti-CD62L-allophycocyanin (APC), anti-HLA-DR-FITC, anti-HLA-DR-PerCP, anti-CXCR4-PE, anti-CCR5-PE, anti-CCR7-PE-Cy7, anti-Ki67-PE, goat anti-mouse immunoglobulin G1 (IgG1)-PE, and goat anti-mouse IgG1-APC were purchased from Becton Dickinson, Pharmingen, Inc. (San Diego, CA). Anti-IL-4-PE, anti-IFN-γ-PE, anti-CD25-FITC, anti-CD25-PE, anti-CD8-FITC, and anti-p24gag-rhodamine were purchased from Beckman Coulter S.A. (Paris, France). Anti-CD45RO-APC, anti-CD4-APC, and anti-CD3-APC were purchased from Caltag (Burlingame, CA). Monoclonal antibody anti-CCR5 (clone 3A9; produced by Pharmingen) was obtained through the AIDS Research and Reference Reagent Program from DAIDS, NIAID, and visualized with goat anti-mouse IgG1.
For analysis of cell surface marker expression, 2 × 105 cells were washed in phosphate-buffered saline containing 0.5% fetal calf serum and 0.02% sodium azide and incubated for 15 min at room temperature in the presence of the appropriate antibodies at concentrations recommended by the producer. For detection of intracellular antigens, the cells were fixed and semipermeabilized with Cytofix-Cytoperm (Pharmingen). Intracellular expression levels of IFN-γ and IL-4 in the total T-cell population were determined after 4 h of stimulation with 100 nM phorbol myristate acetate, ionomycin (1 μg/ml), and brefeldin A (10 μg/ml). After labeling, cells were fixed in 4% paraformaldehyde and analyzed after gating on live lymphocytes with a FACScan using CELLQuest software (Becton Dickinson, Le Pont de Claix, France). Some analyses of CD4+ T cells from lymphoid tissue were performed with a FACSAria using DIVA software (Becton Dickinson, Le Pont de Claix, France).
For evaluation of the presence of cycling cells in the cultures, the method of simultaneous labeling of RNA and DNA with pyronin Y (PY) and 7-amino-actinomycin D (7AAD) was used (23). The proliferation rate of T-cell cultures was determined from decrease in mean intensity of labeling of CFSE [5-(and 6)-carboxyfluorescein diacetate succinimidyl ester]-labeled T lymphocytes. CD4+ T lymphocytes (107 cells) were labeled with 0.6 μM CFSE in phosphate-buffered saline for 10 min at 37°C. Cells were washed and further cultivated in complete culture medium.
The endpoint titers of HIV-1 NL4-3, HIV-1 49.5, and HIV-1 AD8 were determined from syncytium formation after infection of C8166-CCR5 indicator T cells (17) with 10-fold dilutions of virus-containing cell-free supernatant in duplicate.
For the determination of the anti-HIV-1 activities of reverse transcriptase inhibitors, 1 × 105 CD4+ T cells were incubated with various concentrations of AZT or nevirapine (0.005, 0.05, 0.5, 5.0, and 50 μM) in 100 μl of complete culture medium for 2 h. The cell cultures were then exposed to one of the HIV-1 clones (NL4-3, 49.5, and AD8) or to HDV for 20 h, washed, and further treated with the drug at the original concentration for 3 days. After another change of culture medium, the proportion of cells that expressed p24gag was determined by means of fluorescence-activated cell sorter (FACS) analysis 7 days postinfection. All assays were performed in duplicate. The 50% inhibitory concentration (IC50) was determined from analysis of the regression curve obtained by semilogarithmic plot of the proportion of cells productively infected with HIV-1 against the logarithm of concentration of drug.
Real-time reverse transcriptase PCR was performed to quantify thymidine kinase transcripts. The primer set TK+ (forward), 5′-ATGAGCTGCATTAACCTGCC-3′, and TK− (reverse), 5′-AATCACCTCGACCTCCTTCT-3′, was used to detect thymidine kinase mRNA. We performed amplification and detection with an Applied Biosystems Prism 7000 sequence detection system using an Absolute QPCR SYBR green ROX mix (ABgene Life Sciences, Courtaboeuf, France) with the forward primer at 300 nM, the reverse primer at 300 nM, and 100 to 500 ng of template cDNA in a 25-μl reaction volume. We performed and analyzed the reactions by using an ABI Prism 7000 sequence detection system (Perkin-Elmer-Applied Biosystems, Foster City, CA).
