Centre d'Immunologie de Marseille-Luminy, Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale, Université de la Méditerranée Campus de Luminy, Marseille, France
Potential conflict of interest: Nothing to report.
CD8+ T cells represent a sizable component of the liver inflammatory infiltrate in chronic hepatitis C and are thought to contribute to immune-mediated tissue injury. Because chronic stimulation may promote the expression by CD8+ T cells of distinct human leukocyte antigen class I–specific natural killer cell receptors (NKRs) susceptible to both inhibiting effector functions and promoting cell survival, we examined the distribution and characteristics of CD8+ T cells with such receptors in chronic hepatitis C patients. NKR CD8+ T cells were detectable in liver and peripheral blood from hepatitis C virus (HCV)–infected patients but were not major subsets. However, the frequency of NKG2A+ CD8+ in the liver and in a lesser extent in the peripheral blood was positively correlated to histological activity in HCV-infected patients. No such correlation was found with KIR+ T cells in liver in HCV-infected patients and with the both NKR CD8+ T cells in hepatitis B virus (HBV) infected patients. Circulating CD8+ T cells expressing KIRs exhibited phenotypic features of memory T cells with exacerbated expression of the senescence marker CD57 in patients. NKG2A+CD8+ T cells were committed T cells that appeared less differentiated than KIR+CD8+ T cells. In HCV-infected patients, their content in perforin was low and similar to that observed in NKG2A−CD8+ T cells; this scenario was not observed in healthy subjects and HBV-infected patients. Both NKG2A and KIRs could inhibit the response of HCV-specific CD8+ T cells ex vivo. Conclusion: These results support the concept that an accumulation in the liver parenchyma of NKR+CD8+ T cells that have functional alterations could be responsible for liver lesions. They provide novel insights into the complexity of liver-infiltrating CD8+ T cells in chronic hepatitis C and reveal that distinct subsets of antigen-experienced CD8+ T cells are differentially sensitive to the pervasive influence of HCV (HEPATOLOGY 2007.)
Hepatitis C is a liver-targeting infectious disease caused by a noncytopathic RNA virus [hepatitis C virus (HCV)]. If HCV is not rapidly cleared by the immune system, the infection becomes chronic, causes inflammation, mononuclear cell infiltration, hepatocyte destruction, and progression to fibrosis, and can lead to cirrhosis and, in some instances, to hepatocellular carcinoma.1, 2 Although vigorous HCV-specific T cell responses can allow for virus clearance, HCV persists in a majority of cases.3–6 Among liver lymphocytes of chronic hepatitis C (cHC) patients are many activated CD8+ T cells.7 Such a population comprises very few T cells reactive to HCV because HCV-specific CD8+ T cells are difficult to detect within a persistently infected liver.8–10 HCV-specific CD8+ T cells differ from other virus-specific CD8+ T cells by their incomplete differentiation status. For instance, many of them retain the expression of CD27 and CD28 costimulatory molecules that are missing on fully differentiated cytomegalovirus (CMV)-specific CD8+ T cells.11 Our knowledge of CD8+ T cells in cHC has thus progressed. Nevertheless, some features remain poorly explored. In particular, some natural killer cell receptors (NKRs) that bind human leukocyte antigen class I (HLA-I) can be expressed by CD8+ T cells and greatly influence their properties. Although the expression of some NKRs has been examined in some chronic viral infections, including Epstein-Barr virus, CMV, and human immunodeficiency virus (HIV) infections,12–15 less is known about NKR expression by CD8+ T cells in cHC.
