Dysfunction and functional restoration of HCV-specific CD8 responses in chronic hepatitis C virus infection


  • Amalia Penna,

    Corresponding author
    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    • Laboratorio di Immunopatologia Virale-U.O., Malattie Infettive ed Epatologia, Azienda Ospedaliero-Universitaria di Parma, Via A. Gramsci, 14, 43100 Parma, Italy
    Search for more papers by this author
    • fax: (39) 0521-703857

  • Massimo Pilli,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Alessandro Zerbini,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Alessandra Orlandini,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Sergio Mezzadri,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Luca Sacchelli,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Gabriele Missale,

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author
  • Carlo Ferrari

    1. Laboratory of Viral Immunopathology, Unit of Infectious Diseases and Hepatology, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


The functional impairment of HCV-specific T cell responses is believed to be an important determinant of HCV persistence, but the functional T cell defects of patients with chronic hepatitis C (CH-C) are only partially defined. CD8 responses to HLA-A2–restricted epitopes of HCV and other unrelated viruses were studied in 23 HLA-A2–positive patients both ex vivo and after in vitro culture. Degranulation capacity, intracellular perforin, and granzyme-A content and cytokine production (IFN-γ, TNF-α) by HCV- and non–HCV-specific CD8 cells were tested both ex vivo and in vitro, whereas cytolytic activity was studied after 10 days' expansion in vitro. Memory maturation and role of exhaustion were assessed ex vivo by HCV-specific CD8 staining for CD127 and PD-1, and in vitro after peripheral blood mononuclear cells (PBMC) culture in the presence of anti–PD-L1 monoclonal antibodies. IFN-γ production and cytolytic activity were expressed less efficiently by HCV-specific than by non–HCV specific CD8 cells derived from the same CH-C patients. The amount of stored granzyme-A within single cells was always lower in HCV-specific CD8 cells, which were less efficient also in the release of lytic granules and in the production of TNF-α. The CD8 dysfunction was associated with high PD-1 expression by most HCV-specific CD8 cells, and PD-1/PD-L1 blockade by anti–PD-L1 antibodies in vitro was able to improve the HCV-specific CD8 function. Conclusion: Our study characterizes CD8 defects that may be important in maintaining HCV persistence; identification of strategies to correct these defects may help to define novel approaches to treat HCV infection. (HEPATOLOGY 2007;45:588–601.)

Hepatitis C virus infection is a major public health problem, affecting more than 200 million people worldwide and causing chronic liver disease, with possible evolution to cirrhosis and HCC.1 Available therapies can cure no more than 50% of infected patients, making essential the development of new therapeutic strategies for nonresponders.1 Restoration of efficient antiviral T cell responses is one of the possible approaches to improve the efficacy of available antiviral drugs. This is suggested by the observation that HCV-specific CD4 and CD8 cells are hyporesponsive in patients with chronic hepatitis C (CH-C).2, 3 However, the functional basis of these defective T cell responses remains largely undefined.

To better characterize the immune defects underlying chronic viral persistence, in this study we focused our analysis of the adaptive immune response in patients with chronic HCV infection on CD8 lymphocytes, given the central role played by these cells in the control of viral infections. Despite the low frequency of peripheral blood HCV-specific CD8 cells at this stage of infection, functional studies were carried out not only on in vitro expanded CD8 cells, but also ex vivo when HCV-specific CD8 frequencies were sufficient for a reliable analysis. This is particularly important to avoid incorrect conclusions based on the study of limited CD8 subsets selectively expanded by in vitro stimulation but insufficiently representative of the overall CD8 population. All together, our results demonstrate a defective HCV-specific cytolytic activity at the single CD8 T cell level, associated with an impaired de-granulation efficiency and a defective production of IFN-γ, granzyme-A (GRZ-A) and tumor necrosis factor alpha (TNF-α) by HCV-specific CD8 cells. The expression of PD-1 by most HCV-specific CD8 cells and the possibility of improving the CD8 function by PD-1/PD-L1 blockade suggest that CD8 exhaustion represents an important cause of this functional impairment.4–6 This study adds a new piece of information to our current knowledge of HCV immunopathogenesis that may be important for the design of strategies to restore an efficient antiviral CD8 response in patients unable to control infection spontaneously.


CH-C, chronic hepatitis C; CMV, cytomegalovirus; EBV, Epstein-Barr virus; GRZ-A, granzyme-A; ICS, intracellular cytokine staining; mAb, monoclonal antibody; MFI, mean fluorescence intensity; PBMC, peripheral blood mononuclear cell; TNF-α, tumor necrosis factor alpha.

Patients and Methods


Twenty-three HLA-A2+ patients with CH-C were studied (14 women, 9 men; mean age, 49 years; mean HCV-RNA level, 573.858 IU/ml; mean ALT level, 100 IU/L; 79% infected by HCV genotype 1) (Table 1). All patients were negative for antibodies to HBV and to human immunodeficiency virus (HIV-1,2). Other possible causes of chronic liver damage, such as alcohol, drugs, congestive heart failure, and autoimmune diseases, were excluded. The study was approved by the Ethics Committee of the Azienda Ospedaliero-Universitaria di Parma, Italy, and all patients gave written informed consent before entering the study.

Table 1. Virologic, Biochemical, and Demographic Characteristics of the Patients Studied
PatientSexAgeHCV GenotypeALT (n.v. < 40 IU/L)HVC-RNA (IU/ml)
  1. Abbreviations: nt, positive HCV-RNA by qualitative PCR; quantitative HCV-RNA not performed.


HCV-RNA and Genotype Determination.