The intracellular levels of dTTP in activated and resting CD4+ T cells were detected as described previously (69). Briefly, aliquots of 107 cells resuspended in 1 ml of 60% methanol and incubated overnight at 4°C were dried for 2 h under vacuum. The dried pellets were then resuspended in 100 μl of distilled H2O, incubated at room temperature for 5 min, and centrifuged at 14,000 × g for 15 min. The reaction cocktail (50 μl) included 40 mM Tris-HCl (pH 7.5), 0.5 mg of bovine serum albumin per ml, 10 mM MgCl2, 10 mM dithiothreitol, 3.5 μM oligonucleotide template (5′-TTATTATTATTATTATTAGGCGGTGGAGGCGG-3′), 3.5 μM oligonucleotide primer (5′-CCGCCTCCACCGCC-3′), 0.1 U of DNA polymerase I (U.S. Biochemical Corp.), 0.5 μM [3H]dATP, and 5 μl of sample. This mixture was incubated at 37°C for 1 h, and the quantity of incorporated radioactivity was determined on DE81 filters.
R5 variants HIV-1 AD8 and 49.5 are only weakly infectious for PHA-activated CD4+ T cells. First, the subsets of activated memory T cells in PBMCs were examined to see if they were differentially infected by R5 and X4 HIV-1 variants. For this purpose, we characterized the induction of activation and memory markers and HIV-1 coreceptors on monocyte- and CD8+ lymphocyte-depleted CD4+ T cells from PBMCs stimulated with PHA and kept afterwards in the presence of IL-2 (Fig. (Fig.1).1). The majority (>90%) of PHA-activated CD4+ T cells coexpressed CD25, CD45RO, and CD62L molecules on their surfaces 10 days after PHA activation (23) (Fig. (Fig.1A).1A). The peak of intensity of CXCR4 expression, measured as mean fluorescence intensity (MFI), correlated with the maximum CD25 intensity (Fig. 1C and D). After downregulation of the cell surface expression of CD25 (<15% of maximal MFICD25), the expression of CXCR4 remained stable at 40% of the maximal level of MFICXCR4. In contrast, the initial level of 7% of CCR5 expression in CD4+ T cells dropped to less than 1% after PHA activation and then gradually increased to a 10% level over a period of 30 days after PHA activation (Fig. (Fig.1B1B).
The CD4+ T lymphocytes were infected with the X4 variant HIV-1 NL4-3 and with the R5 variants HIV-1 AD8 and HIV-1 49.5 at the same infective doses (MOI of 0.01 TCID/cell) at regular intervals after PHA activation (Fig. (Fig.1E).1E). The R5 variant HIV-1 49.5 is isogenic with HIV-1 NL4-3 except for the HIV-1 Bal-derived V3 loop of gp120 (12). The viral infectivity, defined as the percentage of p24gag+ CD4+ T cells, was determined with FACS at constant periods of 7 days after each challenge. HIV-1 NL4-3 infected up to 55% of the activated CD25+ CD4+ T cells that expressed high levels of the coreceptor CXCR4 (Fig. (Fig.1E).1E). After an eightfold downregulation of the cell surface expression of CD25 (<15% of maximal MFICD25), accompanied by a 2.5-fold decrease of CXCR4 intensity (MFICXCR4) (Fig. (Fig.1D),1D), the infectivity of HIV-1 NL4-3 dropped about 100 times (<1% of p24gag+ T cells) (Fig. (Fig.1B).1B). In contrast, the R5 variants HIV-1 AD8 and HIV-1 49.5 infected only about 10% of activated CD25+ CD4+ T cells; however, their infectivity was maintained after the downregulation of CD25 to a greater extent than that of HIV-1 NL4-3 (Fig. (Fig.1B1B).