Depending on the nature of the ligand-binding domain, HLA-I–reactive NKRs can be subdivided into 2 major groups. The first group corresponds to molecules whose ectodomain belongs to the C-type lectin domain family, such as natural killer group 2A (NKG2A), which associates with CD94 to form inhibitory heterodimers. The second group corresponds to molecules whose ectodomain consists of immunoglobulin (Ig)-like domains. It includes killer cell immunoglobulin-like receptors (KIRs) such as CD158 molecules and leukocyte immunoglobulin-like receptor 1 (LIR1)/immunoglobulin-like transcript 2 (ILT2). Although CD94:NKG2 heterodimers react to the nonclassical HLA-I molecule HLA-E, KIR family members bind to polymorphic residues of classical HLA-I molecules. As for LIR1/ILT2, it reacts to a monomorphic portion of classical HLA-I with poor affinity and with marginal functional results with respect to KIRs or NKG2A.16–20
KIRs are absent on naive CD8+ T lymphocytes, but they can be surface-expressed on CD8+ T cells harboring memory cell–associated phenotypic features. In most cases, they correspond to KIRs with a high affinity for their HLA-I ligands, and that can negatively regulate effector functions (inhibitory KIRs).19 KIR expression by CD8+ T cells reflecting prior T cell receptor (TCR) engagement, is stable, and remains unaltered in a culture.21, 22 With respect to the inhibitory heterodimer CD94:NKG2A, its surface expression on CD8+ T cells depends on TCR engagement and is observed on T cells with a memory/effector phenotype.23 Various observations support the notion that HLA-I–reactive inhibitory NKRs expressed by CD8+ T cells can impair their effector functions.13, 14, 24–26 In addition, KIR and NKG2A expression can have an impact on CD8+ T cell persistence via the modulation of their susceptibility to activation-induced cell death and possibly promote their accumulation.19
In cHC patients, the initiation and progression of liver damage, which is characterized by portal lymphoid infiltration and degenerative lobular lesions, are essentially driven by the host immune response.3 CD8+ T cells are abundant within the diseased liver in both the lobular and periportal areas7, 27 and can be cytotoxic for HCV-infected and noninfected liver cells through various mechanisms.9, 28, 29 Given these characteristics and because HLA-I–specific NKRs can be expressed by committed CD8+ T cells, NKR+CD8+ T cells could contribute to the progression toward chronicity both by impairing the function of HCV-specific CD8+ T cells and by promoting the accumulation of activated CD8+ T cells at large within the liver. We therefore examined the expression of NKG2A (Z199 antibody) and CD158a/h and CD158b/j KIR members (EB6 and GL183 antibodies) by intrahepatic and circulating CD8bright T lymphocytes from patients with a characterized HCV chronic infection in comparison with healthy controls and hepatitis B virus (HBV)–infected patients.
Thirty-four chronically HCV-infected patients seen in our clinical department for a liver biopsy from January to December 2005 were included in the study (Table 1). Each patient had serologic markers of HCV infection and the presence of serum HCV RNA. The criteria for exclusion were an age under 18 years, the presence of serum hepatitis B surface antigen or anti-HIV antibodies, other causes of chronic liver disease, alcohol consumption higher than 30 g/day, and current or past immunomodulatory or antiviral treatment. Serum alanine aminotransferase levels were determined at the time of the liver biopsy and were expressed as a multiple of the upper limit of the normal. Anti-HCV antibodies were tested by a third-generation enzyme-linked immunosorbent assay (ELISA; Ortho Diagnostic Systems, Raritan, NJ). The serum HCV RNA titer was measured with a real-time quantitative polymerase chain reaction (Monitor 2.0 HCV, Roche Diagnostic Systems, Meylan, France) at the time of the liver biopsy. The HCV genotype was determined by direct sequencing (Trugene HCV 5′NC; Visible Genetics, Inc., Toronto, Canada). Patients underwent a percutaneous liver biopsy (16-gauge hepafix needle) as part of their diagnostic evaluation. Blood samples from patients were collected at the time of the biopsy in heparinized vacutainer tubes. Peripheral blood mononuclear cells (PBMCs) were isolated with Ficoll-Hypaque density gradient separation. Two control groups were studied for PBMC experiments: 13 chronically HBV-monoinfected patients seen in the same clinical department and 22 healthy volunteers obtained from our blood bank (Etablissement Français du Sang, La Tronche, France). The study was in agreement with the recommendations of the local ethics committee, and all patients gave written informed consent.
Table 1. Demographic and Clinical Characteristics of the Patients Included in This Study
HCV+ (n = 34)
HBV+ (n = 13)
For technical reasons (the number of cells available per sample), the entire set of experiments described could not be performed for every single patient.
Number of cases (percentage of total). †Median (range).
Liver biopsies were processed for histopathological examination by a single independent pathologist (N.S.) using the Metavir scoring system. Each biopsy was cut into 5-μ-thick cryostat sections, dried, and fixed in formol. Immunostaining was performed with an indirect streptavidin biotin peroxidase method with an anti-CD8 antibody (C8/144B clone, Dako). For the cell suspension analysis, fresh fragments collected in Roswell Park Memorial Institute 1640 10% rhesus AB human serum were washed extensively to minimize blood cell contamination and disrupted mechanically into small pieces in complete media. Cell suspensions were prepared in complete media with a glass mortar. Cells were washed and processed immediately for immunoanalysis.