Serum HCV-RNA was quantified using the HCV RNA 3.0 Assay (bDNA) kit (Bayer Health Care LLC, Tarrytown, NY). HCV genotypes were tested by the Innolipa assay (Bayer Health Care LLC, Tarrytown, NY), according to manufacturer's instructions.

Synthetic Peptides and Peptide-HLA Class I Tetramers.

Five synthetic peptides, 9 to 10 amino acids in length (Mimotopes, Victoria, Australia), containing the HLA-A2 binding motif and representing widely recognized (http://hcv.lanl.gov)7 CD8 epitopes located within nonstructural HCV proteins (NS3/1073-1081, NS3/1406-1415, NS4/1812-1821, NS4/1992-2000, NS5/2627-2635), were used to study CD8 responses. Because of the high degree of HCV variability, different sets of peptides corresponding to the prevalent sequences of different HCV genotypes (1 to 4) (www.ncbi.nlm.nih.gov/BLAST/)8 were synthesized and used for T cell analysis. As control peptides, three HLA-A2–restricted epitopes (9 amino acids in length; Mimotopes, Victoria, Australia), known to be targets of the CD8 response against 3 different HCV-unrelated viruses able to establish acute and chronic infections, were selected (CMV 495-503, lower matrix phosphoprotein pp65; EBV 259-267, BMLF-1 early protein; FLU 58-66, influenza-A-virus M1 matrix protein9–11).

HCV-HLA-A2 tetramers (Proimmune LTD, Oxford, UK) contained the above HCV peptides of genotype 1a. The non–HCV-HLA-A2 tetramers contained the cytomegalovirus (CMV), Epstein-Barr virus (EBV), or FLU peptides reported above. Tetramer staining was performed as described in detail.12 All flow cytometry data were analyzed with the CELLQUEST software (BD Biosciences, Immunocytometry Systems).

Isolation of Peripheral Blood Mononuclear Cells and Production of Short-term T Cell Lines.

Peripheral blood mononuclear cell (PBMC) isolation and generation of T cell lines was done as described in detail.12

Cytokine Production (Intracellular Cytokine Staining).

Freshly isolated PBMC, 3 to 5 × 106 cells were incubated with or without synthetic peptides (1 μM) for 1 hour at 37°C in 5% CO2 and then cultured overnight in the presence of brefeldin-A. The day after, stimulated and unstimulated cells were first stained with the PE-tetramers (Proimmune LTD, Oxford, UK) and anti–CD8-APC mononuclear antibody (MAb) (BD Biosciences) and then fixed and permeabilized (Cytofix-Cytoperm by BD Biosciences) to allow detection of intracellular gamma-interferon using FITC-conjugated anti–IFN-γ mAb (Sigma-Aldrich, St. Louis, MO).

For intracellular cytokine staining (ICS) of T cell lines, 0.2 to 0.3 × 106 cells were incubated with or without peptides for 1 hour and then cultured in the presence of brefeldin-A (10 μg/ml, Sigma-Aldrich) for 4 additional hours. After incubation, cells were stained with PE-tetramers and anti-CD8-APC or anti-CD8-PerCP conjugated mAb (BD Biosciences), then fixed and permeabilized and finally incubated with anti–IFN-γ-PE (Sigma-Aldrich) and anti–TNF-α–FITC or anti–IL2-APC (BD Biosciences) conjugated mAbs as described above.

Cytotoxicity Assay.

The cytotoxic activity was assessed by 4 hours' incubation of CD8 cells expanded in vitro by 10 days of peptide stimulation with peptide-pulsed or unpulsed51 Cr-labeled, HLA-A2–positive, EBV-transformed B cells as targets. After calculation of tetramer-positive CD8 cell frequencies in each in vitro expanded T cell line, graded dilutions of tetramer-positive CD8 cells were seeded in each well to test different E:T ratios (0.5, 0.1, 0.05). The amount of51 Cr released by target cells (5000/well) in the supernatant was quantified with a gamma-counter (TopCount NXT, Packard-PerkinElmer, Boston, MA), and the percentage specific lysis was calculated as described.13

Intracellular Perforin and Granzyme Staining.

Freshly isolated PBMC (3-5 × 106) or in vitro expanded (0.2-0.3 × 106) cells were stained first with the PE-tetramers (Proimmune LTD) and then with anti–CD8-APC mAb (BD Biosciences); finally, cells were fixed and permeabilized (Cytofix-Cytoperm by BD Biosciences) for intracellular staining with anti–perforin-FITC (BD Biosciences) or anti–GRZ-A-FITC (BD Biosciences) mAbs.

Degranulation Experiments.

For CD107a staining, the specific mAb (anti-CD107a-PE-Cy, BD Biosciences) was added to effector CD8 cells (from freshly isolated PBMC or T cell lines) at the beginning of the incubation time with or without the relevant peptide. Staining with tetramers, anti–CD8-APC, and anti–IFN-γ-FITC mAbs was performed 2 hours later. CD107a expression was measured as mean fluorescence intensity (MFI) of the specific antibody, and the degranulation capacity of the antigen-specific cytotoxic T lymphocyte (CTL) was expressed as the ratio between the MFI values of stimulated and unstimulated CD8 T cells.

Staining for CD127 and PD-1/PD-L1 Blocking Experiments.

Freshly isolated PBMC (3-5 × 106) were stained first with the PE-tetramers (Proimmune LTD) and the anti-CD127 purified mAb (BD Biosciences). After 20 minutes' incubation and extensive washing, an APC-conjugated goat anti-mouse-IgG1 antibody (BD Biosciences) was added, and after an additional incubation and washing, the cells were stained with anti–CD8-PerCP and anti–PD-1-FITC mAbs or with a mouse IgG1-FITC, as isotype control for PD-1 staining, (BD Biosciences).

