HIV “Elite Controllers” Are Characterized by a High Frequency of Memory CD8⁺CD73⁺ T Cells Involved in the Antigen-Specific CD8⁺ T-cell Response

Abstract

Human immunodeficiency virus type 1 (HIV-1) infection is characterized by chronic immune activation and suppressed T-lymphocyte functions. Here we report that CD73, both a coactivator molecule of T cells and an immunosuppressive ecto-enzyme through adenosine production, is only weakly expressed by CD8⁺ T cells of HIV-infected patients and only partially restored after successful antiviral treatment. CD73 expression on CD8⁺ T cells correlates inversely with cell activation both ex vivo and in vitro. However, CD8⁺ T cells from HIV controllers (HICs), which spontaneously control HIV replication, express CD73 strongly, despite residual immune activation. Finally, we demonstrate that CD73 is involved in the HIV-specific CD8⁺ T-cell expansion. Thus, we show that CD73 is central to the functionality of HIV-specific CD8⁺ T cells and that the preservation of HIV-specific CD73⁺CD8⁺ T cells is a characteristic of HICs. These observations reveal a novel mechanism involved in the control of viral replication.

activation, CD73, CD8⁺ T-cell response, HIV, HIV controller

The acute phase of human immunodeficiency virus (HIV) infection is characterized by rapid immune activation, the expansion of HIV-specific CD8⁺ T cells, and an associated decrease of plasma viremia [1]. During the chronic phase, CD8⁺ T-cell function is altered, and a high level of viral replication occurs [2]. This absence of viral control is due, in part, to the exhaustion of the immune system, an indirect consequence of both massive CD4 depletion and systemic immune activation [3]. Viral replication associated with the loss of immune control also occurs in the gut [4], where most of the body's immune cells are localized [5]. A state of apparent durable control of HIV replication becomes established in a very small percentage (<1%) of untreated infected patients, called HIV controllers (HICs) [6]. The involvement of the CD8⁺ T-cell response seems to be important for the containment of HIV in HICs [7]. In many of these patients, CD8⁺ T cells can proliferate, produce perforin and numerous cytokines following HIV antigen stimulation [8, 9], and are able to inhibit viral replication when cocultured with infected CD4⁺ T cells [10,–11]. Recently, a genome-wide association analysis confirmed that the major genetic determinants related to the natural control of HIV are those affecting the structure of class I antigen-presenting molecules [12]. These data implicate CD8 T cells in the control of HIV, but because these genotypes are not restricted to HICs, the exact mechanism remains unclear.

CD73, expressed by immune and endothelial cells [13], is a glycosyl-phosphatidylinositol (GPI)–anchored cell surface molecule that hydrolyzes extracellular nucleoside monophosphates, such as adenosine monophosphate (AMP), into adenosine [14]. Adenosine inhibits T-cell proliferation [15, 16] and impairs the cytokine production and cytotoxicity of activated T cells [17]. CD73 may also promote lymphocyte adhesion to endothelial cells [18, 19]. It is also a costimulatory molecule, delivering potent activation signals in T cells [20, 21]. Thus, CD73 may have a dual role in the modulation of T-cell immune responses. Consistent with a complex function, CD73 expression is regulated by the endogenous GPI phospholipase D (GPI-PLD), responsible for the cleavage and release of various GPI-anchored membrane receptors [22–24]. Although abundant in the serum, the GPI-PLD is active in the intracellular compartment [25, 26] but not in the circulation or extracellular fluids [27, 28].

Before CD73 was characterized, it had been reported that the 5-ecto-nucleotidase activity at the surface of CD8⁺ T-cells was lower in HIV-infected patients than controls [29, 30]. It has been shown very recently that this downregulation of CD73 on CD8⁺ T cells correlates with immune activation and leads to functional deficits in HIV infection, as sorted CD73⁺CD8⁺ T cells show a higher cytokine production and a higher proliferative capacity after HIV-specific stimulation than CD73^–CD8⁺ T cells [31]. However, the precise role of CD73 and its involvement in CD8⁺ T-cell function during HIV infection have not been fully described. Here, we first compared the expression of CD73 on peripheral blood mononuclear cells (PBMCs) and gut biopsies in different groups of HIV-infected patients and healthy subjects, analyzed the correlations with clinic and immune activation parameters, and then focused on the effect of in vitro downregulation of CD73 using either anti-CD73 antibody or siRNA on CD8 T-cell response. Our original data show that CD73 plays a key role in the function of HIV-specific CD8⁺ T cells and that the preservation of a population of HIV-specific CD73⁺CD8⁺ T cells is a characteristic of HICs.