At the peak of infectivity of HIV-1 NL4-3, when the cells were challenged 11 days after PHA activation and analyzed 7 days later, 66.9% of p24gag+ T cells were CD62L+ (Fig. (Fig.1F).1F). In contrast, the majority of p24gag+ T cells that were productively infected with HIV-1 AD8 (challenged 28 days after PHA activation and analyzed 7 days later) were CD62L− (66%). Less then 0.2% of control mock-infected cells were p24gag+. Taken together, the CD25+ CXCR4+ CCR5− CD62L+ T-cell-enriched population was strongly productively infected with the X4 variant HIV-1 NL4-3 but only weakly with the R5 variants HIV-1 AD8 and HIV-1 49.5.
In order to study the mechanism of HIV-1 persistence, we prepared long-term cultures of CD4+ T cells. To this end, the PHA-activated PBMCs of normal healthy donors, depleted of monocytes and CD8+ T cells, were repeatedly activated with PHA at intervals of about 20 days (Fig. (Fig.2A).2A). The time of PHA reactivation was determined on the basis of the return of activated T cells into resting phase, monitored as the loss of CD25 expression. CD4+ T cells proliferated during this treatment. After repeated cycles of PHA activation, the majority of CD4+ T cells downregulated expression of CD62L and concomitantly upregulated production of IFN-γ and IL-4. We characterized the activation and cell proliferation statuses of these cultures.
Without subsequent PHA activation, the cell count of the activated CD62L+ CD4+ T-cell-enriched cultures subjected to one cycle of PHA activation increased 10-fold during the 20-day period; then, the cell culture degenerated and died within about 50 days (Fig. (Fig.2B).2B). In contrast, CD4+ T cells reactivated five times with PHA survived for the next 120 days (Fig. (Fig.2C);2C); thus, the total life span of this cell culture, together with the periods of repeated activation, reached 200 days (n = 3). As assessed by means of CFSE assay, the CD4+ T-cell culture subjected to one PHA activation proliferated approximately 6.6 times faster (doubling time, t = 0.9 days; R2 = 0.97) than the culture subjected to five PHA activations (t = 6.0 days, R2 = 0.82) (Fig. 2D and E). In parallel, we examined the cell cycle statuses of both cell cultures by labeling with 7AAD for DNA and with PY for RNA. Approximately 34.1% of CD4+ T cells were found in the S/G2/M phase of the cell cycle 7 days after the first PHA activation (Fig. (Fig.2F);2F); by contrast, ninefold fewer (3.8%) T cells were detected in the same phase 10 days after the fifth PHA activation (Fig. (Fig.2G).2G). The majority (94%) of CD4+ T cells 10 days after the fifth PHA activation were in the G1b phase of the cell cycle. Collectively, our results show that the PBMC culture subjected to repeated PHA activations was enriched with slowly dividing T cells that were CD25− HLA-DR− CD62L− and produced IFN-γ and IL-4. This culture proliferated significantly longer in the absence of further activation signal than the culture subjected to one PHA activation.
The majority (60 to 80%) of the cells subjected to five cycles of PHA activation showed the CD62L− phenotype independently of expression of CD25 and HLA-DR (Fig. (Fig.3A).3A). To investigate the differential susceptibility of this culture to productive infection with X4 and R5 variants of HIV-1, we examined the expression of activation and memory markers and of HIV-1 coreceptors in repeatedly activated CD4+ T cells. After the downregulation of the cell surface expression of CD25 (<10% of maximal MFICD25) (Fig. (Fig.3C)3C) and the concomitant drop in percentage and MFI of CXCR4+ CD4+ cells, the percentages of CCR5+ and CXCR4+ cells gradually increased (up to 30-fold for CCR5) (Fig. (Fig.3B3B).