For the blocking experiments, cells separated from PBMCs by adherence to the plastic of the culture plates were first incubated with 2 μg/ml anti–PD-L1 mAb (functional grade purified anti-human B7-H1) or a mouse IgG1-k as isotype control (both from eBioscience, San Diego, CA) for 1 hour at 37°C, then HCV and non-HCV peptides were added for 30 minutes. After washing, the nonadherent cells were added and cultured for 10 days for generation of short-term T cell lines, which were tested for virus-specific T cell expansion by tetramer staining and for cytokine production by ICS, as described above.

HCV-RNA Sequencing.

Viral RNA was extracted, amplified, and sequenced as described.14 Briefly, total RNA was extracted from 600 μl serum using a kit RNAfast isolation system (RNA fast, Molecular System, San Diego, CA) and reverse-transcribed using random hexamers and MuLV Reverse Transcriptase (GeneAmp RNA PCR, Applied Biosystems). Nested primer sets were constructed to amplify each region of the HCV genome that contained HLA class I–restricted epitopes. PCR products were then sequenced using the ABI Prism 377 DNA Sequencing System (PerkinElmer Applied Biosystems).

Statistical Analysis.

Mann-Whitney, Wilcoxon matched pairs, and Spearman rank-correlation tests were used for statistical analysis, and P values less than or equal to 0.05 were considered significant.


Impaired IFN-γ and TNF-α Production by HCV-Specific CD8 Cells.

Ex vivo studies were performed on HLA-A2–positive CH-C patients with a measurable frequency of circulating virus-specific CD8 cells detected by flow cytometry with 5 HLA-A2/HCV tetramers (NS3-1073, NS3-1406, NS4-1812, NS4-1992, NS5-2627) and 3 HLA-A2/non-HCV tetramers (CMV, EBV, FLU). Measurable CD8 frequencies were observed for all analyzed peptides except NS4-1812. PBMC were stimulated with the positive HCV and non-HCV peptides identified by tetramer staining, and HCV peptides corresponding to the prevalent sequence of the HCV genotype infecting each individual patient were used.

An impaired capacity to synthesize IFN-γ ex vivo in response to peptide stimulation was observed in HCV-specific compared with CMV-specific CD8 cells derived from the same patients (Fig. 1A). The proportion of IFN-γ–producing cells ranged from 0.5% to 17% of the HCV-tetramer-positive CD8+ cells and from 65% to 98% of the CMV-specific CD8 lymphocytes (Fig. 1A). This defect was observed not only in genotype non-1 (CH6, CH8) but also in genotype 1 (CH3, CH7, CH9, CH10, CH11) infected patients, thereby making very unlikely the possibility that it was an artifact caused by sequence discordance between tetramers (based on a genotype 1a sequence) and peptides used for PBMC stimulation (corresponding to the genotype infecting each patient). This possibility was formally ruled out by the evidence that no IFN-γ production was detectable within the CD8+ tetramer-negative T cell population (Fig. 1B). In addition, ex vivo frequencies of circulating HCV-specific T cells calculated as percentage of CD8+ lymphocytes able to bind the specific tetramer (%CD8+ tetramer-positive) were always higher than frequencies of CD8+ T cells able to produce IFN-γ (%CD8+ IFN-γ+) after peptide-specific stimulation (ratio of IFN-γ+ and tetramer-positive CD8 cells ranging between 0.01 and 0.46, independently of the HCV genotype infecting the individual patient, data not shown). In contrast, frequencies of tetramer-positive and IFN-γ–producing CMV-specific CD8 cells were very similar, with ratios between IFN-γ+ and tetramer-positive CD8 cells very close to 1 (ranging from 0.67 and 1.26, data not shown). IFN-γ production was also tested on T cell lines generated from the same CH-C patients analyzed ex vivo (Fig. 1C). Whereas CMV-specific T cell lines expanded easily in all patients, HCV-specific T cell lines could be induced only in four CH subjects. When the ex vivo results (Fig. 1A) were compared with those generated in vitro on expanded T cell lines (Fig. 1C), a partial restoration of IFN-γ production was observed after culture in the presence of cytokines (Fig. 1C; patients CH3, CH7, CH10), suggesting that the functional impairment can be overcome to some extent. However, the percentage of IFN-γ–producing cells was always lower among HCV-specific (range, 13%-57%) compared with non–HCV-specific CD8 lines (range, 77%-98%) (Fig. 1C).

Figure 1.

Comparison between ex vivo and in vitro IFN-γ synthesis by HCV- and non–HCV-specific CD8+ cells. (A) Ex vivo IFN-γ production by HCV- and CMV-specific CD8 cells was tested by ICS in seven CH-C patients. Patients CH3, CH7, CH9, CH10, and CH11 were infected by HCV genotype 1; patients CH6 and CH8 by HCV genotypes 4 and 3a, respectively. CD8 responses were HCV-NS3-1073-specific in patients CH3, CH6, CH7, and CH8, NS3-1406-specific in patient CH11, NS3-1073 and NS4-1992-specific in patient CH9, and NS3-1073 and NS3-1406-specific in patient CH10. (B) Dot plot analysis of ex vivo IFN-γ production by HCV- and CMV-specific CD8 lymphocytes derived from a representative patient (CH3). (C) After 10 days of in vitro culture, 5 HCV-specific and 4 CMV-specific T cell lines were generated from four patients, and their capacity to produce IFN-γ was tested as described. Plotted results are already subtracted of the background levels. Background values ex vivo of IFN-γ–positive cells among the unstimulated tetramer-positive CD8 population ranged from 0.2% to 0.8% in HCV-specific and from 0.16% to 1% in non–HCV-specific lymphocytes. Background values after 10 days' expansion in vitro ranged from 0.1% to 1.5% of both HCV-specific and non–HCV-specific CD8 cells. The white panels refer to ex vivo data; the gray panel refers to T cell lines.

