Hepatitis c virus-specific t-cell-derived transforming growth factor beta is associated with slow hepatic fibrogenesis§


  • Shaoyong Li,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    2. College of Veterinary Medicine, China Agricultural University, Beijing, China
    Search for more papers by this author
  • Lianne E.M. Vriend,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    2. Vrije University Medical Center, Amsterdam, Netherlands
    Search for more papers by this author
    • These authors participated equally in this study.

  • Imad A. Nasser,

    1. Department of Pathology, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Yury Popov,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Nezam H. Afdhal,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Margaret J. Koziel,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Detlef Schuppan,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    2. Molecular and Translational Medicine, Dept. of Medicine I, Mainz University Medical School, Mainz, Germany
    Search for more papers by this author
  • Mark A. Exley,

    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    Search for more papers by this author
  • Nadia Alatrakchi

    Corresponding author
    1. Department of Medicine, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA
    • Instructor in Medicine, Gastroenterology Division, Beth Israel Deaconess Medical Center, Harvard Institutes of Medicine, Dana/RW663C, 330 Brookline Ave., Boston, MA 02215

    Search for more papers by this author
    • fax: 617-667-8154

  • Potential conflict of interest: Nothing to report.

  • Financial support: U19 AI066313 to N.A., M.E., and D.S., R01 DK066917 and R21 CA143748 to M.E., China Scholarship Council to S.L., and U01 AI068636 to N.A.

  • Present address for M.J. Koziel: Vertex Pharmaceuticals, Cambridge, MA.

  • The protocol was reviewed and approved by the Investigational Review Board of Beth Israel Deaconess Medical Center and all subjects gave informed consent for the collection of specimens.


Hepatitis C virus (HCV)-specific immune effector responses can cause liver damage in chronic infection. Hepatic stellate cells (HSC) are the main effectors of liver fibrosis. TGFβ, produced by HCV-specific CD8+ T cells, is a key regulatory cytokine modulating HCV-specific effector T cells. Here we studied TGFβ as well as other factors produced by HCV-specific intrahepatic lymphocytes (IHL) and peripheral blood cells in hepatic inflammation and fibrogenesis. This was a cross-sectional study of two well-defined groups of HCV-infected subjects with slow (≤0.1 Metavir units/year, n = 13) or rapid (n = 6) liver fibrosis progression. HCV-specific T-cell responses were studied using interferon-gamma (IFNγ)-ELISpot ±monoclonal antibodies (mAbs) blocking regulatory cytokines, along with multiplex, enzyme-linked immunosorbent assay (ELISA) and multiparameter fluorescence-activated cell sorting (FACS). The effects of IHL stimulated with HCV-core peptides on HSC expression of profibrotic and fibrolytic genes were determined. Blocking regulatory cytokines significantly raised detection of HCV-specific effector (IFNγ) responses only in slow fibrosis progressors, both in the periphery (P = 0.003) and liver (P = 0.01). Regulatory cytokine blockade revealed HCV-specific IFNγ responses strongly correlated with HCV-specific TGFβ, measured before blockade (R = 0.84, P = 0.0003), with only a trend to correlation with HCV-specific IL-10. HCV-specific TGFβ was produced by CD8 and CD4 T cells. HCV-specific TGFβ, not interleukin (IL)-10, inversely correlated with liver inflammation (R = −0.63, P = 0.008) and, unexpectedly, fibrosis (R = −0.46, P = 0.05). In addition, supernatants from HCV-stimulated IHL of slow progressors specifically increased fibrolytic gene expression in HSC and treatment with anti-TGFβ mAb abrogated such expression. Conclusion: Although TGFβ is considered a major profibrogenic cytokine, local production of TGFβ by HCV-specific T cells appeared to have a protective role in HCV-infected liver, together with other T-cell-derived factors, ameliorating HCV liver disease progression. (HEPATOLOGY 2012;56:2094–2105)

Up to 4 million persons in the United States have chronic hepatitis C (CHC).1 Despite a decline in overall hepatitis C virus (HCV) infections, the number of patients with endstage liver disease due to CHC will increase for the next 2 decades.2 Even with highly effective novel therapies, currently 30%-50% of infected individuals fail treatment.3 Therefore, a better understanding of mechanisms involved in CHC-related liver disease progression could permit more efficient therapies.

