Synergistic Reversal of Intrahepatic HCV-Specific CD8 T Cell Exhaustion by Combined PD-1/CTLA-4 Blockade

Viral persistence is associated with hierarchical antiviral CD8 T cell exhaustion with increased programmed death-1 (PD-1) expression. In HCV persistence, HCV-specific CD8 T cells from the liver (the site of viral replication) display increased PD-1 expression and a profound functional impairment that is not reversed by PD-1 blockade alone. Here, we report that the inhibitory receptor cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) is preferentially upregulated in PD-1+ T cells from the liver but not blood of chronically HCV-infected patients. PD-1/CTLA-4 co-expression in intrahepatic T cells was associated with a profound HCV-specific effector dysfunction that was synergistically reversed by combined PD-1/CTLA-4 blockade in vitro, but not by blocking PD-1 or CTLA-4 alone. A similar effect was observed in circulating HCV-specific CD8 T cells with increased PD-1/CTLA-4 co-expression during acute hepatitis C. The functional response to combined blockade was directly associated with CTLA-4 expression, lost with CD28-depletion and CD4-independent (including CD4+FoxP3+ Tregs). We conclude that PD-1 and CTLA-4 pathways both contribute to virus-specific T cell exhaustion at the site of viral replication by a redundant mechanism that requires combined PD-1/CTLA-4 blockade to reverse. These findings provide new insights into the mechanisms of virus-specific T cell dysfunction, and suggest that the synergistic effect by combined inhibitory receptor blockade might have a therapeutic application against chronic viral infection in vivo, provided that it does not induce autoimmunity.


Introduction
Virus-specific CD8 T cells become progressively exhausted during chronic viral infection due to increased level or duration of antigenic stimulation without sufficient CD4 help [1]. Among the CD28 family of costimulatory molecules, programmed death-1 (PD-1) is an immune inhibitory receptor that is highly expressed on both exhausted and activated T cells [2]. Interactions between PD-1 and its ligands PD-L1/PD-L2 can inhibit antigen-specific T cell proliferation and effector function [2,3]. Importantly, blockade of PD-1 signaling can restore function to exhausted virus-specific CD8 T cells with reduced viral load in mice with chronic lymphocytic choriomeningitis virus (LCMV) infection in vivo [4], thereby raising the possibility that immune exhaustion can be reversed with potentially therapeutic antiviral effects. A role for PD-1 pathway in viral persistence and antiviral T cell exhaustion has been shown in various chronic viral infections including hepatitis B virus (HBV), human immunodeficiency virus (HIV), simian immunodeficienty virus (SIV) and hepatitis C virus (HCV) [2,5,6,7,8].
In particular, HCV is a highly persistent human pathogen that infects the liver and causes significant morbidity and mortality due to chronic liver disease [9]. Patients with chronic HCV infection harbor dysfunctional antiviral T cells with increased PD-1 expression in circulating blood, and PD-1 blockade can restore their antigen-specific effector function in vitro [10,11,12,13,14]. However, HCV-specific CD8 T cells in the liver (the site of HCV infection) display markedly increased PD-1 expression compared to peripheral blood [10,13,15] and a profound functional impairment that is refractory to PD-1 blockade alone [13]. Similarly, highly activated circulating HCV-specific CD8 T cells in acute evolving hepatitis C show markedly increased PD-1 expression with a deep functional impairment that is unresponsive to PD-1 blockade. These results suggested the existence of additional inhibitory mechanisms that contribute to virus-specific CD8 T cell exhaustion in HCVinfected patients, particularly in PD-1 high cells.