METHODS

Human Subjects

Blood samples from HIV-1- (Table 1) and Epstein–Barr virus (EBV)–infected subjects and healthy donors (controls) were obtained at Henri Mondor Hospital, and the Etablissement Français du Sang, Créteil, France, respectively. HICs were selected as HIV-1–infected individuals with a plasma viral load under 300 copies/mL for more than 10 years in the absence of treatment, as previously defined [32]. Gut biopsies (Table 2) were obtained from 14 HIV-1–infected patients by rectoscopy at “Centre Hospitalier Regional d'Orléans.” Gut biopsies from 6 HIV-1–negative patients who underwent colonic surgery were studied as controls. Ethical committee approval and written informed consent from all subjects were obtained before the start of the study.

Table 1.

Characteristics of HIV-1–Infected Patients

Groups of HIV-1–Infected Patients	Chronic Nontreated (c-ART−)	Chronic Treated (c-ART+)	HIV Controllers (HICs)
Number of patients	16	16	7
Plasma viral load (log10 copies/mL)	5 ± 0.3	<1.7	1.9 ± 1.2
CD4+ T cells (cells/µL)	309 ± 24	607 ± 75	741 ± 112
Infection duration (years)	3.2 ± 1	13.4 ± 2	13.9 ± 3
Treatment duration (years)	N/A	9.4 ± 1	N/A

Abbreviations: c-ART, combined antiretroviral therapy; HIV-1, human immunodeficiency virus type 1; N/A, not applicable; SEM, standard error of margin.

Values mentioned represent the mean ± SEM.

Table 2.

Clinical Characteristics of HIV-1–Infected Patients for Gut Biopsies

Groups of HIV-1 Infected Patients	Chronic Nontreated (c-ART−)	Chronic Treated (c-ART+)
Number of patients	3	12
Plasma viral load (log10 copies/mL)	3.9 ± 0.1	UND (<1.7)
CD4+ T cells (cells/µL)	424 ± 71	623 ± 80
Infection duration (years)	6.3 ± 3.4	9.9 ± 2
Treatment duration (years)	N/A	6.2 ± 1

Abbreviations: c-ART, combined antiretroviral therapy; HIV-1, human immunodeficiency virus type 1; UND, undetectable; N/A, not applicable; SEM, standard error of margin.

Values mentioned represent the mean±SEM.

Gut and Blood Lymphocytes Isolation

Lymphocytes were prepared from gut biopsies as described previously [33]. PBMCs were isolated by centrifugation with lymphocyte separation medium (cell culture company: PAA) from fresh blood samples collected onto ethylenediaminetetraacetic acid. PBMCs were cryopreserved in fetal calf serum/10% dimethyl sulfoxide in liquid nitrogen or used directly.

Antibodies and Flow Cytometry

The antibodies used were: anti-CD8-FITC, -PerCP, anti-CD73-PE (clone AD2), anti-CD3-alexa-700, anti-CD4-PacBlue, anti-CD38-APC, anti-CD28-PerCP-Cy5.5 (BD Biosciences), and anti-CD45RA-ECD (Beckman Coulter). The LIVE/DEAD fixable dead cell stain kit (Molecular Probes, Invitrogen) was used in all stainings to exclude dead cells from the analysis. Intracellular staining assays were performed using the Perm2 buffer according to the manufacturer's instructions and with PE-Cy7-conjugated interferon γ (IFN-γ), tumor necrosis factor α (TNF-α), and interleukin 2 (IL-2) antibodies (BD Biosciences). The specific response was evaluated as the percentage of live CD3⁺CD8⁺ cells producing at least 1 of the 3 cytokines (IFN-γ, TNF-α or IL-2). CD107a-FITC (BD Biosciences) staining was performed after long-term culture (7 days) by direct incubation during the restimulation of cells with specific antigens and GolgiStop for 5 hours according to the manufacturer's instructions (BD Biosciences). Cytometer acquisition was performed on LSR II (BD Biosciences), and data were analyzed with Diva (Version 6.1; BD Biosciences) or FlowJo (Version 7.6.5; TreeStar) software.