Adaptive effector T cells (frequently assessed by measuring production of prototypic T helper 1 cytokine interferon-gamma [IFNγ]) play an important role in control of HCV infection during the acute phase.4 In CHC, effector HCV-specific T-cell immune responses are weak in peripheral blood, although they can be/are present in liver.5-8 It is likely that in chronic infection, persistence of inefficient effector T-cell responses cause collateral tissue damage and inflammatory reactions, leading to fibrosis and finally cirrhosis. Regulatory/immunosuppressive T cells (Tregs) are apparently involved in HCV pathogenesis, although it remains largely unclear whether they play a detrimental role by suppressing effector T-cell responses against HCV or are protective by preventing excessive immunological liver damage.

Tregs consist of heterogeneous populations that can be natural or induced, antigen-specific or not. Natural Treg, at least in vitro, function via antigen-independent, contact-dependent, and cytokine-independent mechanisms,9 whereas cytokine-mediated suppression (mostly interleukin [IL]-10 and/or transforming growth factor beta [TGFβ]) has been established for peripheral adaptive Treg in vivo.10 This heterogeneity leads to ambiguous marker(s) for identifying Tregs. Current optimal Treg markers are expression of Foxp3 (forkhead box p3), a transcription factor,11 high levels of CD25 (although both of these markers can also be expressed by activated effector T cells), as well as minimal CD127 (IL-7 receptor) expression.12

In HCV infection, increased circulating CD4+CD25+Foxp3+ T cells were associated with viral persistence13, 14 with suppressive activity independent of cytokines and antigen nonspecific.15, 16 Histological costaining of liver infiltrates showed CD4+Foxp3+ cells at high proportion in livers of CHC patients,17 suggesting their involvement in intrahepatic immune regulation, but possibly also amelioration of fibrosis.18 HCV can prime virus-specific CD4+CD25+Foxp3+ Treg with antigen-specific expansion and suppression of HCV-specific CD8+ T cells.19 Tregs also include IL-10-producing CD4+ HCV-specific T cells,20 and IL-10 dampens hepatic inflammation, but also leads to increased viral load.21 Peripheral CD4+CD25+ Tregs were shown to secrete TGFβ in response to HCV, which was inversely correlated with liver inflammation.22 Suppressive IL-10 producing HCV-specific CD8+ liver infiltrating lymphocytes were also described23 and have been associated with protection against apoptosis and fibrosis-related laminin production, as CD8 T cells were located in liver areas with both low hepatocyte apoptosis and fibrosis.24 A limitation of previous studies on Treg is use of phenotypic markers to characterize Treg before functional analysis, as opposed to functionally defining relevant Treg first, so as to not miss subsets. We identified in CHC novel blood HCV-specific CD8+CD25-Foxp3− Treg secreting TGFβ, first functionally then phenotypically.25 TGFβ production by CD4+ T cells was also observed in one patient. Suppression of peripheral HCV-specific IFNγ was predominantly mediated by TGFβ rather than IL-10. The presence of HCV-specific CD4 and CD8 T cells producing TGFβ was recently confirmed in peripheral blood mononuclear cells (PBMCs) in acute HCV infection.37

Of note, TGFβ is a multifunctional cytokine with the unique ability to direct T cell lineage commitment toward either proinflammatory Th17 T cells or antiinflammatory Treg, depending on the presence of additional factors, such as IL-6.26 Significantly, TGFβ is also a key cytokine driving liver fibrogenesis.27 Disruption of the local balance between opposing effects of TGFβ on liver inflammation and fibrogenesis could underline fibrosis progression in CHC. Here we found that TGFβ produced by HCV-specific T cells significantly masks T-cell effector response only in those patients who show attenuated fibrosis progression. In addition, TGFβ inversely correlated not only with liver inflammation, but also with liver fibrosis progression and fibrogenic hepatic stellate cell (HSC) gene expression. It is possible that in chronic HCV infection immunoregulatory and antiinflammatory functions of TGFβ, produced by certain HCV-specific Treg, ameliorate liver inflammation, while limiting the fibrotic process.

Materials and Methods

Subjects and Samples.

Blood and matched liver biopsy samples were assayed from 19 subjects with CHC who were undergoing routine diagnostic evaluation and who had previously another liver biopsy (Table 1). No patients were being treated for HCV infection. All subjects were HCV RNA+, but none were decompensated. Persons with other forms of liver disease, including due to hepatitis B virus or alcohol, other immunosuppressive conditions, or other comorbidity requiring immunosuppressive therapy were excluded, as were subjects with human immunodeficiency virus (HIV) infection. The protocol was reviewed by the Beth Israel Deaconess Medical Center Investigational Review Board and all subjects gave informed consent.

Table 1. Demographics of Studied Subjects
CharacteristicRapid progressors (n = 6)Slow progressors (n = 13)P
  1. Liver histology scored by Metavir system or modified Ishak converted to Metavir scores. HAI: stage+grade. Progression rate for rapid progressors >0.1 Metavir/yr. ALT: alanine-aminotransferase.

