Impaired NFAT nuclear translocation results in split exhaustion of virus-specific CD8⁺ T cell functions during chronic viral infection

Functional Competence of LCMV-Specific CD8⁺ T Cells After Acute or Chronic Infection.

To compare the functional capacities of LCMV-specific CD8⁺ T cells after acute/resolved (low-dose, l.d.) or chronic (high-dose, h.d.) infection, we analyzed their ability to degranulate and to produce IFN-γ, TNF-α, or IL-2 (Fig. 1A). LCMV-specific CD8⁺ T cell numbers and viral titers are shown in supporting information (SI) Fig. 6. The majority of gp33- and np396-specific CD8⁺ T cells from l.d.-infected mice were able to degranulate and to produce IFN-γ and TNF-α. In contrast, gp33- and np396-specific CD8⁺ T cells from h.d.-infected mice were incapable of producing significant amounts of IFN-γ, TNF-α, or IL-2. To our surprise, LCMV-specific CD8⁺ T cells retained their degranulation capacity. Comparable results were obtained for lymphocytes isolated from lymph node and peripheral tissues (liver and lung, not shown). Programmed death 1 (PD-1) is highly expressed on virus-specific CD8⁺ T cells during chronic infections, and signaling through PD-1 might be implicated in CD8⁺ T cell dysfunction (29–32). We therefore tested whether blocking of PD-1/PD-L1 or cytotoxic T lymphocyte antigen (CTLA)-4/B7 interactions during restimulation would lead to increased cytokine production. Both PD-1 and CTLA-4 blocking did not restore cytokine production capacities of LCMV-specific CD8⁺ T cells (Fig. 1B).

Functional Properties of gp33-Specific TCR Transgenic (tg) CD8⁺ T Cells.

To investigate whether selective dysfunction of CD8⁺ T cells was also apparent in a monoclonal CD8⁺ T cell population, we adoptively transferred a low number (10⁴) of naïve Ly5.1⁺ TCR tg CD8⁺ T cells, specific for the LCMV gp33–41 epitope, into naïve Ly5.2⁺ C57BL/6 recipient mice followed by l.d. or h.d. LCMV infection. Transfer of 10⁴ naïve Ly5.1⁺ TCR tg CD8⁺ T cells did not influence viral kinetics (SI Fig. 7) or induce selection of CTL escape variants in vivo (SI Figs. 8 and 9 and SI Materials and Methods). Eight days after l.d. infection, TCR tg T cells expressed intermediate to high levels of CD43 and were CD62L⁻ and CD127⁻, virtually all cells were able to degranulate and produce IFN-γ, and a majority produced TNF-α (Fig. 2). In h.d. infection, TCR tg cells expressed high levels of CD43, were CD62L⁻ and CD127⁻, and produced IFN-γ but not TNF-α at day 8 after infection. Importantly, degranulation was not affected in h.d. infection. At day 16 after h.d. infection, TCR tg CD8⁺ T cells still expressed high levels of CD43, were CD62L⁻ and CD127⁻, and degranulation remained functional, whereas cytokine production was further compromised, because IFN-γ and TNF-α synthesis were virtually absent. Similar results were obtained for day 50 after h.d. infection. Impaired cytokine production of TCR tg CD8⁺ T cells from chronically infected mice was also evident on mRNA level (data not shown).

It is conceivable that the impaired cytokine production of CD8⁺ T cells from h.d.-infected mice is because of dysfunctional proximal TCR signaling. We therefore compared MHC/peptide stimulation with phorbol 12-myristate 13-acetate/ionomycin stimulation, which bypasses proximal TCR signaling (Fig. 2D). Both stimulation protocols induced degranulation but not IFN-γ production, indicating that impaired cytokine production is not a consequence of dysfunctional proximal TCR signaling.

Degranulation and Cytotoxicity Are Maintained During Chronic Infection.