Cell Culture and Peptide Stimulation

Magnetic beads (Miltenyi Biotec) were used according to the manufacturer's instructions for cell purification. Cultures were performed in 48-well plates with 2.10⁶ thawed PBMCs and cocultures in 96-well plates with 50.10³ CD14⁺ and 200.10³ CD8⁺ cells purified from fresh PBMCs. Cells were incubated with anti-CD73 (clone 4G4, Hycult Biotechnology) or mouse IgG1 antibodies (Ab-1, clone NCG01, Abcam), both at 1 µg/mL on days 0 and 3. A pool of whole HIV-1 Gag 15-mer peptides (1 µg/mL) or a mix of specific peptides from cytomegalovirus, EBV, and influenza virus (CEF pool) at 1 µg/mL was used for stimulation. On day 7, cells were restimulated for 5 hours with the same peptides in the presence of Brefeldin A (10 µg/mL), and then cell surface and intracellular staining were performed as described above.

siRNA Experiments

A total of 10⁷ cells/mL were incubated in Accell siRNA delivery serum-free medium either with scramble (scr) or specific siRNA (Accell SMART pool, Dharmacon, Abgene Ltd) at 10 µM for 5 hours at 37°C. Then, 200.10³ cells/well were cultured in 96-well plates (precoated or not with 2 µg/mL anti-CD3 mAb [UCHT1, Beckman Coulter]) in 100 µL Accell siRNA delivery medium containing 3% fetal bovine serum (Gibco). Both specific siRNA and scr were re-added at 1 µM every 2 days. For GPI-PLD–specific siRNA experiments, CD8⁺ cells were purified after 5 days of culture for cytometry. CD73-specific siRNA experiments with purified CD8⁺ cells were continued for 15 days before peptide stimulation in coculture with CD14⁺ cells as described above.

mRNA Quantification

Total RNA was isolated using Trizol reagent (Invitrogen) according to the manufacturer's instructions. The AffinityScript QPCR cDNA Synthesis Kit (Agilent Technologies) was used for reverse transcription by the supplier's protocol with random primers. Polymerase chain reaction amplification was performed with the Brilliant II SYBR Green QPCR Master Mix (Agilent Technologies) and 5 ng of each primer on a Mx3005 QPCR machine (Agilent Technologies). The housekeeping genes included for reference were 28S rRNA for in vitro lymphocyte stimulation [34], and both 28S rRNA and S14mRNA for ex vivo analysis. The 2^−ΔΔCt method was used, and results are expressed in arbitrary units. Primer sequences were (forward/reverse):

CD73 (GCAACAATGGCACAATTAC)/(ATGCTCAAAGGCCTTCTTCA), CD38 (ACCCTGGAGGACACGCTGCTA)/(TCTGCAAACCTGCGGGAAACCG), GPIPLD (TCACATGGCGGCAGATGTCAGC)/(TCACCAGCCGAATGAGCCTCTGA), 28S (TTGAAAATCCGGGGGAGAG)/(ACATTGTTCCAACATGCCAG), S14 (GGCAGACCGAGATGAATCCTC)/(CAGGTCCAGGGGTCTTGGTCC).

Statistical Analysis

GraphPad Prism (Version 5.03) was used for statistical analyses, and the tests used are given in the corresponding figure legends.

RESULTS

Frequency of CD73⁺CD8⁺ T Cells is Higher in HICs Than Other HIV-1–Infected Patients

We studied CD73 expression on CD8⁺ T cells from different groups of HIV-1–infected individuals and in healthy donors (control) (Figure 1A). CD73⁺CD8⁺ T cells were less numerous in combined antiretroviral therapy nontreated (c-ART–) patients than controls (6.9% ± 1.1% vs 43.7% ± 4.8%, respectively; P < .01), and the value for combined antiretroviral therapy treated (c-ART+) patients was intermediate (37% ± 2.8%). These cells were more abundant in HICs (61.4% ± 6.7%) than in any of the other groups (P < .01 for all HIV-1–infected groups, and P = .04 for controls).

Figure 1.

The CD73⁺CD8⁺ T-cell frequency in HICs is higher than those in c-ART– or c-ART+ HIV-1–infected patients and controls. Expression by flow cytometry of CD73 on CD8⁺ T cells in 3 different groups of HIV-1–infected patients: nontreated (c-ART–, n = 16), treated (c-ART+, n = 16) and HIV controllers (HICs, n = 7), and in a group of healthy donors (controls, n = 12). A, Expression of CD73 on the CD8⁺ T-cell population. B, Expression of CD73 on CD8⁺ T-cell subsets: naive (CD45RA⁺CD28⁺), memory (CD45RA^–CD28⁺ and CD45RA^–CD28^–) and terminal effector (CD45RA⁺CD28^–) cells. C, CD73⁺CD8 absolute cell count in nontreated (c-ART–, n = 16), treated (c-ART+, n = 16), and HIV controllers (HICs, n = 7). Results are expressed as percentages of cells expressing CD73 (A and B) or as number of cells/µL expressing CD73 (C). Horizontal bars and histograms represent mean values. Error bars represent SEM. Statistical analyses were performed with the unpaired t test (*P < .05, **P < .01, ***P < .001). Abbreviations: c-ART, combined antiretroviral therapy; HIC, HIV controllers; HIV-1, human immunodeficiency virus type 1; SEM, standard error of measurement.