The CD4+ T lymphocytes were infected at regular intervals with the X4 variant HIV-1 NL4-3 and the R5 variant HIV-1 AD8 (and with the R5 variant HIV-1 49.5 4 and 14 days after the fifth PHA activation) at the same infective doses (MOI of 0.01 TCID/cell) (Fig. (Fig.3E).3E). We determined viral infectivity by using FACS at constant periods of 7 days after each challenge. Activated CD25+ CD4+ T cells expressing low levels of both coreceptors, CXCR4 and CCR5, were infected only weakly with X4 and R5 variants of HIV-1 (≤4% CD4+ T cells). After the downregulation of cell surface expression of CD25 (<5% of maximal MFICD25) and HLA-DR, accompanied by an increase in CCR5 expression, infectivity of HIV-1 AD8 and 49.5 increased (30 to 70% CD4+ T cells). In contrast, the X4 variant HIV-1 NL4-3 infected less than 4% of the cells in cultures enriched with slowly dividing HLA-DR− CD62L− T cells (Fig. 3E and F). In the cell culture subjected to five PHA activations, HIV-1 AD8 predominantly infected CD62L− CD4+ T cells (89.8% at 30 days after activation) (Fig. (Fig.3F).3F). Less then 0.2% of control mock-infected cells were p24gag+. Taken together, the slowly dividing population of CD4+ T cells that were CD25− HLA-DR− CXCR4− CCR5+ CD62L− after five PHA activations was strongly productively infected with the R5 variant HIV-1 AD8 but only weakly with the X4 variant HIV-1 NL4-3.
To investigate correlations of viral infectivity with the levels of activation and differentiation markers and with the expression of HIV-1 receptors on CD4+ T cells, we compared percentages of cells infected with HIV-1 NL4-3 and HIV-1 AD8 after one cycle or five cycles of PHA activation with percentages or MFIs of CD25, CD62L, CXCR4, and CCR5 expressed in these cell cultures shown in Fig. Fig.11 and Fig. Fig.3.3. In these comparisons, we observed a direct correlation between the infectivity of HIV-1 NL4-3 and the MFI of CD25 expressed both on CD4+ T cells after the first PHA activation (R2 = 0.960, P = 0.003) and on CD4+ T cells after the fifth PHA activation (R2 = 0.838, P = 0.029) (Fig. 4A and B). The second significant correlation was obtained between the infectivity of HIV-1 AD8 in CD4+ T cells after the fifth PHA activation and the percentages of CCR5+ CD4+ T cells in this cell population (R2 = 0.787, P = 0.044) (Fig. (Fig.4C).4C). All comparisons of viral infectivity with the other tested variables, including those between the infectivity of HIV-1 NL4-3 and that of HIV-1 AD8, were not significant. Cross-comparison between the levels of activation and differentiation markers on the one hand and the expression of HIV-1 receptors on CD4+ T cells on the other revealed a negative logarithmic correlation between the percentages of CD25 and CCR5 both after one cycle (R2= 0.840, P = 0.0002) and after five cycles (R2 = 0.955, P = 0.0008) of PHA activation (Fig. 4D and E). Also, the MFI of CD25 correlated with the percentage of CCR5 both after one cycle (R2= 0.603, P = 0.023) and after five cycles (R2 = 0.555, P = 0.045) of PHA activation (not shown). Therefore, productive infection with the X4 variant HIV-1 NL4-3 seems to be closely related to the activation status of CD4+ T cells, whereas productive infection with the R5 variant HIV-1 AD8 is closely related to the expression of the CCR5 coreceptor. Most strikingly, high CCR5 expression levels were not compatible with a high activation status of CD4+ T cells determined by expression of the CD25 molecule.
We then examined whether the NRTI AZT and 3TC are more powerful blockers of HIV-1 replication in activated CD25+ CD62L+ T cells (the preferential target of X4 variant HIV-1 NL4-3) than in slowly dividing CD25− CD62L− T cells (the preferential target of R5 variant HIV-1 AD8) (Fig. (Fig.5;5; Table Table1).1). To this end, we infected the cultures of PBMCs submitted to one or five cycles of PHA activation with HIV-1 NL4-3 soon after PHA activation (6 or 8 days), when T cells expressed high levels of CD25 (MFI > 30), and determined productive viral infection by detection of intracellular p24gag. In parallel, we infected the same PBMC cultures with HIV-1 AD8 or 49.5 late after PHA activation (from 17 to 35 days), when T cells expressed CD25 at low levels (MFI < 8). The effects of different concentrations of AZT on productive infection of PBMCs submitted to one cycle of PHA activation and infected with HIV-1 NL4-3 and the effects of different concentrations of AZT on productive infection of PBMCs submitted to five cycles of PHA activation and infected with HIV-1 AD8 are shown in Fig. Fig.5.5. In addition, we took advantage of the lack of dependence of HDV pseudotyped by vesicular stomatitis virus protein G on the coreceptor and transduced cultures of PBMCs submitted to one or five cycles of PHA activation early or late after PHA activation, when T cells expressed high or low CD25 levels, respectively. In these experiments, the low IC50 levels of AZT or 3TC (<33 nM) correlated with the high MFI of CD25, whereas the high IC50 levels of AZT or 3TC (>113 nM) correlated with the low MFI of CD25. In the same experiments, the activity of the NNRTI nevirapine was independent of CD25 expression levels.