Simultaneous production of IFN-γ and TNF-α was tested on 10 HCV- and 18 non–HCV-specific T cell lines derived from 7 patients (CH1, CH3, CH15, CH16, CH17, CH18, CH19). All T cell lines responded to peptide stimulation with production of at least one cytokine. Although all TNF-α–positive CD8 cells were also able to produce IFN-γ, an average of 52% of IFN-γ–producing cells among the different HCV-specific T cell lines did not synthesize TNF-α. A greater proportion of CD8 cells within the HCV-specific lines was able to produce IFN-γ alone and, conversely, a significantly higher percentage of TNF-α/IFN-γ–producing cells was detected in non–HCV-specific lines (90% versus 48%, Mann-Whitney P = 0.0004) (data not shown)

Defective Cytolytic Activity by HCV-Specific CD8 Cells.

Given the low frequency of HCV-specific CD8 cells, precluding the possibility to use the51 Cr release assay for cytotoxicity analysis ex vivo, the CD8 cytolytic potential was first analyzed ex vivo by assessing the degranulation capacity (CD107a up-regulation) of virus-specific CD8 cells of different specificity. To test this function, the expression of the CD107a molecule (a lysosomal associated membrane glycoprotein expressed on the cell surface after the release of the cytotoxic granules' content15) on peptide-stimulated and unstimulated CD8 cells was measured after 2 hours' incubation, and the level of up-regulation of the molecule on the surface of HCV-specific and non–HCV-specific CD8 cells was compared in 6 CH-C patients. The increment of CD107a expression was evaluated as the ratio between the MFI of CD8+ tetramer-positive cells (stained with the anti-CD107a mAb) after peptide stimulation and the MFI of the same unstimulated CD8 cells. Non–HCV-specific CD8 cells showed in all 6 patients a greater up-regulation of CD107a expression after peptide stimulation, compared with HCV-specific lymphocytes (Fig. 2A). The MFI of HCV-specific CD8 cells upon peptide stimulation never reached values two times greater than the basal values of unstimulated cells. MFI ratios ranged from 1 to 1.4 in HCV-specific and from 2.1 to 17.4 in non–HCV-specific CD8 cell populations (mean values, 1.2 and 5.8, respectively; Mann-Whitney, P = 0.002) (Fig. 2A).

Figure 2.

Cytolytic function of HCV-specific and non–HCV-specific CD8+ cells evaluated as degranulation capacity and cytotoxic activity. (A) Freshly isolated PBMC from 6 CH-C patients were stained with tetramer and anti-CD107a mAb. After 2 hours' incubation with or without the relevant peptide, the expression of CD107a on HCV and non-HCV/tetramer-positive/CD8+ T cells was tested to evaluate their degranulation capacity, which is expressed as the ratio between the MFI value of stimulated and unstimulated CD8 T cells (MFI ratio). (B) CD107 up-regulation after 2 hours' peptide stimulation was compared in 11 HCV-specific and 16 non–HCV-specific T cell lines derived from 8 CH-C patients. The difference of MFI ratios between HCV-specific and non–HCV-specific CD8 cells was statistically significant (Mann-Whitney test, P = 0.005). (C) CD107a expression on unstimulated (hatched line) and stimulated (solid line) CD8 cells from T cell lines of the indicated specificity from 2 representative patients (CH3, CH17). The numbers in parentheses express the MFI ratio. (D)Cytolytic activity by HCV-specific and non–HCV-specific T cells after in vitro expansion. Graded numbers of tetramer-positive HCV-specific (black symbols) and non–HCV-specific (white symbols) CD8 cells were tested against a constant number of EBV-transformed B cells (HLA-A2 positive) as targets. 2.5 × 103, 5 × 102, or 2.5 × 102 tetramer-positive effector CD8 cells were cultured for 4 hours with 5 × 103peptide pulsed or unpulsed targets. Levels of specific lysis ≥10% were considered significant.

The degranulation capacity was evaluated also after in vitro expansion in 11 HCV- and 16 non–HCV-specific T cell lines (Fig. 2B). MFI ratio levels ranged from 1 to 9.25 in HCV-specific and from 2.67 to 18.40 in non–HCV-specific CD8 cell populations (mean values, 3.46 and 7.31, respectively; Mann-Whitney, P = 0.005). In 7 of 8 patients tested, CD107a up-regulation was minor in HCV-specific compared with non–HCV-specific CD8 cells, as illustrated in Fig. 2C for 2 representative patients (CH3 and CH17).

The cytotoxic efficiency of effector CD8 cells of different specificity was then carefully analyzed at different effector-to-target ratios in nine subjects after in vitro expansion. Graded numbers of tetramer-positive cells specific for HCV, FLU, EBV, or CMV were seeded in each well. In all patients, at equal numbers of seeded tetramer-positive cells, non–HCV-specific T cells expressed better cytotoxicity compared with HCV-specific CD8+ cells (Fig. 2D). In patients CH2, CH15, CH18, CH19, CH21, and CH22, as few as 250 CMV-, FLU-, or EBV-specific CD8 cells (E:T = 0.05) were still able to express detectable levels of cytotoxicity, whereas 10 times more (2500 tetramer-positive cells; E:T = 0.5) HCV-specific CD8 cells were barely able to induce levels of chromium release slightly above the sensitivity threshold of the method (Fig. 2D, patients CH2 and CH18). The only exception was a single line specific for NS4 1992-2000 in patient CH15 that expressed significant cytotoxicity at the lowest E:T ratio tested (0.05).

GRZ-A and Perforin Expression.