Next, we studied expression of CD73 on the CD8⁺ T-cell population at various stages of differentiation: naive (CD45RA⁺CD28⁺), memory (CD45RA^–CD28⁺ and CD45RA^–CD28^–) and terminal effectors (CD45RA⁺CD28^–) (Figure 1B and Supplementary Figure 1). As previously shown [35], CD73 was mainly expressed by naive CD8⁺ T cells. The frequency of CD73 in each subset was significantly lower in c-ART– than control subjects (39.5% ± 5.4% vs 83.1% ± 3.9% for naive, 12.4% ± 2.2% vs 35.3% ± 3% for memory CD45RA^–CD28⁺, 0.8% ± 0.1% vs 16.9% ± 2.9% for memory CD45RA^–CD28^–, and 5% ± 1.3% vs 41.4% ± 5.9% for terminal effectors; P < .001 for all comparisons). Under antiviral treatment, the frequency of CD73⁺CD8⁺ T cells was found to be restored in every T-cell subset except in terminal effectors (25.6% ± 4.5% vs 41.4% ± 5.9% in controls; P = .05). In HICs, the frequency of CD8⁺ T cells expressing CD73 was higher in every subset. This difference with HIV-1–infected patients was substantial (P < .01 for all comparisons) and that with controls was most pronounced for the CD45RA^–CD28⁺ memory subset (55% ± 5.3% vs 35.3% ± 3%, respectively; P < .01). Then, we conclude that the high expression of CD73 on whole CD8⁺ T cells was due to the memory CD45RA^–CD28⁺ subset. The CD45RA^–CD28⁺ subset represented around 30% of CD8⁺ T cells in all groups of HIV patients and controls (data not shown). Beside the frequency within CD8⁺ T-cell population or subsets, we assessed the analysis of absolute CD73⁺CD8 cell count (Figure 1C). As it was for the frequency, the absolute CD73⁺CD8 cell count was found lower in c-ART– compared to c-ART+ (72 ± 13 vs 179 ± 30 cells/µL, respectively; P < .01) and was found higher in HICs compared to c-ART+ (288 ± 38 vs179 ± 30 cells/µL, respectively; P < .05).

Then, we investigated the expression of CD73 on CD8⁺ T cells from gut biopsies from c-ART– and c-ART+ patients (Table 2). A group of HIV-1–negative patients who underwent colonic surgery was used as control (Figure 2). The CD73⁺CD8⁺ T-cell frequency was lower in c-ART– patients than in controls (8.7% ± 0.6% vs 45.2% ± 5.5%, respectively; P < .01; Figure 2A) and in c-ART+ patients than controls (27.3% ± 2% vs 45.2% ± 5.5%, respectively; P < .05). CD73 expression was then evaluated on memory (CD45RA^–CD28⁺ or CD28^–) CD8 subsets within the gut. First, we found that, at the opposite of the blood, the memory (CD45RA^–) subset was the main CD8 subset present in the gut with more than 80% of whole CD8⁺ T cells in all groups of subjects. This memory CD8⁺ T cells consisted of CD28⁺ cells at 60% in controls, 54% in c-ART+, and 26% in c-ART–, the cells in this last group being more differentiated toward the CD28^– phenotype, certainly due to the presence of virus (data not shown). Concerning CD73, its expression on CD45RA^–CD28⁺CD8⁺ T cells was lower in c-ART– and c-ART+ than in controls (9.2% ± 1.6% and 24.9% ± 1.5% vs 39.9% ± 5.5%; P < .01 and <.05, respectively) (Figure 2B). The same results were found on CD45RA^–CD28^–CD8⁺ T cells (data not shown). The frequencies of CD73⁺ on CD45RA^–CD28⁺CD8⁺ T cells in PBMCs and gut were positively correlated (P < .001; R = +0.82) (Figure 2C). These results suggest that during HIV infection, there is a deficit of CD73⁺CD8⁺ T cells rather than a delocalization of the cells in the gut. However, we cannot rule out that these cells are localized in others peripheral lymphoid tissues. z

Figure 2.