Since intrahepatic PD-1 + CD8 T cells also express increased levels of immune inhibitory receptor cytotoxic T lymphocyteassociated antigen-4 (CTLA-4) [13], we asked if CTLA-4 might contribute to virus-specific T cell dysfunction in HCV-infected patients. We show that CTLA-4 is preferentially co-expressed in PD-1 high CD8 T cells (particularly HCV-specific CD8 T cells) in peripheral blood during acute hepatitis C and in the liver during chronic HCV infection. PD-1/CTLA-4 co-expression was associated with marked antigen-specific effector T cell dysfunction that was dramatically and synergistically reversed by combined PD-1/ CTLA-4 blockade in vitro. The response to combined PD-1/ CTLA-4 blockade was directly associated with CTLA-4 expression, independent of CD4 T cells including FoxP3 + Tregs and dependent on CD28 expression. Collectively, these findings suggest that both CTLA-4 and PD-1 pathways contribute to HCV-specific T cell exhaustion in a distinct manner in HCV infection and that combined inhibition of CTLA-4 and PD-1 pathways may have potential therapeutic application in reversing immune exhaustion.

CTLA-4 expression is increased in PD-1 + HCV-specific CD8 T cells from the liver during chronic HCV infection and in the blood during acute hepatitis C
In chronically HCV-infected patients (C), CTLA-4 expression was greater in CD8 T cells from liver infiltrating lymphocytes (LIL) compared to peripheral blood lymphocytes (PBL) (p,0.0001) ( Figure 1A). This compartmental difference was further amplified in HCV-specific CD8 T cells both in MFI (p = 0.023) and percentages (p = 0.005), but not in CD8 T cells specific for influenza virus (Flu), cytomegalovirus (CMV) or Epstein-Barr virus (EBV) epitopes (p.0.3) ( Figure 1B). These expression patterns mirrored PD-1 expression, which is also upregulated in intrahepatic HCVspecific CD8 T cells [10,12,13,16]. Of note, CTLA-4 expression was low in CD8 T cells from liver explants of two HCV-seronegative patients with nonalcoholic steatohepatitis and alcoholic liver disease ( Figure S1). In peripheral blood, CTLA-4 expression levels were uniformly low in total, HCV-specific and non-HCV-specific CD8 T cells during chronic or resolved (R) HCV infection; this contrasts with PD-1 expression, which is elevated in circulating HCV-specific CD8 T cells from chronic compared to resolved patients [13]. CTLA-4 expression was also upregulated in total and HCV-specific (but not non-HCV-specific) CD8 T cells from the peripheral blood of patients with acute hepatitis C ( Figure 1A/B). Figure 1C shows the representative PD-1 and CTLA-4 expression for peripheral blood and intrahepatic CD8 T cells specific for HCV and non-HCV epitopes detected by HLA-A2/peptide tetramers. Of note, CTLA-4 was preferentially expressed in PD-1-high, but not PD-1intermediate or PD-1-negative CD8 ( Figure 1D) and CD4 T cells (data not shown) with a strong association between PD-1 MFI and CTLA-4 expression in percentage (R = 0.81, p,0.0001) ( Figure 1E) as well as MFI (R = 0.69, p,0.0001, data not shown). Taken together, CTLA-4 expression is induced early with PD-1 in HCV-specific CD8 T cells during acute hepatitis C but becomes compartmentalized to the liver with chronic infection.
Combined PD-1/CTLA-4 blockade can synergistically enhance intrahepatic HCV-specific CD8 and CD4 T cell cytokine response We previously showed that PD-1 blockade failed to restore function to highly PD-1 + HCV-specific CD8 T cells from HCVinfected liver, although it enhanced the functionality of PD-1 intermediate CD8 T cell from blood [13]. As CTLA-4 was preferentially expressed in PD-1 high CD8 T cells ( Figure 1D), we asked if CTLA-4 blockade might reverse their dysfunction, either alone or combined with aPD-L1. To this end, intrahepatic and peripheral lymphocytes from HCV-infected patients were cultured with 15mer peptides spanning the entire HCV NS3 or Flu matrix protein in the presence of blocking aPD-L1 and/or aCTLA-4 or isotype control antibodies. On day 7, the cultures

Author Summary
Hepatitis C virus (HCV) is an important human pathogen with a high rate of persistence associated with chronic liver disease that can progress to cirrhosis and hepatocellular carcinoma. Chronic HCV infection occurs in the setting of impaired antiviral T cells that over-express an inhibitory receptor PD-1 (programmed death-1 receptor). Recent studies showed that in vitro inhibition of the PD-1 pathway via an inhibitory antibody can reverse the functional impairment in HCV-specific CD8 T cells from blood but not the liver (the site of viral infection and disease progression). In this study, we show that a second co-inhibitory receptor, CTLA-4, is upregulated in HCV-specific CD8 T cells from the liver and that combined PD-1/CTLA-4 blockade (but not single blockade of PD-1 or CTLA-4) can synergistically enhance their function. This functional enhancement was CD28-dependent but CD4-independent. This effect also differed between viruses, tissue compartments (liver vs. periphery) and clinical status (acute vs. chronic). We conclude that PD-1, CTLA-4, and CD28 expression profiles define a novel hierarchy in HCVspecific CD8 T cell exhaustion than can be synergistically reversed by combined inhibitory receptor blockade. These findings have potential immunotherapeutic applications, provided that no autoimmunity is induced.