We next assessed whether the sustained degranulation potential of CD8⁺ T cells from h.d.-infected mice was associated with cytotoxic activity, because we have previously shown that degranulation is necessary but not sufficient for immediate cytotoxicity (33). We used the adoptive transfer system and isolated CD8⁺ T cells 8, 16, 50, and 90 days after l.d. or h.d. infection. Cells were adjusted to contain equal numbers of Ly5.1⁺ TCR tg CD8⁺ T cells and were tested in ⁵¹Cr release assays (Fig. 3A). TCR tg CD8⁺ T cells from h.d.-infected mice were at least as cytolytic as cells from l.d.-infected mice. We confirmed that cytotoxicity was perforin-mediated, because killing of fas-negative mbl-2 targets was comparable to fas-transfected mbl-2/fas targets (not shown).

We next compared the cytotoxic activity of gp33-specific CD8⁺ T cells directly in vivo in l.d.- or h.d.-infected mice. Sixteen days after infection, carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled target cells were transfused, and elimination of gp33-loaded target cells was assessed after 15 min, 4 h, and 16 h (Fig. 3B). Specific target cell elimination occurred with similar kinetics in l.d.- and h.d.-infected mice, corroborating that the cytotoxic activity of CD8⁺ T cells was maintained in vivo during persistent LCMV infection.

In Vivo Elimination of gp33-Expressing Antigen-Presenting Cells (APCs) Is Perforin-Dependent in Chronic Infection.

To assess whether maintained cytotoxicity of gp33-specific CD8⁺ T cells during h.d. infection was associated with enhanced virus control in vivo, we compared T cell functionality, LCMV titers, and frequencies of gp33-expressing APCs in C57BL/6 and perforin-deficient mice (PKOB) 19 days after h.d. infection. Comparable frequencies of gp33- and np396-specific CD8⁺ T cells were able to degranulate upon specific stimulation but failed to produce IFN-γ (Fig. 4A). LCMV titers in spleens, lymph nodes, and liver were 90% reduced in C57BL/6 mice compared with PKOB mice (data not shown). A more direct evaluation of antigen load, as encountered by gp33-specific CD8⁺ T cells, is the direct ex vivo analysis of gp33-expressing cells. We therefore isolated splenic dendritic cells (DCs), which are preferential in vivo targets for LCMV (34) from h.d.-infected C57BL/6 and PKOB mice and tested their ability to stimulate naïve TCR tg CD8⁺ T cells (Fig. 4B). Only DCs from h.d.-infected PKOB mice were able to induce proliferation of TCR tg CD8⁺ T cells, corroborating that perforin-mediated cytotoxicity is indeed functional in vivo during chronic LCMV infection.

Degranulation Is Ca²⁺-Dependent but Calcineurin-Independent.

Our results indicate a functional dichotomy of CD8⁺ T cells during chronic infection. To address the signaling requirements for degranulation and IFN-γ production in LCMV-specific CD8⁺ T cells directly ex vivo, we tested signaling inhibitors, including the mitogen-activated protein kinase kinase (MEK)1/2 Ca²⁺ chelator EGTA, the calcineurin inhibitor cyclosporine A (CsA), the protein kinase C inhibitor calphostin C (35), or the MEK1/2 inhibitor PD98059 (36) (Fig. 5A). EGTA completely abrogated degranulation and IFN-γ production, whereas CsA selectively affected IFN-γ production. Both calphostin C and PD98059 only marginally affected degranulation and IFN-γ production. Thus, Ca²⁺ influx without involvement of the downstream calcineurin pathway is required for degranulation, implying that degranulation-competent CD8⁺ T cells from h.d.-infected mice are capable of TCR-triggered Ca²⁺ influx. To directly test this, we measured Ca²⁺ flux in response to engagement with gp33-presenting APCs. Significant Ca²⁺ influx was observed in gp33-specific CD8⁺ T cells from mice with l.d. or h.d. infection, indicating that Ca²⁺ influx, triggered by TCR MHC/peptide interactions on APCs, is functional in CD8⁺ T cells from h.d.-infected mice (Fig. 5B).

Impaired Cytokine Production Due to Defective NFAT Nuclear Translocation.