CD73 expression on CD8⁺ T cells in gut biopsies from HIV-1–infected patients. A, Representative dot plots of CD73⁺ and CD8⁺ T cells in 1 HIV-, 1 c-ART–, and 1 c-ART+ patients. B, Frequency of CD73 on CD45RA^–CD28⁺CD8⁺ T cells in gut biopsies from HIV- (n = 6), c-ART– (n = 3), and c-ART+ (n = 12) patients. C, Correlations between the frequency of CD73 on CD45RA^–CD28⁺CD8⁺ T cells in PBMCs and gut biopsies from c-ART– (n = 3) and c-ART+ (n = 12) patients. Histograms report mean values and error bars represent 1 SEM from the mean. Statistical analyses were performed with the unpaired t test (*P < .05, **P < .01). Pearson's correlations were used for bivariate correlations. Abbreviations: c-ART, combined antiretroviral therapy; HIV-1, human immunodeficiency virus type 1; PBMCs, peripheral blood mononuclear cells; SEM, standard error of measurement.

Frequency of CD73⁺CD8⁺ T Cells Is Inversely Correlated With Immune Activation but Remains High in HICs Despite Residual Cell Activation

In c-ART– patients, CD73 expression by CD45RA^–CD28⁺CD8⁺ T cells was inversely correlated with plasma viral load (log₁₀ copies/mL) (P < .01; R = −0.67) (Figure 3A) and positively correlated with CD4⁺ T-cell counts (P < .05; R = +0.55) (Figure 3B). Accordingly, the frequencies of CD73⁺ and CD38⁺ on CD45RA^–CD28⁺CD8⁺ T cells (Figure 3C and 3D) were inversely correlated, both in c-ART– (P < .01; R = −0.80) and c-ART+ individuals (P < .01; R = −0.78), whereas no negative correlation was found in HICs (P = .05; R = +0.81). We also assessed correlation analysis between the same parameters (viral load, CD4 absolute count, CD38 expression) and CD73⁺ expression either on whole CD8⁺ T cells or whole memory (CD45RA^–) CD8⁺ T cells in the different groups of patients. In c-ART–, only negative correlations between viral load and CD73⁺ on CD8⁺ T cells and CD45RA^–CD8⁺ T cells were found significant. In c-ART+, only a negative correlation between CD73 and CD38 expressions on CD45RA^–CD8⁺ T cells was found (data not shown). Thus, we followed our investigation on the CD45RA^–CD28⁺CD8⁺ T-cells subset. We next investigated CD73 expression by activated (CD38⁺) and nonactivated (CD38^–) CD45RA^–CD28⁺CD8⁺ T cells in HICs and c-ART+ patients (Figure 3E). The expression of CD73 was greater on CD38^– than CD38⁺CD8⁺ T cells for both groups (27.7% ± 3.5% vs 13% ± 2.5% in c-ART+, and 50.7% ± 5.5% vs 35.5% ± 5.1% in HICs; P = .05 for both comparisons). However, CD73 remained more abundant in HICs than c-ART+, both on CD38^– and CD38⁺ CD45RA^–CD28⁺CD8⁺ T cells (P < .05 and <.01, respectively).

Figure 3.

CD8⁺CD73⁺ T-cell frequency is inversely correlated with chronic immune activation during HIV-1 infection. A, Correlation between CD73⁺ frequency on CD45RA^–CD28⁺CD8⁺ T cells and plasma viral load (log₁₀ copies/mL) in c-ART– patients (n = 15). B, Correlation between CD73 frequency on CD45RA^–CD28⁺CD8⁺ T cells and CD4⁺ T-cell count (cells/µL) in c-ART– patients (n = 15). C, Correlation between CD73⁺ and CD38⁺ frequencies on CD45RA^–CD28⁺CD8⁺ T cells in c-ART– (n = 14). D, Correlation between CD73⁺ and CD38⁺ frequencies on CD45RA^–CD28⁺CD8⁺ T cells in c-ART+ patients (n = 14) and HICs (n = 6). E, CD73⁺ frequency on CD38^– (open circles) or CD38⁺ (black circles) CD45RA^–CD28⁺CD8⁺ T cells in c-ART+ patients (n = 6) and HICs (n = 4). F, Representative flow cytometry dot plot of CD73 versus CD38 on CD8⁺ T cells after 0, 2, and 4 days of anti-CD3 stimulation of PBMCs from c-ART+ (n = 3) and HIC patients (n = 3). Pearson's correlations and unpaired or paired (linked dots) t tests were used (*P < .05, **P < .01). Abbreviations: c-ART, combined antiretroviral therapy; HICs, HIV controllers; HIV-1, human immunodeficiency virus type 1; PBMCs, peripheral blood mononuclear cells; SEM, standard error of measurement.