were examined for antigen-specific IFN-c and TNF-a expression by intracellular cytokine staining. As shown in Figure 3A, there was little to no HCV-specific CD4 or CD8 T cell cytokine expression in intrahepatic lymphocytes cultured with aPD-L1 or isotype control antibodies, although low level responses were occasionally seen with aCTLA-4 alone. Remarkably, combined blockade with aPD-L1 and aCTLA-4 resulted in a marked enhancement of intrahepatic HCV NS3-specific CD8 and CD4 T cell cytokine production from 4/6 patients. The comparison of total NS3-specific cytokine responses showed a significant difference between combined PD-1/CTLA-4 blockade and single PD-1 blockade (2.3% vs 0%, p = 0.0037 by Mann Whitney U). However, in blood, HCV-specific CD8 T cell cytokine response was augmented by aPD-L1 without further enhancement by aCTLA-4. Thus, combined PD-1/CTLA-4 blockade resulted in marked HCV-specific cytokine response in LIL but not PBL in HCV-infected patients: LIL 4/6 patients (67%) vs PBL 0/8 patients (0%), p = 0.015 by Fisher's Exact. The effect of inhibitory receptor blockade is shown in representative flow cytometry plots for intrahepatic CD8 ( Figure 3B); notably, the intrahepatic Fluspecific cytokine response was readily detectable without blockade and not enhanced by PD-1/CTLA-4 blockade ( Figure 3B, right panels).
Combined PD-1/CTLA-4 blockade can synergistically enhance intrahepatic HCV-specific CD8 T cell expansion The effect of combined PD-1/CTLA-4 blockade on HCVspecific CD8 T cell function was more directly examined in HLA A2+ patients using HLA-A2/peptide tetramers. As shown in Figure 4A, aPD-L1 alone did not enhance intrahepatic HCV NS3 1073-specific CD8 T cell expansion compared to isotype in patient C57, and a 2-fold enrichment was observed with aCTLA-4. However, combined PD-1/CTLA-4 blockade induced a dramatic increase in the HCV NS3 1073 tetramer + CD8 T cell frequency compared to isotype control or single blockades with aPD-L1 or aCTLA-4. In peripheral blood, aPD-L1 enhanced HCV-specific CD8 T cell expansion as expected while aCTLA-4 provided only an additive effect ( Figure 4B). As for antigenspecific effector function, the frequency of tetramer + CD8 T cells with HCV-specific IFN-c production, CD107a mobilization and perforin expression increased with aCTLA-4 and combined aPD-L1/aCTLA-4 but not with aPD-L1 alone; in peripheral blood, the functional enhancement also occurred with aPD-L1 alone. Overall, the combined PD-1/CTLA-4 blockade enhanced the expansion and effector function of liver-derived HCV tetramer + CD8 T cells compared to isotype control in 2/3 patients ( Figure 4C). By contrast, intrahepatic CMV-specific CD8 T cells displayed little to no PD-1 or CTLA-4 expression ex vivo and were highly functional without further enhancement by PD-1/ CTLA-4 blockade ( Figure 4D). Collectively, these results show that combined PD-1/CTLA-4 blockade can restore proliferative capacity and effector function to deeply exhausted intrahepatic HCV-specific CD8 T cells in a synergistic manner.