The CD8 cytotoxic potential was also analyzed by assessing ex vivo the expression of GRZ-A by virus-specific CD8 cells. GRZ-A was evaluated on tetramer-positive CD8 cells in terms of both frequency of positive cells and intracellular content of the molecule (as MFI) in 12 CH-C patients. Both proportion of tetramer-positive GRZ-A+ cells and amount of stored GRZ-A within individual cells were reduced in HCV-specific compared with non–HCV-specific CD8 cells in most of the patients tested. MFI values ranged from 13 to 47 and from 28 to 61 in HCV- and in non–HCV-specific cells, respectively (Mann-Whitney, P = 0.0086) (Fig. 3A). A mean of 76% and 92% of HCV- and non–HCV-specific CD8 T cells, respectively, were GRZ-A+ (Mann-Whitney, P = 0.01). A representative patient (CH4) is reported in Fig. 3B with percentages of tetramer-positive GRZ-A+ CD8 cells of 23% and 93% among the HCV- and non–HCV-specific CD8 T cell populations, respectively.

Figure 3.

Intracellular GRZ-A content in HCV and non-HCV specific CD8+ cells. (A) The GRZ-A content of freshly isolated HCV and non–HCV-specific CD8 cells was compared in 12 CH-C patients. The MFI values of GRZ-A expression in HCV- and non–HCV-specific CD8 cells from each patient (identified by different symbols) are represented (Mann-Whitney test, P = 0.0086). (B) Percentage of GRZ-A–positive cells among HCV-specific (23%) and CMV-specific (93%) CD8 cells in the representative patient CH4. The corresponding GRZ-A content (MFI values) is illustrated in panel A. (C) Mean fluorescence intensity (MFI) values of all HCV-specific and non–HCV-specific T cell lines derived from the 12 patients studied (Mann-Whitney test, P = 0.0001). (D) Anti–GRZ-A Ab fluorescence intensity profile expressed by tetramer-positive CD8+ cells of different specificity in each individual patient. Fourteen HCV-specific (solid lines) and 20 non–HCV-specific (dotted or hatched lines) T cell lines were tested. In most HCV-specific CD8 cell populations, GRZ-A expression showed a homogeneous pattern with a single peak of fluorescence, with the exception of patients CH15, CH17, CH19, and CH23, who showed two distinct peaks of fluorescence intensity, indicating that a proportion of the cells had a particularly low GRZ-A content. In contrast, GRZ-A expression was always homogeneous in non–HCV-specific lines and higher than in HCV-specific CD8 cells in all patients tested.

Frequency of perforin-positive and granzyme-positive cells and the intracellular content of these molecules were also assessed on in vitro expanded HCV- and non–HCV-specific CD8 cells in 12 CH-C patients. The proportion of tetramer-positive perforin-positive CD8 lymphocytes within HCV-specific T cell lines was similar to that observed in non–HCV-specific lines (94% and 97%, respectively) and similarly behaved to the intracellular content of perforin, which was comparably expressed by HCV- and non–HCV-specific CD8 cells (mean MFI values, 83 and 98, respectively). Although the proportion of tetramer-positive/GRZ-A+ CD8 lymphocytes was similar in T cell lines of different specificity (mean values, 93% and 98% in HCV- and non–HCV-specific lines, respectively; data not shown), the average amount of stored GRZ-A within single cells was always lower in HCV-specific (mean value, 153 MFI) compared with non–HCV-specific CD8 cells (mean value, 277 MFI) derived from the same subjects (Mann-Whitney, P = 0.0001; Fig. 3C,D). MFI values ranged from 77 to 332 and from 130 to 630 in HCV-specific (17 lines tested) and in non–HCV-specific CD8 cells (25 lines tested), respectively.

CD127 and PD-1 Expression on Tetramer-Positive CD8 Cells.

To assess the possible role of T cell exhaustion in the HCV-specific CD8 impairment, PD-1 expression was analyzed on tetramer-positive HCV-specific and non–HCV-specific CD8 cells derived from eight CH-C patients. CD127 staining was simultaneously performed to assess whether memory maturation is inhibited by chronic HCV persistence.16 Most HCV-specific CD8 cells expressed the CD127 marker (an average of 73%; range, 58%-92%) despite chronic exposure to HCV and the low response to HCV antigens. However, the percentage of CD127-positive CD8 cells was significantly lower (Mann Whitney, P = 0.00004) among the HCV-specific than among the CMV-specific or FLU-specific CD8 populations (average, 97%; range, 89%-100%) (Fig. 4A). In contrast, the frequency of antigen-specific CD8 cells expressing PD-1 was significantly higher within the HCV-specific compared with the non–HCV-specific T cell population (average values, 66% and 14%, respectively: Mann Whitney, P = 0.00002) (Fig. 4A). Most HCV-specific/PD-1+ CD8 cells co-expressed CD127 at percentages ranging from 21.5% in patient CH12 to 63.2% in patient CH20, with a mean value of 45.5%. In contrast, only a very limited fraction of non–HCV-specific CD8+CD127+ cells were PD-1 positive, and this lack of PD-1 was associated with the expression of an efficient effector function. Also, the amount of surface CD127 and PD-1 molecules was significantly different in HCV-specific and non–HCV-specific CD8 cells (Mann Whitney, P = 0.0014 for both CD127 and PD-1 comparisons). On average, the surface expression of CD127 and PD-1 was 45 and 32 MFI, respectively, on HCV-specific T cells versus 103 and 21 MFI, respectively, on non–HCV-specific lymphocytes (Fig. 4B). High PD-1 expression was associated with high levels of circulating HCV-RNA, and a positive correlation was observed between both frequency (%) and surface expression (MFI) of PD-1 and levels of viremia (Fig. 4C) (Spearman rank, P < 0.03 and P < 0.02, respectively).