We next investigated changes in CD73 expression following direct activation of CD8⁺ T cells in vitro (Figure 3F): anti-CD3 activation of CD8⁺ T cells from c-ART+ and HIC individuals led to acquisition of CD38 expression and concomitant loss of CD73 expression. The results for fresh PBMCs from controls activated with phytohemagglutinin were similar (data not shown). We also tested whether this negative correlation between activation and CD73 expression could be found in other infections than HIV. We phenotyped CD8⁺ T cells in individuals with primary EBV infection, which is characterized by strong immune activation [36] (Supplementary Figure 2A and 2B). The frequency of memory CD8⁺CD73⁺ T cells was inversely correlated with the frequency of memory CD8⁺CD38⁺ T cells. In conclusion, the loss of CD73 from memory CD8⁺ T cells correlates with cell-activation markers during both HIV and EBV infections, and is directly due to cell-activation pathways.

Downmodulation of CD73 protein upon anti-CD3 activation in vitro was associated with a decrease of CD73 mRNA level (1 arbitrary unit for nonstimulated CD8⁺ T cells vs 0.27 ± 0.15 arbitrary units after anti-CD3 activation; P < .05, Supplementary Figure 2C). We also studied GPI-PLD, a membrane enzyme involved in the release the CD73 by cleavage of the GPI anchor (Supplementary Figure 2D). The frequency of CD8⁺CD73⁺ T cells was higher in samples incubated with a GPI-PLD–specific siRNA than with the scramble control, both in resting conditions (61.1% ± 7.8% vs 53.3% ± 9.4%; P < .05) and after anti-CD3 stimulation (53.3% ± 6% vs 32.9% ± 4.7%; P < .05). Thus, CD73 expression is regulated both at the transcriptional level during cell activation and by the cleavage of the GPI anchor under the control of GPI-PLD.

However, as CD73 baseline expression on memory CD8⁺ T cells appears to be higher in HICs than other individuals, we wondered whether the cells of HICs could present a lower activation level by measuring CD38 frequency in CD45RA^–CD8⁺ T cells and CD38 mRNA in CD8⁺ T cells from HICs and c-ART+ patients. The frequency of CD38⁺CD45RA^–CD8⁺ T cells was higher in HICs than c-ART+ (12.2% ± 1.8% vs 8.3% ± 1.2%). In term of mRNA, CD38 mRNA level was significantly higher in CD8⁺ T cells from HICs than those from c-ART+ (9.1 ± 1.7 vs 4.9 ± 1 arbitrary units; P < .05). Taking in consideration that the frequency of CD45RA^– on activated CD8⁺ T cells was equivalent in both populations of patients and that no difference was found in term of CD38 frequency on CD45RA⁺CD8⁺ T cells, this difference of CD38 mRNA level must be driven by the memory subset (data not shown). Hence, these results suggest that the higher expression of CD73 on CD8⁺ T cells from HICs is not due to a lower level of T-cell activation.

As the level of CD73 expression depended on the mRNA levels of both CD73 and GPI-PLD, we assayed both mRNAs in the few number of HIC samples available (n = 4) and in c-ART+ samples (n = 10). CD73 transcripts were more abundant in CD8⁺ T cells from HIC individuals than T cells from c-ART+ patients (9.3 ± 2.5 vs 4.0 ± 0.7 arbitrary units, respectively; P < .05), whereas no difference was found between groups for GPI-PLD mRNA (data not shown). In conclusion, the strong CD73 expression in HICs seemed to be due to high levels of CD73 mRNA, despite the residual cell activation.

In Vitro Downmodulation of CD73 Is Associated With a Decrease of the Specific CD8⁺ T-cell Response

We studied the role of CD73 in Gag-specific CD8⁺ T-cell responses. We first analyzed the frequency of memory HIV-specific CD8⁺ T cells expressing CD73 in HIC and c-ART+ patients. Following overnight stimulation of PBMCs with Gag peptides, the frequency of Gag-specific CD45RA^–CD28⁺CD8⁺ T cells expressing CD73 and producing at least 1 of the 3 cytokines (IFN-γ, TNF-α, or IL-2) was higher in HICs than in c-ART+ (0.15% ± 0.02% vs 0.09% ± 0.02%, respectively; P = .05) (Figure 4A), whereas no difference was found in the frequency of Gag-specific CD45RA^–CD28⁺CD8⁺ T cells not expressing CD73 (data not shown).