Bulk Expansion of Intrahepatic Lymphocytes (iHLs).
Infiltrating T lymphocytes were isolated from fresh liver tissue and expanded as we described elsewhere.30 Briefly, cells were expanded polyclonally (unprimed, bulk-expanded) in the presence of irradiated (30 Gy) autologous PBMCs (105 cells/well), recombinant interleukin-2 (100 IU/mL), and phytohemagglutinin A (2 μg/mL) in 96-well plates (105 cells/well) for 2 weeks in Roswell Park Memorial Institute 1640 medium containing 10% AB human serum, 1 mM glutamine, 1% nonessential amino acids, 1% sodium pyruvate, and 50 μg/mL penicillin/streptomycin (Gibco Life Sciences, Rockville, MD) at 37°C in a 5% CO2 atmosphere. Half of the medium was replaced twice a week, and the cells were split at confluence. Expanding cells were maintained for 2–3 weeks without restimulation with the addition of media and interleukin-2. Bulk-expanded lymphocytes were used directly or cryopreserved.
Antibodies and Multimers.
The antibodies were anti–HLA-A2-fluorescein isothiocyanate (FITC; clone BB7.2), anti–CD3-R-phycoerythrin (PE)-cyanine 5 (UCHT1), anti–CD8α-FITC (SK1), anti–CD28-FITC (CD28.2), anti–CD27-FITC (M-T271), anti–CD45RA-FITC (HI100), anti–CD45RO-FITC (UCHL1), anti–CD57-FITC (NK-1), anti–CD85j/LIR1-PE/FITC (GHI/75), and anti–perforin-FITC (δG9) from Becton-Dickinson Biosciences (Le Pont-de-Claix, France). Anti–CD8α-allophycocyanin (APC; B9.11), purified or PE/APC-coupled anti-CD158a,h (EB6), purified or PE/APC-coupled anti-CD158b1,b2,j (GL183), purified or PE-coupled anti-NKG2A/CD159A (Z199), and anti–CD94-PE (HP3B1) were obtained from Beckman Coulter France (Villepinte, France). Anti–NKG2A-APC (131411) and purified anti-CD158b2/KIR2DL3 (180701) were from R&D Systems (Lille, France). Anti–CD62L-FITC (FMC46) was from Serotec (Oxford, United Kingdom). F(ab′)2 of goat anti-mouse IgG-FITC was from Jackson ImmunoResearch (West Grove, PA). Soluble R-PE–labeled HCV peptide:HLA-A*0201 multimers containing the NS31073-1081, NS31406–1415 and NS52594–2602 peptides (NS = nonstructural) were obtained from Proimmune, Ltd. (Oxford, United Kingdom). Staining was performed according to the manufacturer's instructions.
Immunostaining and Flow Cytometry.
Fresh PBMC/intrahepatic samples (0.2–1 × 106 cells) were incubated with a saturating amount of appropriately labeled antibodies in phosphate-buffered saline (30 minutes, 4°C). Cells were washed twice and analyzed immediately. For indirect staining, cells first were incubated with unconjugated primary monoclonal antibodies (mAbs), which were followed by an F(ab′)2 fragment of goat anti-mouse FITC. For intracellular staining, cells were first surface-stained, washed, and subjected to fixation/permeabilization (Cytofix/Cytoperm kit, BD Pharmingen) prior to intracellular staining according to the manufacturer's instructions. A multiparameter flow cytometry analysis was performed on a FACSCalibur instrument with CellQuest software (Becton Dickinson, Mountain View, CA). Nonviable cells were excluded by electronic gating or propidium iodide. Depending on the conditions, 20,000–200,000 gated events were acquired.