Figure 4.

Ex vivo expression of CD127 and PD-1 by HCV- and non–HCV-specific CD8+ cells. (A) Freshly isolated PBMC from eight CH-C patients were stained with tetramers corresponding to the indicated HCV- and non–HCV-specific epitopes, anti-CD127, and anti-PD-1 mAbs. Numbers indicate the percentages of tetramer-positive cells expressing CD127 or PD-1 or both markers. (B) Surface amount of CD127 and PD-1 on HCV- and non–HCV-specific T cells (expressed as MFI); values were compared by the nonparametric Mann-Whitney test. Patient CH12 had detectable CMV- and FLU-specific T cells. However, dot plot analysis is presented only for CMV-tetramer positive T cells. (C) Correlation between MFI of PD-1 expression or percentage of tetramer-positive PD-1–positive HCV-specific T cells and HCV-RNA levels; values were analyzed by the Spearman rank-correlation test.

Effect of PD-1/PD-L1 Blockade on the CD8 Function.

To assess whether inhibition of PD-1 interaction with its ligand PD-L1 can lead to restoration of the HCV-specific CD8 cell function, blocking experiments with an anti–PD-L1 (B7-H1) mAb were performed in 7 CH-C patients. HCV-specific and non–HCV-specific T cell lines were induced by PBMC stimulation with synthetic peptides for 10 days in the presence of an anti–PD-L1 or a control antibody and the expansion of peptide-specific T cells, as well as the production of IFN-γ and IL-2, were compared by tetramer staining and ICS.

In 6 of 8 cases, PBMC stimulation in the presence of anti–PD-L1 led to a better expansion of HCV-specific CD8 lymphocytes compared with cultures performed in the presence of the control antibody (Fig. 5A,D) (Wilcoxon test, P ≤ 0.015). Also the frequency of both IFN-γ– and IL-2–secreting HCV-specific T cells slightly increased after incubation with anti–PD-L1 compared with the control antibody (Wilcoxon test, P ≤ 0.03 in both cases) (Fig. 5B,C). As expected, because of the lower level of PD-1 expression on influenza-specific and CMV-specific CD8 cells, induction of non–HCV-specific T cell lines in the presence of the anti–PD-L1 mAb led to an increase of FLU-specific or CMV-specific T cells in only 2 patients (CH01 and CH12, respectively; Fig. 5A), and no significant differences were observed in the production of IFN-γ and IL-2 in the presence or absence of the anti–PD-L1 mAb (Fig. 5B,C).

Figure 5.

Effect of PD-1/PD-L1 blockade on the function of HCV-specific and non–HCV-specific CD8 cells. HCV-specific and non–HCV-specific T cell lines were induced in the presence of an anti–PD-L1 mAb or a control mouse immunoglobulin. The expansion of HCV-specific (top of panel A) and non–HCV-specific (bottom of panel A) CD8 T lymphocytes was quantified after 10 days of culture by tetramer staining and was expressed as fold expansion, that is, the ratio between percent tetramer-positive CD8+ T cells on day 10 and percent tetramer-positive CD8+ T cells on day 0 (Wilcoxon matched pairs test, P ≤ 0.015). (B, C) Frequency of IFN-γ– and IL-2–producing T cells among HCV-specific and non–HCV-specific CD8 cells. Statistical analysis was performed using the Wilcoxon matched pairs test. The antigen specificity of the T cell lines tested is the same as described in Fig. 4. (D) HCV-specific CD8 T cell frequencies in the presence of anti–PD-L1 or the control antibody in three representative patients (CH01, CH12, and CH17).

Although a trend toward a better functional recovery was observed in the presence of lower PD-1 and HCV-RNA levels, a statistically significant inverse correlation was detected only between improvement of IFN-γ production as a consequence of the PD-1/PD-L1 blockade and levels of viremia (Spearman rank test, P < 0.02, data not shown).

Sequencing Analysis and CD8 Functional Characterization with Homologous Peptides.

To rule out the possibility that the observed functional impairment of HCV-specific CD8 cells was due to sequence mismatch between reagents and infecting virus (even using genotype-specific peptides), sequencing of the NS3 1073-1081 (in patients CH17, CH21, CH22), NS3 1406-1414 (in patients CH10, CH11), NS4 1992-2000 (in patient CH9) and NS5 2627-2635 (in patient CH1) regions was performed (Fig. 6, upper panel). Patients CH9, CH21, and CH22 showed sequences identical to those of the peptides used for functional analysis. Sequencing of HCV viruses infecting patients CH1, CH10, and CH17 identified single conservative substitutions within peptides NS5 2627 (S to T at position 2633), NS3 1406 (V to I at position 1412), and NS3 1073 (A to V at position 1077), respectively (Fig. 6, upper panel). Patient CH11 was infected by a genotype 1b virus carrying N to G and L to I substitutions at positions 1409 and 1412, respectively, of the NS3 1406-1414 sequence (Fig. 6, upper panel). The use of peptides corresponding to the autologous viral sequences neither improved the proliferative capacity of HCV-specific T cells assessed as a percentage of tetramer-positive CD8+ lymphocytes after 10 days of in vitro culture (Fig. 6A) nor ameliorated IFN-γ production by HCV-specific T cells both ex vivo and after in vitro expansion (Fig. 6B,C) nor increased the cytolytic function (Fig. 6D). In particular, in patient CH17, the autologous NS3 1073-1081 sequence was much less efficient than the genotype-specific peptide in stimulating CD8 cell expansion (3% vs. 14%, respectively) and IFN-γ production (26% versus 63%, respectively) (Fig. 6A,C) and in patient CH1 the autologous NS5 2627-2635 peptide was unable to induce significant levels of cytotoxicity even at the highest E/T ratio tested (Fig. 6D).