Activity of thymidine kinase in resting and activated T cells was estimated indirectly from the levels of thymidine kinase mRNA and from the pools of intracellular triphosphates. The levels of thymidine kinase mRNA were assessed by means of quantitative real-time reverse transcriptase PCR and standardized to mRNA of actin. In comparison with activated CD4+ T cells, these levels were reduced approximately 20 times in resting CD4+ T cells 29 days after the first PHA activation and approximately 47 times in CD4+ T cells 29 days after repeated cycles of PHA activation (Table (Table2).2). Similarly, the pool of intracellular dTTP was reduced approximately 17 times in resting CD4+ T cells 29 days after the first PHA activation, approximately 30 times in CD4+ T cells 18 days after repeated cycles of PHA activation, and approximately 80 times in CD4+ T cells 46 days after repeated cycles of PHA activation, in comparison with activated CD4+ T cells (Table (Table2).2). Thus, lower levels of thymidine kinase mRNA and dTTP were detected in metabolically weakly active resting memory CD4+ T cells than in activated CD4+ T cells.
Next, we examined whether the difference in the susceptibilities of the subsets of resting and activated CD4+ T cells to productive infection with X4 and R5 variants of HIV-1 observed with PBMC-derived CD4+ T cells also occurs in a system of human lymphoid tissue infected ex vivo, without exogenous activators or cytokines. Expression of Ki67, a nuclear antigen associated with all phases of the cell cycle (with the exception of G0), was used to determine whether CD4+ T cells present in lymphoid tissue are actively cycling, like CD4+ T cells from PBMCs subjected to repeated PHA activations. Approximately 40% of cycling CD4+ Ki67+ lymphocytes in noninfected control tissue obtained from seronegative donors were CD45RO+ CD62L− (Fig. (Fig.6A).6A). Approximately 42.5% of cycling CD45RO+ CD62L− cells were HLA-DR−. In contrast, only 8% of cycling CD4+ Ki67+ lymphocytes were CD45RO+ CD62L+. Memory T cells were defined in subsequent experiments by expression of the CD45RO molecule, since more than 95% of CD45RO+ cells in lymphoid tissue coexpressed the CD3 molecule on their surface. CD45RO+ T lymphocytes infected with HIV-1 NL4-3 and HIV-1 AD8 that were p24gag positive were analyzed for expression of HLA-DR, CCR7, and CD62L markers 12 days after infection of the lymphoid tissue (Fig. 6B to D). Productive HIV-1 NL4-3 infection predominated in HLA-DR+ CD45RO+ T cells (89%), whereas HIV-1 AD8 predominantly infected HLA-DR− CD45RO+ T cells (56%) (Fig. (Fig.6B).6B). Then, we examined productive infection of CD45RO+ cells, defined by expression of CCR7 (Fig. 6C and D) or CD62L (Fig. (Fig.6E)6E) molecules, which differentially expressed HLA-DR. HIV-1 NL4-3 productive infection predominated in HLA-DR+ CCR7+ and HLA-DR+ CD62L+ T cells (48.5% ± 8.2% in HLA-DR+ CD62L+ T cells compared with 23.9% ± 6.0% in HLA-DR− CD62L− T cells and 10.0% ± 2.2% in HLA-DR+ CD62L− T cells; n = 10). In contrast, HIV-1 AD8 predominantly infected CCR7− and CD62L− T cells that did not express HLA-DR (53.0% ± 4.9% in HLA-DR− CD62L− T cells compared with 15% ± 5% in HLA-DR+ CD62L+ T cells and 12.4% ± 3.0% in HLA-DR+ CD62L− T cells; n = 13).