Combined PD-1/CTLA-4 blockade can reverse HCVspecific CD8 T cell dysfunction during acute hepatitis C The effect of PD-1/CTLA-4 blockade was examined in 2 HLA-A2+ patients (A29, A35) with acute hepatitis C characterized by markedly elevated serum alanine aminotransferase (sALT) activities and viral titers. As shown ( Figure 5A/B), circulating HCVspecific CD8 T cells displayed increased PD-1 (A29: 95%; A35: 92%) and CTLA-4 (A29: 28%; A35: 14%) expression. As we previously reported [13], HCV-specific CD8 T cells expanded poorly in vitro when stimulated in the presence of aPD-L1 alone. With aCTLA-4 alone, small increases in HCV tetramer + CD8 T cell frequencies were observed in both patients, with proliferation directly measured by CFSE dilution in A35 ( Figure 5B). However, a marked enhancement of HCV-specific CD8 T cell expansion occurred with combined PD-1/CTLA-4 blockade, mirroring the scenario with intrahepatic T cells. Notably, HCVspecific CD8 T cell dysfunction during acute infection did not persist after spontaneous (A29) or treatment-induced (A35) viral clearance (data not shown), suggesting that PD-1 and CTLA-4 inhibitory pathways can downregulate immune function upon active antigenic encounter, but without necessarily defining the ultimate virological outcome. Effect of inhibitory receptor blockade on HCV-specific IFN-c and TNF-a production by CD8 and CD4 T cells from liver and blood. The bar graphs show the frequency of CD8 (upper graphs) and CD4 (lower graphs) T cells with HCV-specific intracellular IFN-c and/or TNF-a expression in liver-derived (left panel, n = 6) and blood-derived (right panel, n = 8) lymphocytes isolated from chronic HCV patients and cultured for 7 days in vitro with 15mer overlapping peptides spanning the entire HCV NS3 protein (pNS3) in the presence of isotype control or blocking antibodies. The 7-day cultures were further stimulated for 6 hours with media alone or with pNS3 before intracellular IFN-c and TNF-a staining. The %IFN-c 2 TNF-a + (blue bar), %IFNc + TNF-a + (red bar) and %IFN-c + TNF-a 2 (yellow bar) T cells are stacked together in each case to show total cytokine + cells. (B) Representative flow cytometry plots showing HCV NS3 and Flu-specific IFN-c and TNF-a production in vitro in liver-derived CD8 T cells from chronic HCV patient C21 following 7 days of culture with NS3 or Flu-derived peptides in the presence of isotype or blocking antibodies. Flow cytometry plots on the far left shows the PD-1 and CTLA-4 expression in liver-derived CD8 T cells directly ex vivo. doi:10.1371/journal.ppat.1000313.g003 The functional response to combined PD-1/CTLA-4 blockade is associated with CTLA-4 expression on HCVspecific effector CD8 T cells but not FoxP3 + Tregs The HCV-specific cytokine response to combined PD-1/ CTLA-4 blockade was tightly correlated with CTLA-4 expression in CD8 T cells directly ex vivo (R = 0.83, p = 0.0026) ( Figure 6A); a similar positive trend was noted with PD-1 expression, although this did not reach a statistical significance (R = 0.36, p = 0.20). Notably, the HCV-specific cytokine response during combined PD-1/CTLA-4 blockade did not correlate with the Foxp3 + Treg frequency directly ex vivo. Moreover, the functional restoration by PD-1/CTLA-4 blockade persisted after CD4-depletion that resulted in complete loss of CD4 + FoxP3 + Tregs ( Figure 6C); these data indicate that the functional response to inhibitory receptor blockade is independent of CD4 + FoxP3 + Tregs. Among tetramer + CD8 T cells, increased PD-1 expression associated with poor augmentation in antigen-specific expansion with the addition of aPD-L1 (data not shown), as previously reported [13]. However, the addition of aCTLA-4 to aPD-L1 resulted in marked expansion of tetramer + CD8 T cells in direct correlation with CTLA-4 expression ex vivo ( Figure 6B), suggesting that aCTLA-4 acts directly on effector CD8 T cells expressing CTLA-4.