For a stimulation assay of T cells from HLA-A*0201+ patients, the cells used were total PBMCs or unprimed, bulk-expanded iHLs plus autologous, T-depleted PBMCs. Cells were cultured for 16–18 hours with a mixture of HCV peptides (3 μg/mL) that bind to HLA-A2 in the presence of anti-KIRs (EB6 and GL183), anti-NKG2A (Z199), or isotype-matched (IgG1), irrelevant antibodies (5 μg/mL). Cultures were set up in duplicates in a final volume of 200 μL of complete media. The peptides were Core 35–44, 90–98, 131–140, 156–165, 177–185, NS31073-1081, 1131–1139, 1169–1177, 1284–1298, 1406–1415, 1443–1451, and 1585–1593. Supernatants were harvested for the detection of tumor necrosis factor α (TNF-α) by ELISA (OptEIA ELISA kit, BD Biosciences).
Comparisons between the group of patients and controls were performed with the nonparametric Mann-Whitney U test and Kruskall-Wallis tests. Within-group comparisons were evaluated with the Wilcoxon signed rank test. Bivariate correlations were made through the computation of Spearman's correlation coefficient. All comparisons were 2-tailed. When multiple comparisons were performed for a given experiment, the P value was corrected by multiplication by the number of comparisons. P values less than 0.05 were considered significant. The statistical analysis was conducted with SPSS 9.0 software (SPSS Inc., Chicago, IL).
CD8bright T Lymphocyte Enrichment in Damaged Livers from cHC Patients.
We previously observed a positive correlation between the severity of liver lesions and the frequency of intrahepatic CD3+ lymphocytes in cHC patients.30 We also found an accumulation of CD8β transcripts in liver biopsies with a high histological activity index with respect to samples with a milder index.31 We therefore examined here the frequency of CD3+CD8bright lymphocytes (hereafter called CD8+ T cells) in liver biopsies from cHC patients presenting with various degrees of histological lesions. The flow cytometry analysis showed that an elevated fraction of CD3+CD8+ iHLs was associated with a more pronounced histological activity (P < 0.04; Fig. 1). This observation was corroborated by an in situ analysis by immunostaining showing a large number of CD8+ T cells surrounding areas of lobular and periportal necrosis (Fig. 2). The latter result is in line with a recent immunochemistry study that showed a higher CD8+ T cell infiltrate in livers from patients with Metavir activity superior to 1 in comparison with those who had mild activity.32 Taken together, these results strongly support the concept that in cHC patients, the progression from mild to marked histological lesions is associated with an increase in the size of the liver CD8+ T cell pool.
Ex Vivo Detection of CD8+ T Cells Expressing HLA-I–Specific NKRs.
Because NKR+CD8+ T cells are antigen-experienced cells whose presence might be associated with chronic stimulation,12, 33 we investigated whether substantial fractions of intrahepatic CD8+ T cells from cHC patients could express CD158a/h and CD158b/j KIRs and NKG2A and CD94 receptors. The data in Fig. 3 indicate that both PBMCs and iHLs from cHC patients did comprise CD8+ T cells with surface expression of NKG2A, CD94, or CD158a/h-CD158b/j KIR molecules. These populations were not major subsets among intrahepatic T cells, with frequencies in the liver ranging on average from 4% for CD158a to 12% for CD94 (Table 2). Interestingly, the percentage of NKR+CD8+ T cells visualized in the liver was significantly correlated to that observed in the blood. The strongest correlations were observed for NKG2A (r = 0.72, P < 0.006) and CD94 (r = 0.82, P < 0.001; Fig. 4). For each NKR+CD8+ T cell subset, there was a trend toward an enrichment in the liver compared to the blood that was significant only for CD158b (11% versus 4%, P < 0.01). Finally, we compared circulating NKR+CD8+ T cell frequencies observed in cHC patients to those of both chronic hepatitis B (cHB) patients and healthy controls, and we observed no significant differences between the 3 groups (Table 2).
Table 2. Frequencies of NKR+CD8+ T Cells in the Liver and Blood
The results are expressed as percentages (medians and ranges).
HCV IHLs (n = 17)
HCV blood (n = 28)
HBV blood (n = 13)
Controls (n = 22)
Increased Liver Injury Correlates with a High Frequency of NKG2A+CD8+ T Lymphocytes.