As we did with PBMCs, we compared the levels of activity of the NRTI AZT in blocking productive infection by R5 and X4 HIV-1 variants in memory subsets that differentially expressed HLA-DR and CD62L. In these comparative experiments, AZT was highly active in blocking the infectivity of HIV-1 in activated HLA-DR+ CD62L+ T cells (IC50 with NL4-3 was 5 ± 3 nM) (Fig. (Fig.6B).6B). In contrast, AZT was weakly active in resting HLA-DR− CD62L− T cells (IC50 with AD8 was 741 ± 398 nM). Thus, viral replication in HLA-DR− CD62L− T cells was more resistant to AZT than in activated Tcells.
In contrast to the conventional wisdom that HIV-1 replicates primarily in activated cells, several recent papers have clearly demonstrated that simian immunodeficiency virus and HIV preferentially infect and deplete resting cells present in the gastrointestinal tracts of infected monkeys and humans (8, 33, 38, 44, 64). Here, we developed two model systems for the study of HIV-1 replication in HLA-DR− CD62L− CD4+ T cells, one based on CD4+ T cells from PBMCs repeatedly activated with PHA and the other based on the highly physiologically relevant ex vivo model of lymphoid tissue explants in which CD4+ T cells are exposed to HIV-1 in the absence of any exogenous cytokine or cell activator (20, 22, 25). Our results demonstrate that in both systems R5 variants of HIV-1 preferentially infect CD62L− CD4+ T cells that do not express activation markers such as CD25 and HLA-DR and are potentially resistant to NRTI in these cells. In comparison with complex lymphoid tissue, repeated activation of primary CD4+ T cells with a nonphysiological polyclonal stimulator, PHA, resulted in a better defined and more simple model, enriched by CD25− HLA-DR− CD45RO+ CD62L− CD4+ T cells that produce IFN-γ and IL-4 after adequate stimulation. The majority of these slowly proliferating cells remained in the G1b phase of the cell cycle. We have shown that infectivity of R5 variants of HIV-1 in this cell system correlates directly with level of expression of the CCR5 coreceptor. More interestingly, activation of CD4+ T cells monitored from expression levels of CD25 correlated negatively with expression of CCR5. Whereas recent activation of CD4+ T cells is a prerequisite for CCR5 expression (7, 36), CCR5 was expressed in our experiments only later after activation, reciprocally to downregulation of CD25. The exponential character of the negative correlation between the levels of CD25 and CCR5 suggests that small changes in the activation status of CD4+ T cells may result in significant downregulation of CCR5. Thus, the surprisingly high tropism of R5 variants of HIV-1 for slowly dividing CD4+ T cells could be related to the negative correlation between expression levels of CD25 and CCR5 molecules on CD4+ T cells. Actually, mucosal CD4+ T lymphocytes in human and monkey organisms appear to be derived from blood cells that have recently divided and then migrated into mucosa, where they lose expression of Ki67 (48, 71) and express high levels of the CCR5 receptor (8, 33, 44, 46, 50, 64).