Functional restoration by combined PD-1/CTLA-4 blockade is CD28-dependent
Since CD28 is over-expressed in PD-1 + CTLA-4 + CD8 T cells ( Figure 2A) and mediates positive costimulatory signaling for T cell activation [18,19], we asked if the functional response to PD-1/CTLA-4 blockade is mediated by CD28 + T cells. To this end, the effect of PD-1/CTLA-4 blockade on the HCV-specific CD8 T cell IFN-c response was examined in CD4-depleted lymphocyte subsets with and without CD28-depletion in 3 patients (2 chronic HCV patients with liver-derived lymphocytes, 1 acute HCV patients in peripheral lymphocytes). As shown ( Figure 7A), the HCV-specific CD8 T cell IFN-c response was markedly enhanced by combined PD-1/CTLA-4 blockade even in CD4-depleted cells. However, this response was lost with CD28-depletion in all 3 patients, suggesting that the functional response to combined blockade in CD28-dependent.

Discussion
CTLA-4 is an immune inhibitory receptor within the CD28 family of costimulatory molecules [18,20,21]. Induced in activated T cells and constitutively expressed in FoxP3 + Tregs [22,23], CTLA-4 shares its ligands B7-1 and B7-2 with CD28 but binds them with differential kinetics [19,24,25]. CTLA-4 inhibits T cell activation by engaging specific signaling pathways and by outcompeting the positive costimulatory receptor CD28 [20,26]. A critical immune regulatory role for CTLA-4 is evident from the massive and fatal lymphoproliferation that occurs in CTLA-4-deficient mice [27,28]. Antibody-mediated blockade of CTLA-4 signaling can augment antigen-specific CD8 T cell responses in a CD4-independent manner, promoting anti-tumor and autoimmune effects [21,29]. CTLA-4 also contributes to immune regulation and pathology in animal models of bacterial or parasitic infection [30,31]. CTLA-4 may play a more variable role in viral infections. For example, in LCMV-infected mice, the disruption of CTLA-4 signaling fails to modify the course of infection or antiviral T cell responses in vivo [32,33], unlike PD-1 blockade [4]. By contrast, increased CTLA-4 expression on HIV-specific CD4 (but not CD8) T cells is strongly associated with disease progression and reversible immune dysfunction in HIV-infected patients [34]. Little is known about the relevance of CTLA-4 in HCV infection. We show here that CTLA-4 (together with PD-1) contributes to HCV-specific T cell dysfunction in HCV-infected liver that can be dramatically reversed by combined CTLA-4 and PD-1 blockade.
There are several notable findings in our study. For example, CTLA-4 was upregulated in deeply exhausted, HCV-specific PD-1 high CD8 T cells at the site of viral replication (i.e. liver). The increased CTLA-4 and PD-1 expression was functionally relevant, since intrahepatic HCV-specific CD8 T cells regained their function with combined PD-1/CTLA-4 blockade, although not with single-blockade of PD-1 or CTLA-4. Overall, combined PD-1/CTLA-4 blockade strongly enhanced the HCV-specific CD8 T cell response in vitro in 6/9 patients from the liver (4/6 HLA-A22; 2/3 HLA-A2+), compared to the peripheral blood response which was enhanced in none of the 11 chronically infected patients (p = 0.0016). The combined blockade enhanced antigen-specific T cell cytokine production (e.g. IFN-c and TNF-a) and cytolytic potential (e.g. perforin expression and CD107a degranulation) as well as their expansion. Thus, PD-1/CTLA-4 blockade promoted a polyfunctional HCV-specific T cell response, perhaps acting beyond the inhibition of cellular apoptosis which is increased in highly PD-1 + CD8 T cells in the liver or during acute hepatitis  C [35,36]. Interestingly, while combined blockade enhanced both HCV-specific T cell IFN-c and TNF-a production (Figure 3), the increase was particularly evident for TNF-a, suggesting that PD-1/CTLA-4 blockade may promote a cytokine profile that differs from a preferential (albeit weak) IFN-c rather than TNF-a production by dysfunctional HCV-specific CD8 T cells in chronic hepatitis C [11]. The combined PD-1/CTLA-4 blockade also enhanced intrahepatic HCV-specific CD4 T cell function, an important consideration given the relevance of CD4 T cells in immune regulation [37,38].