Because KIR and NKG2A expression by CD8+ T cells reflects prior TCR-mediated activation in vivo,19, 33 we searched for a possible correlation between the frequency of NKR+CD8+ T cells and the disease severity. We found a significant correlation between the histological activity and frequency of intrahepatic NKG2A+CD8+ T cells (r = 0.645, P < 0.02) and to a lesser extent the circulating peripheral NKG2A+CD8+ T cells (r = 0.41, P < 0.05). In contrast, the frequency of liver CD94+CD8+ T cells seemed to negatively correlate with lesion severity, but this trend failed to reach statistical significance (r = −0.543, P = 0.055). No correlation was found with the fibrosis stage. Correlations between the frequency of KIR+CD8+ T cells and biochemical and virological characteristics were not significant. The correlation between the circulating NKG2A+CD8+ T cells and histological activity was not observed in the cHB group (r = 0.04, P = 0.82). Thus, the salient feature of this analysis is that in cHC patients, but not in cHB patients, an increased percentage of the NKG2A+CD8+ T cell subset is associated with enhanced histological activity.
NKG2A or KIR Expression by Intrahepatic HCV-Specific CD8+ T Cells in cHC Patients.
To determine whether NKR expression could be detected among HCV-specific liver CD8+ T cells, we performed costaining with HCV-peptide/HLA-I multimer reagents. Although the frequency of CD8+ T cells specific for a given peptide/HLA-A2 complex may be higher in the liver than in the blood,34, 35 it usually remains very low. For this reason, we used a cocktail of 3 multimer reagents made of HLA-A*0201 molecules complexed to the NS31073–1081, NS31406–1415, and NS52594–2602 HCV epitopes. Figure 5A shows that it is possible to detect very low fractions of HCV-specific CD8+ T cells expressing NKG2A receptors or KIRs among unprimed, bulk-expanded iHLs. Both NKG2A+ and KIR+ HCV-specific CD8+ T cell frequencies were lower than that of HCV-specific cells expressing LIR1/ILT2 molecules. We included the LIR1/ILT2 control because in other infectious diseases these molecules can be expressed by higher fractions of virus-specific CD8+ T cells than KIRs.19 Although a limited number of epitopes were studied here, the data indicate that liver CD8+ T cells from cHC patients can comprise fractions of NKR+ virus-specific cells. Because KIR expression is a feature of antigen-experienced, terminally differentiated CD8+ T cells,12 the data indicate that some HCV-specific CD8+ T cells, albeit at a low frequency, do undergo extensive differentiation during cHC.
NKG2A and KIRs Modulate the Reactivity of HCV-Specific CD8+ T Cells from cHC Patients.
To determine whether NKG2A and KIR expression can have an impact on the reactivity of HCV-specific CD8+ T cells, we stimulated PBMCs from HLA-A2+ patients with various HCV peptides, that bind to HLA-A2 in the presence or absence of anti-NKG2A or anti-KIR antibodies, and measured the TNF-α secretion in the supernatant. We observed that neutralizing KIR engagement and, to a minor extent, NKG2A engagement did enhance the response of virus-specific CD8+ T cells to specific stimulation in vitro (Fig. 5B). Similar results were obtained when unprimed, bulk-expanded iHLs were used in the assay. Thus, NKRs expressed by HCV-specific CD8+ T cells from patients are functional and are capable of interfering with their specific response in vitro.
Phenotypic Features of NKR+CD8+ T Cells.
To study the differentiation status of NKR+CD8+ T cells, we examined their phenotype. Because of the limited number of CD8+ T cells that can be recovered from biopsies, we used PBMCs (Table 3). In all groups, most KIR (CD158a/h plus CD158b/j)+CD8+ T cells expressed CD45RA, and only a few expressed CD45RO. Concerning the replicative senescence marker CD57, we observed that, although the frequency of positive cells was comparable between KIR+CD8+ T cells from patients and controls, there was a marked increase in the CD57 expression level on CD57+KIR+CD8+ T cells from cHC patients, as revealed by the mean fluorescence intensity (MFI; Fig. 6A,B). This effect was not observed in the case of cHB patients. KIR+CD8+ T cells from all cohorts comprised a minority of cells positive for CD27 and CD28 with respect to their KIR− counterparts with, however, a trend toward a further reduced CD28+ cell frequency in HCV+ patients (Fig. 6C), which was also detected for cHB patients. With respect to NKG2A+CD8+ T cells, the high proportions of CD45RO+, CD27+, and CD28+ cells seen in controls were virtually unchanged in cHC patients. As for CD57, the proportion of positive cells among NKG2A+CD8+ T cells was lower than that among KIR+CD8+ T cells, and there was no change in the expression level (not shown). Overall, the analysis of CD8+ T cells expressing either KIRs or NKG2A indicates that both subsets display characteristics of effector/memory CD8+ T cells. NKG2A+CD8+ T cells appeared less differentiated than KIR+CD8+ T cells. Finally, the phenotype of distinct NKR+CD8+ subsets can show subtle distinctions between cHC patients and controls.