Age: median (range), yr57(38-59)57(38-66)0.67
Race5 C; 1 Asian10C;1AA;2Asian 
Male / female sex5/19/4 
HCVRNA, median (range), copies × 103 IU/mL2,025(594-17,800)1,210(31.5-14,600)0.80
HCV genotype (no.)1(6)1(12), 3(1) 
ALT: median (range)81 (34-101)59(34-417)0.83
Liver Inflammation Grade: median (range)1(1-3)1(1-2)0.47
Liver Fibrosis Stage: median (range)2(2-4)1(0-2)0.006
Histological activity index (HAI): median (range)3(3-5)2(1-4)0.009
Fibrosis progression rate: median (range), Metavir/yr0.2(0.1-0.67)0.0 (−0.167-0.074)0.0005
Time between2 biopsies: median (range), yr4.9(3.46-5.21)4.26 (3.85-6.76)0.44

Histology of adequate liver biopsies were staged and graded by both Ishak and Metavir scoring systems and histological activity index (HAI) calculated as total score (grade+stage). Liver fibrosis progression rate was calculated based on histological staging as the difference in Metavir stage between two biopsies divided by years between biopsies. Establishing the cutoff rate of liver fibrosis progression at >0.1 Metavir-units/year for relatively rapid progression allowed studying subjects as two groups: rapid and slow progressors (Table 1).

PBMCs and Intrahepatic Lymphocytes (IHLs).

Extracted PBMC25 and expanded IHL28 were viably cryopreserved for later use. IHLs were expanded using CD3 monoclonal antibody (mAb) (gift of J. Wong, MGH/Boston) to uniformly expand T cells. Autologous Epstein-Barr virus (EBV)-transformed B cell lines (B-LCL) were prepared as described28 for use as antigen-presenting cells for assaying expanded IHL.

Antigens and Media.

Two sets of synthetic peptides were used to stimulate PBMC and IHL. Set 1 consisted of 29 18-mer peptides spanning the entire HCV-Core region derived from HCV type 1a strain H77 (BEI resources). Although Core protein has been reported to have immunosuppressive properties,29 peptides stimulate CD8 and CD4 cells, but cannot exhibit Core protein function. Choice of HCV-Core peptides was because it has relatively low variability and to economize on cells, because Core is covered by only 29 × 18-mer peptides, as opposed to, for example, >100 for NS3. HCV peptides were split into three pools of ∼10 peptides (10 μg/mL each peptide within pool). For analysis, results from the pools were analyzed individually and summed. Set 2 (CEF) (National Institutes of Health [NIH]), a pool of 23 major histocompatibility complex class-I restricted T-cell 11-18-mer peptides from human CMV, EBV, and influenza virus, was used as a control (2 μg/mL each peptide within pool). Positive control was phytohemagglutinin (5 μg/mL; Sigma-Aldrich, St. Louis, MO). Negative control was vehicle (dimethyl sulfoxide solvent [DMSO]).

Blocking Monoclonal Antibodies.

Effector T-cell responses to antigens were studied by IFNγ ELISpot ± blockade of Treg associated cytokines IL-10 and TGFβ, as described.25 Blocking mAbs anti-IL-10 and anti-TGFβ1,2,3 (clone-DII) or immunoglobulin G1 (IgG1) and IgG2b isotype controls (R&D Systems, Minneapolis, MN) were simultaneously added at optimized concentration (10 μg/mL).


IFNγ ELISpot ± Treg cytokines blocking mAbs were performed, adapted as described,25 to detect the presence of suppressive cytokine activity on HCV-specific effector (IFNγ) T-cell response. Capture and detection antibodies were used at optimized concentrations of 5 μg/mL and 0.2 μg/mL for IFNγ (Endogen, Woburn, MA). PBMC (2 × 105cells/well) or IHL (0.5 × 105cells/well) were cultured in triplicate for 20 hours with antigens and in the presence or absence of blocking antibodies or isotype controls. Antigen-specific spot-forming cell (SFC) frequencies were measured with an automated microscope (Zeiss, Munich, Germany) and expressed after background subtraction (SFC observed with buffer media).

Multiparameter Intracytoplasmic Cytokine Staining (ICS).

ICS was performed on PBMC after 6 hours of stimulation as described.25 The following fluorochrome-labeled antibodies were used: FITC-CD8, PE-Cy5-CD3, APC-TGFβ (BD Biosciences Pharmingen), PerCP-Cy5.5-CD4, Alexa-Fluor 700-IFNγ, PE-Cy7-IL-10 (Biolegend), and Pacific blue-viability (eBiosciences). Cells were analyzed using LSR-II multicolor flow cytometer (BD Biosciences Pharmingen) and FlowJo software (v. 9.4.5; TreeStar).