Although HCV persistence is associated with increased levels of circulating CD4 + FoxP3 + Tregs that constitutively express CTLA-4 [17,39,40,41], the level of functional restoration in CD8 T cells by PD-1/CTLA-4 blockade correlated directly with the frequency of CTLA-4 + CD8 T cells but not FoxP3 + CD4 + Tregs ex vivo. Furthermore, while the functional response to combined blockade differed between the liver and blood, FoxP3 + Treg frequencies did not differ between the two compartments ex vivo or following in vitro culture with PD-1/CTLA-4 blockade (data not shown). Finally, the response to combined blockade persisted after CD4 T cells (including CD4 + FoxP3 + Tregs) were depleted. These results suggest that PD-1/CTLA-4 blockade targets effector T cells directly in a manner independent of CD4 T cells including FoxP3+ Tregs [42].
PD-1 and CTLA-4 inhibit T cell activation through distinct mechanisms that converge on Akt: PD-1 inhibits CD28-mediated activation of phosphatidylinositol 3-kinase (PI3K) and CTLA-4 activates the type II serine/threonine phosphatase PP2A, both leading to the inhibition of Akt phosphorylation [43]. By competing with CD28 for B7-1 and B7-2 [24,44], CTLA-4 can reduce CD28-mediated PI3K activation, further enhancing the negative signaling through PD-1. Since PD-L1 also interacts with B7-1 [45], both aPD-L1 and aCTLA-4 can increase the accessibility of B7-1 to CD28. In our study, HCV-specific CD8 T cells co-expressing PD-1 and CTLA-4 (e.g. from HCV-infected liver) were deeply exhausted and resistant to PD-1 blockade alone, whereas combined PD-1/CTLA-4 blockade had a synergistic effect in restoring their function. However, the two negative regulators may act redundantly to inhibit T cell function in this setting, such that both must be blocked to regain normal functions. The PD-1 + CTLA-4 + phenotype with a functional response to PD-1/CTLA-4 blockade also occurred in circulating HCV-specific CD8 T cells during acute hepatitis C. By contrast, circulating HCV-specific CD8 T cells from chronic HCV patients (e.g. with intermediate PD-1 expression without CTLA-4 co-expression) were functionally augmented by PD-1 blockade alone. These findings suggest that the CTLA-4 and PD-1 pathways are induced early in HCV infection to co-regulate HCV-specific CD8 T cell T cells compared to PD-1 + CTLA-4 2 (Green; 19.5%) and PD-1 2 CTLA-4 2 (Blue; 12.2%) subsets and isotype control (gray shade) in histogram. (C) Effect of CD28-depletion on antigen-specific expansion in the presence of PD-1 and/or CTLA-4 blockade is shown by CFSE-dilution for HCV NS3 1073specific tetramer + CD8 T cells from patient A47. CD4 depleted PBL with and without CD28-depletion were CFSE-labeled and stimulated for 7 days in vitro with HCV NS3 1073 peptide in the presence of isotype or blocking antibodies before flow cytometric analysis. Note that HCV tetramer + CD8 T cells remain detectable with CD28 depletion in the bottom graphs. doi:10.1371/journal.ppat.1000313.g007 function in a redundant manner that differs between tissue compartments over the course of infection.