Figure 6.

Epitope sequencing and effect of amino acid substitutions on the HCV-specific T cell function. The HCV regions corresponding to the HCV peptides recognized by CD8 cells from patients CH1, CH9, CH10, CH11, CH17, CH21, and CH22 were sequenced. The homologous peptides carried by the infecting viruses are compared in the upper panel with the sequences of the genotype-specific peptides. The homologous viral sequences differing from those of the genotype-specific peptides used for initial analysis (patients CH1, CH10, CH11, CH17) were synthesized and used in parallel with genotype-specific peptides for functional analysis of HCV-specific CD8 cells. (A, C, and D) CD8 expansion, IFN-γ production, and cytolytic activity after 10 days of in vitro peptide stimulation; (B) Ex vivo IFN-γ production. Because of the low expansion of the HCV-specific CD8 T cells at day 10 of culture, in patient CH10 the cytolytic activity was tested at a single E/T ratio (0.1:1). The white panel refers to ex vivo data; gray panels refer to in vitro–generated T cell lines.


HCV-specific CD8 responses are believed to be essential for successful control of HCV infection, as shown by in vivo CD8 depletion studies in the chimpanzee model of HCV infection17 and functional analysis of CD8 responses in acutely infected humans.12, 18, 19–28 Although readily detectable in the acute stage of infection, HCV-specific CD8 responses are difficult to analyze in chronic hepatitis C because of the low peripheral blood CD8 cell frequency at this stage of infection.29, 30 Frequency is greater within the infected liver, but the number of HCV-specific CD8 cells that can be isolated from a liver biopsy specimen is too limited to allow comprehensive ex vivo functional studies.

A functional impairment of HCV-specific CD8 cells has been reported in chronic HCV infection,2, 3, 30, 31 and different mechanisms have been proposed to explain this defective function. These include continuous exposure to viral antigens causing chronic T cell stimulation with subsequent exhaustion of CD8 functions and lack of memory T cell generation2, 16, 32, 33; a direct effect of HCV gene products on T cell or innate immune responses2, 3; mutational escape from CD8 surveillance with the emergence of poorly immunogenic variant epitopes.12, 14, 34–37 Better understanding of the functional CD8 defects and clarification of the mechanisms responsible for this impairment is important to define therapeutic strategies to restore an efficient CD8 function and to facilitate control of infection.

Because of the low peripheral blood HCV-specific CD8 frequency, analysis of the T cell function has generally been carried out after in vitro expansion to obtain a sufficient number of cells for functional studies. In vitro culture, however, can lead to selective expansion of limited T cell subsets insufficiently representative of the overall CD8 population. To circumvent this problem, we initially screened a wide population of CH-C patients, to identify subjects with CD8 frequencies sufficiently elevated for a reliable ex vivo analysis. By this approach, 16 patients were selected for deep functional characterization from an initial screening of 48 HLA-A2+ CH-C patients. Peptides corresponding to the most frequently reported sequences of the HCV genotype infecting each patient were used for T cell stimulation, and sequencing of individual epitopes was performed in 7 patients to analyze the CD8 function after stimulation with homologous peptides carried in vivo by the infecting virus.

A first clear finding emerging from our ex vivo analysis of the CD8 response is that HCV-specific CD8 cells are deeply defective in IFN-γ production, as shown by their comparison with non–HCV-specific CD8 cells derived from the same patients with chronic HCV infection. This functional impairment can be only partially overcome by in vitro culture in the presence of cytokines. Although the proportion of IFN-γ+ tetramer-positive cells increased after 10 days of culture, it never reached the values detected among non–HCV-specific CD8 cells.

A second important finding derived from analyses performed ex vivo and after in vitro expansion is the impairment of the cytolytic function of HCV-specific CD8 cells, which was not restored by in vitro culture with cytokines. The possible explanation that expansion of HCV-specific CD8 cells was insufficient to permit detection of cytolytic activity was ruled out by comparing the cytolytic function of HCV-specific CD8 cells and CD8 cells specific for other unrelated viruses (CMV, EBV, FLU) at equivalent ratios of effector tetramer-positive CD8 cells to target cells. Thus, HCV-specific CD8 cells were less efficient than their non–HCV-specific counterparts at the single tetramer-positive T cell level, thereby demonstrating that the low cytotoxic activity is an expression of a real functional impairment that appears to be selective for HCV-specific CD8 cells. An impaired cytotoxic function in chronic HCV infection has previously been reported by Wedemeyer et al.30; however, the efficiency of the lytic activity per tetramer-positive cell was not significantly lower in CH-C patients than in subjects recovered from HCV infection who were analyzed for comparison. Our data extend these observations because lower levels of HCV-specific cytotoxicity were reproducibly detected by comparing equivalent numbers of HCV-specific and non–HCV-specific tetramer-positive CD8 cells derived from individual patients with chronic HCV infection. This functional difference was detectable not only by the conventional chromium release assay after CD8 expansion in vitro but also by comparing the degranulation capacity of HCV-specific and non–HCV-specific CD8 cells ex vivo. Moreover, the impairment of the HCV-specific cytolytic activity was associated with a lower intracellular content of GRZ-A in HCV-specific compared with non–HCV-specific CD8 cells, which was detectable both ex vivo and in vitro after CD8 cell expansion.