Enzyme-Linked Immunosorbent Assay (ELISA) and Searchlight Multiplex Assays.

Supernatants from cultured cells ± HCV peptide stimulation were harvested. Cytokines released upon antigen stimulation were measured using standardized methods: TGFβ and IL-17 by ELISA (Quantikine) and, 11 additional Th1 and Th2 cytokines (IL-1α/IL-1β/IL-2/IL-4/IL-5/IL-6/IL-8/IL-10/IL-12/IL-13/TNF-α) using multiplex cytokine array (Endogen).

Quantification of Fibrosis-Related Gene Expression in HSCs by Real-Time Quantitative Reverse-Transcription Polymerase Chain Reaction (qRT-PCR).

IHLs were stimulated with media or HCV-Core pool-1, pool-2, or pool-3 peptides in the presence of autologous B-LCL. Fibrogenic/fibrolytic effects of conditioned supernatants (dilution 1:20) from these stimulated IHL or direct coculture were assessed using human LX-2 HSC30 cultured in Dulbecco's modified Eagle's medium (DMEM) plus 2% fetal bovine serum (FBS) in triplicate. HSC transcripts for procollagen-α1 (COL1A1) and matrix metalloproteinase-1 (MMP-1) were quantified, using β2-microglobulin for normalization on a LightCycler 1.5 instrument (Roche, Mannheim, Germany) using TaqMan methodology, as described.31 One μL of complementary DNA (cDNA) (derived from 1 μg of total RNA in 20 μL reaction volume) was used per reaction. Taqman 5′-FAM/3′-TAMRA dual-labeled probes and forward/reverse primers were: COL1A1 (5′-TCGATGGCTGCACGAGTCACACC-3′ and 5′-CAGCCGCTTCACCTACAGC-3′/5′-TCAATCACTGTCTTGCCCCA-3′); MMP-1 (5′-CATCCAAGCCATATATGGACGTTCCCAAA-3′ and 5′-CAGTGGTGATGTTCAGCTAGCTCA-3′/5′-GCCGATGGGCTGGACA-3′). The effect of supernatants from B-LCL alone on HSC showed no significant difference from the effect of media control (IHL+B-LCL+media) (not shown).

Statistical Analysis.

Nonparametric tests were used: Wilcoxon sign rank test to compare each subject results before and after addition of blocking mAbs; Mann-Whitney U test to compare results between two groups of subjects; Spearman rank tests for correlations. Unpaired Student's t test was used to compare gene expression in HSC treated with HCV-stimulated IHL or supernatants versus control-treated HSC. Tests were performed using STATview (Cary, NC, v. 6.0l). P < 0.05 was considered significant.


CEF, pool of peptides derived from CMV, EBV, and influenza-virus; CHC, chronic hepatitis C; CMV, cytomegalovirus; CTL, cytotoxic T lymphocyte; DMSO, dimethyl sulfoxide solvent; EBV, Epstein-Barr virus; Foxp3, forkhead box p3; HAI, histological activity index; HSC, hepatic stellate cells; IFNγ, interferon gamma; IHL, intrahepatic lymphocytes; IL, interleukin; mAbs, monoclonal antibodies; MMP-1, matrix metalloproteinase-1; PBMC, peripheral blood mononuclear cells; RP, rapid liver disease progressor(s); SFC, spot-forming cell; SP, slow liver disease progressor(s); TGFβ, transforming growth factor beta; TNF, tumor necrosis factor; Treg, regulatory T cell(s).


Subject Characteristics.

Subjects' characteristics at the time of study entry are shown in Table 1. Subjects were split into two groups according to the liver fibrosis progression rate: 13 slow progressors (SP) and 6 rapid progressors (RP) with >0.1 Metavir/year. As expected, liver fibrosis stage and progression rate were significantly higher in the RP group (P < 0.006). No difference was observed in alanine aminotransferase (ALT) serum levels, nor in liver inflammation, whereas the HAI, a score combining both liver inflammation and fibrosis, was significantly higher in RP (P = 0.009). The two groups were comparable in terms of age, race, gender, HCV genotype, RNA levels, and number of years separating liver biopsies.

Presence of T-Cell Regulatory Activity in PBMC and Liver, with Greater Suppression of HCV-SpecificEffector T-Cell Responses in Slow Progressors.