Notably, CD28 was highly expressed in intrahepatic PD-1 + CTLA-4 + CD8 T cells compared to CTLA-4 2 CD8 T cells. CD28 expression may be induced to counter the inhibitory signals mediated by PD-1 and CTLA-4. Conversely, CTLA-4 may be induced in CD28 + PD-1 + CD8 T cells to downregulate the active inflammation at the site of viral replication. In either case, increased CD28 expression may enable greater functional enhancement upon PD-1/CTLA-4 blockade due to unhampered positive signaling through CD28. Indeed, the functional response to PD-1/CTLA-4 blockade was abolished in HCV-specific CD8 T cells by CD28-depletion in our study. Along these lines, direct CD28 costimulation enhanced HCV-specific CD8 T cell IFN-c response in HCV-monoinfected patients but not in HIV/HCVcoinfected patients with reduced CD28 expression on CD8 T cells in one study [46]. Since loss of CD28 expression is a marker of T cell senescence and functionality [47], detection of CD28 expression in PD-1 + CTLA-4 + CD8 T cells provides an additional marker for reversible functional exhaustion. Collectively, these costimulatory receptors may define a dynamic and complex functional hierarchy for antigen-specific CD8 T cells at various stages and types of viral infections that may respond to distinct therapeutic modulation. If this is correct, combined blockade could have potential therapeutic implications in chronic viral infection, provided it does not trigger autoimmunity.
There are distinct differences between our study and those in human HIV or murine LCMV infections [4,34]. In peripheral blood of HIV-infected patients, increased CTLA-4 expression with a functional response to CTLA-4 blockade was limited to HIVspecific CD4 but not CD8 T cells. This difference may be explained by the reduced CTLA-4 and CD28 expression in HIVspecific CD8 T cells [34], since the functional response to combined PD-1/CTLA-4 blockade depended on both CTLA-4 and CD28 expression in our study. Alternatively, HIV-specific CD8 T cells might exhibit compartmental differences (e.g. between blood and tissue compartments) similar to HCV. In LCMV-infected mice, LCMV-specific CD8 T cells displayed increased CTLA-4 and PD-1 expression, but responded only to PD-1 but not CTLA-4 blockade in vivo. Furthermore, HIV-specific and LCMV-specific T cells were functionally augmented by aPD-L1, without a synergistic response to combined aPD-L1/aCTLA-4 [4,34]. Thus, PD-1 may play a more universal role in antiviral T cell exhaustion whereas the effect of CTLA-4 may differ between viral infections, T cell subsets and even anatomical locations.
In conclusion, both CTLA-4 and PD-1 contributes to HCVspecific T cell exhaustion in a redundant manner in human HCV infection, particularly in HCV-infected liver; this intrahepatic virus-specific T cell dysfunction can be synergistically reversed by combined PD-1/CTLA-4 blockade in vitro in a CD4-independent and CD28-dependent manner. These findings provide new insights to the mechanisms that regulate virus-specific T cell dysfunction and suggest that immune exhaustion at the site of antigen expression may be reversed by combined inhibitory receptor blockade.

Study subjects
All subjects were recruited with informed consent approved by the Institutional Review Boards. All investigations have been conducted according to the principles expressed in the Declaration of Helsinki. Patients were recruited at the Philadelphia Veterans Affairs Medical Center (PVAMC) and the Hospital of the University of Pennsylvania. A total of 47 patients with chronic hepatitis C without HIV coinfection (group C) were examined, including 33 cirrhotic patients undergoing liver transplantation and 14 patients with chronic stable HCV infection. Control groups included 10 healthy HCV-seronegative subjects (group N), 4 HCV-seropositive but RNA-negative patients with spontaneous resolution of HCV infection without prior antiviral therapy (group R) and 6 patients with acute hepatitis C (group A) diagnosed by acute serum alanine amino-transferase (sALT) elevation with documented HCV-seroconversion and/or viremic fluctuations greater than 10-fold without prior liver disease as described previously [48]. The patient characteristics are shown in Table 1.

Immunophenotyping by flow cytometry
Cells were stained with fluorescent antibodies according to the manufacturer's instructions; events were acquired with a FACS-Calibur or FACSCanto (Becton Dickinson, San Jose, CA) and analyzed with FlowJo software (Tree Star Inc., San Carlos, CA). Compensations were established using single color controls. As CTLA-4 is more readily detected in the cytoplasm due to rapid internalization [19,34,52], CTLA-4 expression was assessed by intracellular staining following permeabilization [34]. Cutoffs for CTLA-4 expression was defined by isotype control where 99.9% of the events were negative. Figure S2 further confirms the preferential CTLA-4 expression on PD-1high CD8 T cells with corresponding isotype and unstained controls.