Our model based on CD4+ T cells from PBMCs repeatedly activated with PHA provides a novel and well-defined very-long-term culture system to evaluate mechanisms for one of the major limitations of current antiretroviral therapy, i.e., persistence of a reservoir of virus in what is arbitrarily described as resting or quiescent CD4+ T cells. The dichotomous nomenclature of “resting” and “activated” and of “central” and “effector” memory CD4+ T cells is difficult to apply to in vitro-developed cell systems, and it may obscure our ability to understand the range of cellular states that affect HIV replication. In spite of seductive parallels between some phenotypic characteristics and the outcome of HIV-1 infection in this cell culture in vitro and in resting CD4+ TEM cells in infected organisms, the in vivo relevance of this cell system remains limited by the absence of immunobiological markers that would relate their expression to a real biological function and by the inherent problem of defining subsets of CD4+ T cells by markers such as HLA-DR and CD62L. Thus, in our model both “resting” CD45RO+ CD62L+ CD4+ T cells resistant to HIV-1 infection in nonstimulated PBMCs and somewhat more “activated” CD45RO+ CD62L+ CD4+ T cells permissive for HIV replication can be categorized as TCM cells on the basis of the indicated phenotypic markers. Analogically, whereas “resting” HLA-DR− CD62L− CD4+ TEM cells in peripheral blood are profoundly resistant to HIV infection, recent in vivo studies have shown that immunophenotypically similar mucosal “resting” HLA-DR− CD4+ TEM cells in the gastrointestinal tract are actually the preferential target of HIV-1 infection (8, 33, 38, 44, 64). The slowly dividing HLA-DR− CD62L− CD4+ T cells examined in our study mimic, in their phenotypic characteristics and sensitivity to infection with the R5 variants of HIV-1, “recently activated” resting CD4+ T cells in the gastrointestinal tract. However, it is possible that these repeatedly activated primary CD4+ T cells reflect selective survival and growth of cells with unusual properties that are not representative of CD4+ T cells in vivo. Use of some antigens that are less potent but more physiologic and specific than PHA, like tetanus toxoid or influenza virus antigen (31), could be tested in future experiments.
Whereas percentages of repeatedly stimulated cells infected with HIV-1 AD8 correlated with percentages of CCR5+ cells (Fig. (Fig.4C),4C), the absolute numbers of HIV-1 AD8-infected cells largely exceeded the numbers of CD4+ T cells that expressed CCR5. We suppose that, as with the macaque system (44), the levels of CCR5 protein expressed in these cells may be sufficient to render CD4+ T cells permissive to HIV infection but too weak to be detected in flow cytometry. Thus, no mechanism other than the CCR5 receptor-dependent one needs to be invoked to account for the infectivity of R5 variants of HIV-1 in repeatedly stimulated cells. In contrast to infectivity of R5 variants of HIV-1, infectivity of the X4 variant HIV-1 NL4-3 correlated directly with the level of activation status of CD4+ T cells (percentage of CD25+ cells) and showed only a trend in correlation with the expression levels of its coreceptor, CXCR4. During progression to AIDS, associated with persistent immune activation (28), X4 HIV-1 variants are frequently selected and massively produced, presumably by activated lymphoid tissue-addressed CD45RO+ CD62L+ CD4+ T cells, which are depleted from peripheral blood and sequestered in lymph nodes (3). Also, the recent study of Mengozzi et al. (45) has shown that an appropriate strength of T-cell activation plays a critical role in the outcome of HIV infection despite downregulation of HIV coreceptor and secretion of high levels of CCR5 ligands. Collectively, the viral infectivity does not need to exclusively reflect the receptor levels and efficiencies with which the virus enters the cell. The virus infectivity in the cell culture could be affected by several other variables, such as (i) differences in the production of progeny virions caused either by cellular differences or by viral-strain differences, (ii) differences in the growth rates of distinct cell populations in the cultures, and (iii) differences in cell death caused by distinct HIV-1 strains. Further studies will be necessary to elucidate these points. However, the outcomes of these studies would not change the major conclusions of our work, the demonstration that R5 variants of HIV-1 preferentially infect CD25− HLA-DR− CD62L− CD4+ T cells and are potentially resistant to NRTI in these cells.