Remarkably, a defective GRZ-A production was detected in all patients studied, and lower expression profiles were observed for all HCV specificities analyzed in each patient. The granule-exocytosis pathway is known to be a highly redundant system with a number of different molecules able to induce cell death by different and complementary mechanisms.38, 39 Moreover, CTL from GRZ-A–deficient mice have normal cytotoxic responses in vitro.40 Therefore, that this functional impairment is sufficient per se to render HCV-specific CD8 cells less efficient at killing virus-infected cells and to explain the decreased cytotoxic activity detected in vitro is unlikely. However, the synergistic effect of a decreased degranulation activity and an impaired TNF-α expression may contribute to reduce the overall antiviral function of HCV-specific CD8 cells. Indeed, a sizable population of HCV-specific CD8 cells failed to produce TNF-α even if they were able to secrete IFN-γ, and the percentage of these tetramer-positive IFN-γ+ TNF-α–negative cells was in each patient significantly more represented among the HCV-specific CD8 population than among the non–HCV-specific counterpart.

An impairment of the cytotoxic function similar to that described in our study has been reported in HIV infection and was correlated with decreased perforin expression.41–43 Moreover, tetramer-positive CD8 cells that lacked effector function, with a weakly cytotoxic activity associated with variable capacity to produce cytokines and perforin, have been described in a murine model of LCMV infection,44 in murine polyomavirus infection,45 in SIV infection,46 and in metastatic melanoma,47 showing that T cells coexisting with persisting viruses can be widely heterogeneous at the functional level.48

The functional features of CD8 cells described in our study cannot be simply attributable to sequence mismatch between reagents and infecting HCV strains because the level of CD8 dysfunction was similar when CD8 responses were analyzed with homologous peptides designed on the sequence of the infecting viruses. Moreover, the behavior of CD8 responses cannot merely reflect the stage of differentiation of circulating CD8 cells because IFN-γ production and cytotoxicity were impaired also when tetramer-positive HCV-specific CD8 cells were predominantly effector memory (CD45-/CCR7-), as shown in two of five patients in whom ex vivo phenotype was performed (data not shown), and thus expected to express these functions efficiently ex vivo.49 Conversely, in vitro maturation by culture in the presence of cytokines and antigenic stimulation did not significantly improve expression of effector functions when CD8 cells were CD127 positive and a central memory phenotype was predominant.

Our results can fit with a model of exhaustion because of chronic stimulation of HCV-specific CD8 cells induced by repeated contact with HCV antigens leading to a hierarchical loss of CD8 functions,33 as suggested by the higher expression of the PD-1 molecule on HCV-specific compared with non–HCV-specific CD8 cells. Although HCV-specific CD8 cells appeared to be profoundly exhausted when their function was analyzed ex vivo, a different level of functional impairment was detectable in vitro after culture with cytokines. Indeed, incubation of HCV-specific T cells with IL-12, IL-7, and IL-2 allowed a partial restoration of IFN-γ production, which was likely less profoundly affected by exhaustion than other CD8 functions, but had no effect on the cytolytic activity, which was probably more deeply impaired because it was more sensitive to negative regulation during chronic infection.33 Moreover, exhaustion was more efficiently overcome by PD-1/PD-L1 blockade, as shown by the increased capacity of CD8 cells to expand and to produce IFN-γ and IL-2 after incubation with anti–PD-L1 antibodies, similarly to what was recently reported in HIV infection.50–52 The functional improvement tended to be greater at lower levels of viremia and PD-1 expression, although a significant inverse correlation was only found between levels of IFN-γ increase and levels of viremia (P < 0.02). In general, a better improvement of CD8 expansion capacity rather than IFN-γ and IL-2 production was observed. This may reflect a hierarchical effect of PD-1/PD-L1 blockade on different CD8 functions more or less profoundly affected by exhaustion, although in patients CH1, CH11, and CH12 the high levels of IFN-γ detectable after culture in IL-2, IL-7, and IL-12 without anti–PD-L1 may have masked the effect of PD-1/PD-L1 blockade on IFN-γ production.

The HCV-specific CD8 dysfunction was detectable in our patients despite CD127 expression by most circulating HCV-specific CD8 cells,53 suggesting that this marker per se does not allow identification of a fully functional memory CD8 population. Indeed, more than 50% of CD127+ CD8 cells were positive for PD-1, which has been reported to down-regulate TCR signal transduction.4–6 Thus, circulating HCV-specific CD8 T lymphocytes appear to be dysfunctional memory cells that can be variably rescued in their function by in vitro PD-1/PD-L1 blockade. This CD8 phenotype of long-term chronic HCV infections differs from the CD8 phenotype observed early after acute infection,27, 28 when HCV-specific CD8 cells in chronically evolving infections are highly PD-1 positive but predominantly CD127 negative. A continuous priming and recruitment of new thymic emigrants has been suggested to contribute to the maintenance of the virus-specific CD8 population in long-lasting chronic viral infections.54 This may lead with time to deep changes of the initial phenotype with progressive accumulation of memory-incompetent cells resulting from abortive CD8 differentiation.

In conclusion, the cytolytic function of HCV-specific, but not EBV-specific, CMV-specific, and FLU-specific CD8 cells, is defective in patients with chronic hepatitis C, as a result of a poor cytolytic efficiency at the single CD8+ T cell level. This functional defect is associated with a decreased expression of granzyme-A, a reduced ability to degranulate in response to peptide stimulation, and a depressed TNF-α production. Moreover, the effector function of HCV-specific CD8 cells is further affected by impaired IFN-γ production. Although the presence of cytotoxicity-incompetent cells is likely the result rather than the cause of chronic uncontrolled virus replication in CH-C patients, once established these molecular defects can certainly contribute to reducing the ability of T cell immunity to control virus replication adequately. Therefore, the possibility of partially restoring CD8 function by blocking PD-1/PD-L1 interaction may provide an additional tool to improve available strategies to cure chronic hepatitis C.