Isolation of peripheral blood lymphocytes (PBL) and liverinfiltrating lymphocytes (LIL)
PBL were isolated by standard Ficoll-Histopaque (Sigma Chemical Co., St Louis, MO) density centrifugation [39,53]. LIL were isolated from 20-50 gm of fresh liver explant tissue that was transported in complete media and processed within 24 hours of explant (usually 1-3 hours) as described previously [13]; briefly, this procedure incorporated careful dicing of liver into 5 mm 3 pieces, incubation of the liver slurry at 37uC with 1 mg/ml collagenase (Type 1a; Roche Molecular) and 1 mg/ml DNase (Sigma Aldrich) for 30 minutes, further mechanical dissociation using the Seward Stomacher 400 Lab Blender (Brinkman Instruments, Westbury, NY), filtration through a 70 mm nylon filter and Ficoll-Histopaque density centrifugation. Control experiments showed that collagenase digestion for 30 minutes did not alter PD-1, CTLA-4, or CD28 expression (data not shown).
Analysis of antigen-specific T cell expansion and effector function in the presence or absence of PD-1 and/or CTLA-4 blockade PBL and LIL (2610 6 cells/ml/well) were stimulated on day 0 with overlapping HCV NS3 or influenza matrix 15mer peptides (2 mM) in complete media in the presence of isotype control antibodies, aPD-L1 [13], aCTLA-4 [34] or both aPD-L1 and aCTLA-4 (10 mg/ml for each mAb). Cell cultures were stimulated with rIL-2 (100 IU/ml) on day 4 and examined by flow cytometry on day 7 with CD107a, intracellular cytokine or perforin staining as previously described [13,54]. For intracellular cytokine staining, expanded PBL and LIL cultures were stimulated for 6 hours with HCV or Flu peptides in the presence of brefeldin A (10 mg/ml) before surface staining, permeabilization and intracellular staining with aIFN-c and aTNF-a. Antigen-specific CD107a mobilization was quantified by adding FITC-labeled aCD107a before peptide stimulation. In selected HLA-A2 + subjects with available cells, PBL and LIL were stimulated with HLA-A2 restricted antigenic peptides (10 mg/ml) with the blocking conditions described above.
In select experiments, antigen-specific IFN-c + T cell response was quantified by IFN-c ELISPOT assay in which cultured lymphocytes were stimulated for 45 additional hours with antigenic peptides or control media (200,000 cells/well in triplicates) as previously described [39,50,53]. HCV-specific IFNc + T cell frequency was calculated by subtracting the mean IFN-c spot forming units (SFU) in control wells from the mean SFU in antigen-stimulated wells and expressed as IFN-c SFU/10 6 cells.

CFSE proliferation assay
Lymphocytes were labeled with 5 mM CFSE (Molecular Probes, Eugene, OR) as described previously [13,17] before 7 days of culture with antigenic peptides (10 mg/ml) in the presence of isotype control or blocking antibodies as described above. Cell cultures were stimulated with rIL-2 (100 IU/ml) on day 4 and examined by flow cytometry on day 7 for antigen-specific T cell expansion.

Statistics
Clinical and immunological parameters were compared using the Mann-Whitney U-test, the paired t-test and the Kruskal-Wallis test. Frequency differences were compared by Fisher's Exact test or the Chi-square test as appropriate. Correlations were tested for significance by the Spearman rank correlation test. P values below 0.05 were considered significant. Figure S1 CTLA-4 and PD-1 expression in intrahepatic CD8 T cells from HCV-seronegative subjects. CTLA-4 and PD-1 expression in gated CD8 T cells from liver explants are shown for 2 HCV seronegative subjects (NC091, NC109) with nonalcoholic steatohepatitis (NASH) and alcoholic cirrhosis and 2 representative chronic HCV patients (C97, C08). Bottom plots show the isotype control staining of the intrahepatic lymphocytes.