Next, we investigated the mechanisms that lead to impaired cytokine production despite maintained TCR-triggered Ca²⁺ flux. We used the adoptive transfer system and isolated TCR tg Ly5.1⁺ CD8⁺ T cells 18 days after l.d. or h.d. infection or from naïve TCR tg mice. Sorted cells were incubated for 16 h in vitro in the presence or absence of gp33 peptide, and thereafter nuclear and cytoplasmic extracts were prepared for Western blot analysis. Stimulation of naïve TCR tg CD8⁺ T cells or TCR tg CD8⁺ T cells from l.d.-infected mice resulted in pronounced NFAT2 nuclear translocation (Fig. 5C). In sharp contrast, no NFAT2 nuclear translocation was observed in CD8⁺ T cells from h.d.-infected mice, despite comparable levels of cytoplasmic p-NFAT2 in CD8⁺ T cells from h.d.- or l.d.-infected mice (Fig. 5 C and D). We further analyzed NF-κB nuclear translocation upon peptide stimulation (Fig. 5E). TCR tg CD8⁺ T cells from both l.d.- and h.d.-infected mice showed nuclear import of NF-κB upon stimulation. Furthermore, TCR tg CD8⁺ T cells from l.d.- or h.d.-infected mice showed significantly increased levels of p-ERK upon activation (SI Fig. 10 and SI Materials and Methods). Thus, there is a selective defect in NFAT nuclear translocation, whereas Ca²⁺ influx, NF-κB nuclear transport, or ERK phosphorylation are functional in CD8⁺ T cells from chronically infected mice.

Maintained Degranulation and Impaired Cytokine Production of CD8⁺ T Cells During HIV Infection.

Finally, we investigated whether this split functionality of CD8⁺ T cells is a general feature in situations of persistent high antigen load. We recruited 12 HLA-A2⁺ patients with advanced untreated HIV-1 infection (SI Table 1) and compared the ability of their HIV- and CMV-specific CD8⁺ T cells to degranulate and to produce IFN-γ upon specific stimulation (SI Fig. 11 and SI Materials and Methods). A significant proportion of CD8⁺ T cells were able to degranulate and to produce IFN-γ upon staphylococcal enterotoxin B or CMV stimulation, whereas a majority of HIV-specific CD8⁺ T cells tended to degranulate but not produce IFN-γ, which is reminiscent of LCMV-specific CD8⁺ T cells during chronic infection.

Mice.

Tg mice expressing a TCR specific for LCMV peptide gp33–41 and PKOB mice were described previously (51, 53). C57BL/6 mice were purchased from Harlan (Indianapolis, IN) and kept under specific pathogen-free conditions. Animal experiments were performed according to the regulations of the Cantonal Veterinary Office.

Virus, Virus Load Determination, and Peptides.

LCMV Docile was propagated on MDCK cells at low multiplicity of infection. Quantification of infectious virus titers was performed as described (54).

LCMV peptides gp33–41 (gp33, KAVYNFATM) and np396–404 (np396, FQPQNGQFI) were purchased from NeoMPS (Strasbourg, France).

Antibodies and Peptide MHC Class I Tetramers.

APC- or PE-conjugated peptide/MHC class I tetrameric complexes were generated as described (55). Anti-CD107a, -IFN-γ, -TNF-α, -IL-2, -CD8, and -CD45.1 were purchased from BD (Basel, Switzerland).

Lymphocyte Isolation and Stimulation.

Lymphocytes were harvested from spleen or lymph node or isolated from lung and liver of perfused mice and stimulated with 1 μg/ml gp33 or np396 peptide in the presence of anti-mouse CD107a and Monensin A (Sigma, St. Louis, MO). Where indicated, lymphocytes were pretreated for 1 h at 37°C with 10 μg/ml anti-PD-L1 or -CTLA-4 or isotype control antibody (eBioscience, San Diego, CA) before stimulation in continued presence of antibodies. Where indicated, cells were pretreated with 2 mM EGTA (Sigma), 1 μM cyclosporine A (Novartis, Basel, Switzerland), 10 μM calphostin C (Chemie Brunschwig, Basel, Switzerland), or 100 μM PD98059 (Sigma) for 2 h before addition of gp33 and Monensin A in the continued presence of inhibitors (for effectiveness of drugs, refer to SI Fig. 10).