The situation in the G1b phase of the cell cycle, required for completion of reverse transcription (32), and high levels of CCR5 expression seem to be necessary for a high efficiency of infection with R5 variants of HIV. Yet, slow proliferation is not sufficient for productive infection with HIV, as evidenced by resistance of PBMCs to HIV infection in vitro, although slowly proliferating (40) and Ki67 antigen-expressing (26) CD4+ TEM cells are present in peripheral blood. Appropriate stimulation provided by different soluble factors is necessary for productive infection of resting CD4+ T cells with HIV (15, 53, 59-61, 63, 65). Among them, IL-2, a cytokine necessary for the survival of cells in the cultures used in our experiments, is apparently an essential factor for productive infection of resting CD4+ T cells with HIV-1. Different mechanisms can modify an intracellular environment of resting memory CD4+ T cells in lymph nodes and gastrointestinal tissue that can favorably support HIV-1 replication. Recent transcriptome analysis has demonstrated that a number of genes involving transcription regulation, RNA processing and modification, and protein trafficking and vesicle transport are significantly upregulated in resting CD25− HLA-DR− CD4+ T cells in PBMCs of viremic patients compared with those of aviremic patients (14). Interestingly, HIV-1 infection itself, via production of the Nef and Tat proteins, activates both nuclear factor AT and nuclear factor κB, resulting in increased IL-2 secretion and T-cell priming (37, 39, 42, 66). These observations point to a close relationship between viral replication and T-cell activation (4, 6, 11, 13, 43, 45, 58).
Importantly, HIV-1 infection in slowly dividing HLA-DR− CD62L− CD4+ T cells is productive and is less sensitive to NRTI, which need phosphorylation, than in metabolically more active activated CD4+ T cells. Previous studies showed a lower potency of NRTI to block HIV replication in quiescent (G0) CD4+ T lymphocytes than in activated CD4+ T cells (18, 19, 55). However, HIV-1 infection of quiescent (G0) CD4+ T cells is nonproductive, and it is cleared with a half-life of about 1 day (9, 49, 54, 56, 63, 68). In contrast, slowly dividing (G1b) CD4+ T cells and HLA-DR− Ki67+ CD4+ T cells, the preferential target of the R5 variant of HIV-1 and the principal object of the present study, are infected productively. As with quiescent (G0) T cells (18, 55), slowly dividing CD62L− CD4+ T cells contain lower levels of thymidine kinase than do activated CD4+ T cells. The NNRTI nevirapine was equally efficient in both cell systems. Therefore, the relative resistance of HIV-1 replication in slowly dividing HLA-DR− CD62L− CD4+ T cells to NRTI could have a direct importance for antiretroviral therapy.
The demonstration of a lower potency of NRTI in stopping HIV replication in lymphoid tissue is particularly important because the HIV reservoir is formed predominantly in this compartment of the lymphatic system. CD4+ TCM and TEM cells are naturally present in the lymphoid tissue explants without any ex vivo manipulation (5, 8, 10). We have previously characterized HIV-1-infected T cells in lymphoid tissue from the standpoint of the central and effector memory phenotypes (24). In the present study, we investigated their activation and proliferation status. Expression of the Ki67 antigen shows that a significant proportion of TEM cells in lymphoid tissue proliferate, in a way similar to peripheral blood (26, 40). However, unlike peripheral blood, TEM cells are highly permissive to HIV-1 infection, like resting cells present in the gastrointestinal tracts of infected monkeys and humans (8, 33, 38, 44, 64).
Zhang and colleagues (70) have demonstrated propagation of HIV in resting CD4+ T lymphocytes in tonsils and lymph nodes of infected individuals and have shown relative resistance of this cell population to HAART. The same results were obtained with resting T cells from PBMCs of viremic patients (14, 27, 33, 34). Resistance of HIV-1 replication to antiretroviral drugs in resting memory CD4+ T cells of infected individuals highlights an urgent need to search for new HIV therapies. Here, we mimic the lower potency of antiretroviral drugs in blocking HIV replication in two laboratory models, one based on in vitro-infected individual T cells derived from PBMCs and the other based on ex vivo-infected lymphoid tissue.
We thank L. Margolis, J.-C. Grivel, D. Olive, and J. Nunez for helpful suggestions.
Our work was supported by grants from the French National Agency for AIDS Research (ANRS) and INSERM and by fellowships from the French Doctors (M.A.F.), the Ministère de la Recherche et de la Technologie (A.B.), the ACS-Sidaction (L.B., K.T., and M.P.), the Fondation pour la Recherche Medicale (M.P.), and CNOUS (J.B.).
Zidovudine, lamivudine, and monoclonal antibody anti-CCR5 (clone 3A9; produced by Pharmingen) were obtained through the AIDS Research and Reference Reagent Program from DAIDS, NIAID.
The authors have no conflicting financial interests.