Figure 4.

Downmodulation of CD73 in vitro decreases Gag- and CEF-specific CD8⁺ T-cell responses. A, Frequency of CD73⁺ Gag-specific cells among CD45RA^–CD28⁺CD8⁺ T cells after overnight stimulation with Gag peptides in c-ART+ (n = 10) and HICs (n = 7). B, Gag-specific CD8⁺ T cells after 7 days’ stimulation with Gag peptides in the presence of anti-CD73 antibody (clone 4G4) or an isotype control in samples from c-ART+ patients (n = 5) and HICs (n = 4). Cells expressing at least 1 of the 3 cytokines (IFN-γ, TNF-α, or IL-2) after Gag stimulation were considered to be Gag-specific. Unpaired or paired (linked dots) t tests were used for statistical analyses (*P < .05, **P < .01). Effect of in vitro CD73 downmodulation on CEF-specific CD8⁺ T-cell response (C, D, E). C, Frequency of CEF-specific CD8⁺ T cells in the presence of anti-CD73 antibody (clone 4G4) or an isotype control (n = 4) after 7 days of coculture of purified CD8⁺ and CD14⁺ cells in the presence of CEF peptides. D, Frequency of CD107a⁺CD8⁺ T cells in the same experimental conditions as in (C) (n = 4). E, CD8⁺ T cells were precultured for 15 days in the presence of a pool of specific CD73 siRNA or a pool of nonrelevant siRNA (scramble). The frequency of CD73 on CD8⁺ T cells at this time point is presented on flow cytometry histograms (upper part). Then, CD8⁺ T cells were cocultured in the presence of purified CD14⁺ cells from the same donor. The frequency of CEF-specific CD8⁺ T cells are presented in the dot plots (lower part). Cells expressing at least 1 of the 3 cytokines (IFN-γ, TNF-α, or IL-2) were considered to be CEF-specific. Statistical analyses were performed with the paired t test (*P < .05). Abbreviations: c-ART, combined antiretroviral therapy; CEF, cytomegalovirus, EBV, and influenza virus; HICs, HIV controllers; HIV-1, human immunodeficiency virus type 1; IFN-γ, interferon γ; IG, immunoglobin; IL-2, interleukin 2; PBMCs, peripheral blood mononuclear cells; SEM, standard error of measurement; TNF-α, tumor necrosis factor α.

PBMCs from c-ART+ and HICs were stimulated with Gag peptides in the presence of the anti-CD73 mAb clone 4G4, which reduced CD73 expression on T cells by 90% [13]. In control conditions, the frequency of HIV-specific CD8⁺ T cells expressing at least 1 of the 3 cytokines (IFN-γ, TNF-α, or IL-2) was significantly higher in HICs than c-ART+ (6.2% ± 0.8% vs 2% ± 0.6%, respectively; P < .01) (Figure 4B). These frequencies were significantly lower in both groups of patients in the presence of anti-CD73 mAb than in isotype control (2.6% ± 0.6% and 1.5% ± 0.4%, respectively, in HIC and c-ART+ individuals; P < .05 for both comparisons); the frequency of CEF (mix of peptides from cytomegalovirus, EBV, and influenza virus)–specific CD8⁺ T cells was also lower in c-ART+ samples (2.7% ± 1.5% with anti-CD73 mAb vs 4.5% ± 2.4% with isotype control; P = .05) and HICs (2% ± 0.6% vs 5.9% ± 3%, respectively; P = .05) (data not shown). These results implicate CD73 in antigen-specific T-cell expansion.