Immunofluorescent Staining and Analysis.

Cells were stained as described (33). Four-color flow cytometric analysis was performed by using a FACSCalibur flow cytometer (BD) with CellQuest software (BD). List-mode data were analyzed by using WinList software (Verity Software House, Topsham, ME).

Ca²⁺ Flux Analysis.

Cells were adjusted to 10⁷ cells/ml and labeled with Fura-red according to the instructions of the supplier (Invitrogen, Carlsbad, CA). Cells were stained for Ly5.1 at 4°C followed by washing and resuspension in medium at 4°C. Peritoneal macrophages (10⁷) were labeled with 1 μM gp33 peptide or mock-treated for 1 h at 37°C, washed twice in PBS, and resuspended in medium at 4°C. Fura-red-labeled cells (5 × 10⁶) were mixed either with 3 × 10⁶ mock- or 3 × 10⁶ gp33-pulsed macrophages on ice, centrifuged at 250× g for 1 min at 4°C, and kept on ice. Cells were warmed to 37°C for 3 min just before and during analysis. Ly5.1⁺ cells were gated within a large forward-scatter/sideward-scatter gate containing T cell–macrophage conjugates. Data acquisition and analysis were done on a LSRII flow cytometer (BD) by using FACSDiva software (BD).

Cell Purification.

CD11c⁺ DCs were purified from spleens by magnetic activated cell sorting (MACS) according to the instructions of the manufacturer (Miltenyi Biotec, Bergisch Gladbach, Germany). TCR tg Ly5.1⁺ CD8⁺ T cells were purified from l.d.- or h.d.-infected mice by MACS sorting of CD8⁺ T cells followed by FACS sorting of Ly5.1⁺ cells on a FACS Aria cell sorter (BD).

Ex Vivo Quantification of gp33⁺ DCs.

Purified DCs were either used directly ex vivo or pulsed with 1 μM gp33 peptide for 1 h at 37°C followed by two washing steps. Serial dilutions of DCs were incubated with 2 × 10⁵ CFSE-labeled CD8⁺ T cells purified from naïve Ly5.1⁺ TCR tg mice. After 3 days, cells were harvested and stained for CD8 and Ly5.1, and CFSE dilution profiles of TCR tg CD8⁺ T cells were analyzed by flow cytometry.

Ex Vivo Cytotoxicity Assay.

Ex vivo ⁵¹Cr release assays were performed on EL4 target cells as described (33). Spontaneous release was <45% in 16-h assays.

In Vivo Cytotoxicity Assay.

Splenocytes from naïve C57BL/6 mice were labeled with either 25 or 2.5 μM CFSE (Invitrogen) and pulsed with 10⁻⁶ M gp33 peptide for 1 h at 37°C or incubated in medium, washed three times with PBS, mixed at equal amounts, and a total of 1 × 10⁷ cells were adoptively transferred.

Western Blots.

Ly5.1⁺ CD8⁺ T cells were purified from adoptively transfused and l.d. or h.d. LCMV-infected mice or from naïve TCR tg mice, and 2–3 × 10⁶ sorted cells were stimulated for 16 h in vitro ± 1 μg/ml gp33 peptide. Cytoplasmic and nuclear extracts were prepared (SI Fig. 12) according to the instructions of the supplier (NE-PER, Perbio, Bonn, Germany). Nuclear and cytoplasmic fractions corresponding to 6 × 10⁵ cells were separated by SDS/PAGE (8.5%) followed by blotting onto a PVDF membrane (Millipore, Billerica, MA). Membranes were blocked with 5% nonfat dry milk in Tris-buffered saline (100 mM NaCl/20 mM Tris·HCl, pH 7.6) with 0.1% Tween20, incubated with anti-NFAT2 (clone 7A6), anti-p-NFAT2 (polyclonal), or anti-NF-κB (clone F-6) antibodies (LabForce, Basel, Switzerland), followed by a secondary HRP-labeled antibody, and developed by chemiluminescence (Chemie Brunschwig).