Table 3. Phenotypic Features of Peripheral Blood CD8+ T Cell Subsets Expressing HLA-I–Specific NKRs During cHC
The values are the percentages (medians and ranges) of positive cells among CD8bright T cells expressing or lacking KIRs or NKG2A receptors. The differences in the frequencies were not statistically significant.
Scarcity of Perforin+ Cells Among Blood NKG2A+CD8+ T Cells from cHC Patients.
Because perforin is central to granule exocytosis-dependent cytotoxicity, the lytic potential of NKR+CD8+ T cells was estimated by the study of the perforin cellular content. Intracellular stainings in controls showed that blood KIR+CD8+ and NKG2A+CD8+ T cells were both enriched in perforin+ cells with respect to NKR−CD8+ T cells (Fig. 7A). However, although this enrichment was fairly stable among KIR+CD8+ T cells, the proportion of perforin+ cells among NKG2A+CD8+ T lymphocytes was reduced in cHC patients (Fig. 7B). This difference is reminiscent of the reduced frequency of perforin+CD8+ T cells previously noticed in HCV+ patients36 and was not noticed in the case of HBV+ patients (Fig. 7B). Thus, during cHC, a differential modulation of perforin expression might exist among subsets of CD8+ T cells expressing distinct NKRs.
Conventional CD8+ T lymphocytes are largely suspected to be involved in the pathogenesis of cHC.7, 27, 30–32, 37, 38 The first finding of our study supports this concept by unraveling a significant correlation between the frequencies of intrahepatic CD8bright T cells and the histological activity and by showing in situ CD8+ immunostainings in close contact with lobular and periportal hepatocyte necrosis. It has been reported in various pathologies that KIR and NKG2A receptors can be expressed on antigen-experienced CD8+ T cells under conditions of chronic stimulation/inflammation and have an impact on their functions and/or persistence.12, 33 We have therefore examined CD8bright T cells expressing these markers in the liver and blood of cHC patients in terms of the frequencies, functionalities, and association with liver pathogenesis. Although there was no overt accumulation of KIR+CD8+ or NKG2A+CD8+ T cells in the patients, our results indicate that within the liver, the frequency of NKG2A+CD8+ T cells, but not KIR+CD8+ T cells, correlates with lesion severity. The increased frequency of NKG2A+CD8+ T cells that we found in liver samples with marked lesion severity could possibly be related to the fact that the liver can retain committed CD8+ T cells,39–41 a phenomenon that could be reinforced during chronic inflammation. In theory, liver NKG2A+CD8+ T cells could contribute to lesions because some of them were perforin+ and there were blasting cells with a high forward scatter (FSChigh) (not shown). In addition, the inhibitory effect of NKG2A is not necessarily as efficient as that triggered by inhibitory KIRs at interfering with CD8+ T cell lytic activity.19, 20, 25, 42 We have indeed shown that KIRs, rather than NKG2A, are effectively capable of modulating the reactivity of specific anti-HCV CD8+ T cells. It is worth noting that the scarcity of HCV-specific CD8+ T cells that we observed in the liver is consistent with the estimated frequency of intrahepatic cytotoxic T lymphocytes usually reported.8–10 Therefore, the substantial deleterious role for the liver of HCV-reactive NKG2A+CD8+ T cells may not only result from their direct action but also rely on bystander phenomena. On the other hand, liver NKG2A+CD8+ T cells could be neutralized because NKG2A can, in some systems, efficiently control CD8+ T cell functions.26, 43 It is known that such an inhibition greatly depends on HLA-E availability,19, 20 and in the case of hepatitis C, liver HLA-E expression appears enhanced;44 this makes it very likely for NKG2A+CD8+ liver T cells to have their cytotoxic potential constrained in situ. In addition, because repeated TCR engagement is necessary for high NKG2A expression,45 and because the interaction with HLA-E does not modulate NKG2A expression,46 our observation that NKG2A MFI was slightly lower on liver NKG2A+CD8+ T cells from patients than on their circulating counterparts (not shown) suggests that intrahepatic NKG2A+CD8+ T cells are not overtly reactive in situ. Finally, an elevated HLA-E expression is likely to provide NKG2A+CD8+ liver T cells with antiapoptotic signals because CD94 