However, CD73 is expressed by approximately 70% of the B cells and 10% of the CD4⁺ T cells in PBMCs [37]. To avoid any bystander effect due to CD73 expressed on non-CD8⁺ T cells, we repeated the experiments as described above but using purified CD8⁺ T cells from controls cocultured with autologous CD14⁺ antigen-presenting cells. CD14⁺ T cells did not express CD73. This confirmed the findings for PBMCs showing a lower frequency of CEF-specific CD8⁺ T cells after CD73 downmodulation (0.9% ± 0.4% vs 5.6% ± 2%, with anti-CD73 mAb and isotype control, respectively; P = .05) (Figure 4C). We also quantified the cell-surface expression of CD107a in the same experimental conditions, as an indirect measure of degranulation. The percentage of cells expressing CD107a was also lower after anti-CD73 downmodulation than in controls (1.3% ± 0.6% vs 5.8% ± 3.1%, respectively; P < .05) (Figure 4D). We performed a similar coculture experiment with a pool of CD73-specific siRNA. The expression of CD73 on CD8⁺ T cells and the CEF-specific CD8⁺ T-cell response were both lower following treatment with CD73-specific than scramble siRNA (Figure 4E).

These various results indicate that the costimulatory function of CD73 is involved in antigen-specific T-cell responses against recall antigens.

DISCUSSION

We report that both circulating and in situ CD73⁺CD8⁺ T cells are downregulated in HIV-infected patients in correlation with immune activation. The association between CD73 expression and the level of cell activation is in agreement with the observation of a significant reduction of 5′-nucleotidase activity in rat lymphocytes after concanavalin A stimulation [38]. It is also consistent with work on CD4⁺ T cells showing that the CD4⁺CD73⁺ T-cell frequency and number are decreased and inversely correlated with CD4⁺CD38⁺DR⁺ during HIV infection [39]. During the time this work has been finalized, another team has published similar results by screening the CD73 expression on different T-cells subsets of HIV patients [31]. However, they did not show the functional impact of CD73 deficiency by using functional assay based on in vitro–mediated CD73 downregulation.

Here, we show, for the first time, that CD73 is involved in the expansion of antigen-specific CD8⁺ T cells. This is in accordance with the observation of a significant correlation between in vitro proliferative capacities in response to mitogens and the number of CD73 molecules per CD73⁺CD8⁺ lymphocyte in patients with primary immunoglobulin deficiency [40]. Moreover, CD73 can protect against TNF-related apoptosis-inducing ligand (TRAIL)–induced apoptosis [41] and the concentration of plasma-soluble TRAIL increases during the acute phase of HICs [42], where the viral load is found detectable in HICs (for approximately 6 months after primary HIV-1 infection) [43]. Then, we suggest that CD73 may play a key role during the acute phase in HICs by contributing to the establishment of a strong anti-HIV memory CD8⁺ T-cell response leading to the control of viral infections other than HIV.

Regarding HLA-B27 (or -B57) genotypes, which are the major genetic determinants related to the natural control of HIV [12], we did not find any correlation between these genotypes and CD73 expression in HICs (data not shown). Nevertheless, all HICs did not express this phenotype and it has been shown that HICs could better control other chronic infections, including HCV infection, and this independently of HLA-B27 or -B57 [44]. These results suggest that mechanisms other than restricted HLA could contribute to the broad protection of HICs against viral infections.

By contrast, in the context of cancer, CD73 expressed by both tumor and host immune cells has been well established to be harmful [45]. The mechanism proposed is that CD73, by generating adenosine within the antitumor microenvironment, inhibits antitumor-specific T-cell function. However, we suggest that in the context of generalized immune activation and inflammation, the balance is tipped toward the coactivation and away from the enzymatic function of CD73. Nevertheless, this balance is presumably dependent on cell type and concentration of purinergic molecules in the microenvironment: we recently showed that regulatory T cells/CD39⁺ cells suppress IL-2 expression by activated CD4⁺ T cells in a medium supplemented with adenosine triphosphate [46].

In conclusion, during HIV infection, CD73 expression on CD8⁺ T cells decreases substantially due to their high level of activation. This decrease is likely to be involved in the reduced expansion capacity of HIV-specific CD8⁺ T cells, as a consequence of the loss of the coactivator function of CD73. Therefore, the control of the viral replication in HICs by the CD8⁺ T-cell immune response is likely to involve the overexpression of CD73⁺CD8⁺ T cells.

Notes

Acknowledgments. We thank the Mondor Immunomonitoring Center (MIC) for providing technical facilities.

Financial support. This work was supported by the Agence Nationale de Recherche contre le Sida et les Hépatites Virales (ANRS).

Potential conflicts of interest. All authors: No reported conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

a

Vaccine Research Institute.

b

Present affiliation/address: Chronic Viral Illnesses Service of the McGill University Health Center, Montreal, QC, Canada.

Presented in part: 19th Conference on Retroviruses and Opportunistic Infections (CROI 2012), Seattle, Washington, 5–8 March 2012.