expression is high on NKG2A+CD8+ T cells47, 48 and high CD94 expression is associated with an enhanced resistance to cell death.49 Thus, the increased frequency of NKG2A+CD8+ T cells in biopsies with marked lesions could reflect a selective survival advantage. It is possible that this association indeed reflects a combination of several factors, including the retention of activated/memory CD8+ T cells, HLA-E–mediated resistance to cell death, and, perhaps, some level of activation/expansion. Although technically difficult, further experiments are hence needed to fully understand the properties of liver NKG2A+CD8+ T cells during cHC. With respect to blood cells, we detected the increased frequency of NKG2A+CD8+ T cells reported by Nattermann et al.50 in some (nearly 20%), but not all, cHC patients. This difference could be related to genetic factors, risk factors, and differences in the virus genotype and in the duration of the infection. Therefore, the reduction of cytotoxic functions of NKG2A+CD8+ T at the single cell level could be compensated by increased survival favoring a long-lasting accumulation of this population, which eventually induces hepatocyte lesions.
Phenotypically, KIR+CD8+ T cells from the patients were committed cells with a memory-like phenotype resembling that of terminal effector cytotoxic CD8+ T cells.19 NKG2A+CD8+ T cells were also antigen-experienced cells because they were enriched in CD45RO+ cells, but they were less differentiated than KIR+CD8+ T cells because many were CD27+CD28+ cells. TCR engagement is required for NKG2A surface expression by activated CD8+ T cells during infection.23 Therefore, NKG2A+CD8+ T cells from the patients must have responded to TCR-mediated stimulations in vivo. Given that the frequency of NKG2A+CD8+perforin+ T cells seemed reduced in the patients, the data suggest that the differentiation of NKG2A+CD8+ T cells might be altered in a persistent HCV infection. Although the fractions of NKG2A+CD8+ and KIR+CD8+ T cells that expressed CD57 were quite comparable between the patients and controls, the CD57 expression was enhanced on KIR+CD8+ T cells from HCV patients but not HBV patients. Because CD57 expression by CD8+ T cells reflects a history of intense mitosis and a status of proliferative inability,51 we conclude that the majority KIR+CD8+ T cells from cHC patients are truly terminally differentiated cells that have undergone more extensive in vivo division than their counterparts from controls.
Through yet unknown mechanisms, HCV infection can influence the phenotype of antigen-specific CD8+ T cells. Thus, CMV-specific CD8+ T cells from HCV+ patients are less differentiated than their counterparts from controls. This phenomenon is called pervasive influence.36 In this line, we found phenotypic distinctions between NKR+CD8+ T cells from cHC patients and both healthy individuals and cHB patients. For example, most KIR+CD8+ T cells from cHC patients showed signs of enhanced replicative senescence, and perforin+ cells tended to be less frequent among NKG2A+CD8+ T cells. This suggests therefore that HCV pervasive influence can affect various subsets of committed CD8+ T cells and that such distinct subsets can be differentially susceptible to this influence.
In summary, CD8+ T lymphocytes that express either KIRs or NKG2A are T cell subsets that have undergone TCR-mediated activation in vivo in cHC patients. Their expression on HCV-specific CD8+ T cells is rare but associated, especially for KIRs, with a clear impairment of HCV-specific reactivity. Within the inflamed liver, CD8+ T cells expressing KIRs or NKG2A are variable among patients in terms of frequencies and display phenotypic and functional alterations that are not observed in controls or cHB patients. We hypothesize that these changes are the result of the known pervasive influence of HCV, which differentially affects CD8+ subsets. Finally, NKG2A+CD8+ T cells, whose functional impairment seems to be compensated by a survival advantage favoring accumulation in the liver, appear to be major players in HCV pathogenesis.
We are grateful to the patients who enrolled in this study. We thank Professor Jean-Claude Bensa (Etablissement Français du Sang Rhône-Alpes Grenoble, La Tronche, France) for providing blood samples from healthy donors. We thank C. Viret and M. Faure for helpful discussions.