Virus-specific effector T-cell responses were studied by IFNγ-ELISpot with or without mAbs against the Treg-associated cytokines TGFβ and IL-10. There was no significant difference in HCV-specific effector IFNγ response between SP and RP in either PBMC and IHL when measured without blocking Abs (P = 0.37). Treg-associated cytokine blockade significantly increased HCV-specific IFNγ response in SP only, in PBMC (P = 0.003) (Fig. 1A) but also in IHL despite their enrichment (P = 0.01) (Fig. 1B). No significant increase was observed in CEF-specific responses (as defined in Materials and Methods) in either compartment. The increase in HCV-specific IFNγ response upon use of Treg cytokine blocking Abs, measured as “block − isotype,” was greater in SP than in RP: in PBMC (P = 0.047) and in IHL (P = 0.08). Of note, undetectable PBMC HCV-specific IFNγ responses of healthy donors were not increased upon Treg-associated cytokine blockade.25 In addition, IHL and PBMC IFNγ responses revealed upon Treg-associated cytokine blockade significantly correlated in their response to HCV peptides (R = 0.6, P = 0.038) (Fig. 2A). Interestingly, in response to HCV peptides, PBMC IFNγ responses revealed upon Treg blockade strongly correlated with IHL IFNγ responses assayed without blockade (R = 0.8, P = 0.006) (Fig. 2B). Again, there was no such correlations in response to control CEF (R < 0.23, P > 0.36). These findings imply similar regulatory T-cell populations suppressing HCV-specific effector T-cell responses in both periphery and liver, and suggest that suppression of effector HCV-specific T-cell responses by way of the Treg-associated cytokine system might be associated with slower HCV-related liver disease progression.

Figure 1.

Greater suppression of HCV-specific effector T-cell responses in slow progressors. PBMC (A) and IHL (B) IFNγ ELISpot assays for HCV-Core and CEF peptides in the presence of isotype controls (Iso.) or anti-Treg-associated cytokine Abs (TGFβ and IL-10) (Block) of six rapid progressors (>0.1 Metavir units/year) and 13 slow progressors. Each line corresponds to one subject. Spot forming cells (SFC)/106 PBMC over background (unstimulated cells) were expressed for HCV as the sum of three pools of HCV-Core peptides. There was no difference between media alone and isotype (data not shown). In both compartments, addition of blocking Abs significantly enhanced median HCV-specific T-cell responses in the slow progressor group only. No enhancement of median T-cell responses to non-HCV-antigens CEF was observed (Wilcoxon sign rank test). The change from Iso to block in HCV-specific T-cell responses is significantly greater in SP versus RP in PBMC (P = 0.047) with a trend to significance in intrahepatic lymphocytes (P = 0.08) (Wilcoxon rank sum test) (not shown).

Figure 2.

Correlations between peripheral and intrahepatic HCV-specific IFNγ. Peripheral IFNγ response to HCV-Core peptides revealed by ELISpot upon blockade of TGFβ and IL-10 significantly correlated (Spearman rank test) with (A) intrahepatic IFNγ response to HCV-Core peptides revealed upon use of blocking Abs, and (B) to directly assayed intrahepatic IFNγ (without Treg cytokines blockade).

Predominant Involvement of TGFβ in the Regulatory/Suppressive Process.

Correlations of T helper (Th)1, Th2, and Treg-associated cytokines secreted by PBMC in response to HCV-Core peptides with peripheral IFNγ response, as revealed by ELISpot upon use of Treg-associated TGFβ and IL-10 blocking Abs, were studied. Total TGFβ secreted by T cells in response to HCV peptides without blocking Treg cytokines significantly correlated with HCV-specific T cell IFNγ, as revealed by Treg cytokine blockade (R = 0.84; P = 0.0003) (Fig. 3A). There was a trend toward correlation between HCV-specific IL-10 secretion without Treg blockade and HCV-specific T cell IFNγ response, as revealed upon Treg blockade (R = 0.43; P = 0.08) (Fig. 3B). These results suggest that the predominant cytokine involved in regulatory/suppressive activity is Treg-associated TGFβ, although IL-10 might also participate.

Figure 3.

Involvement of TGFβ in suppression. Peripheral IFNγ response to HCV-Core peptides revealed upon blockade of TGFβ and IL-10 (ELISpot) significantly correlated (Spearman rank test) with (A) TGFβ, but not (B) IL-10 secreted by PBMC in response to HCV-Core peptides (ELISA). As for ELISpot, ELISA results are expressed as total responses to HCV-Core peptides (summed for three pools).

T Cells Are the Source of HCV-Specific Treg-Associated Cytokine Production.

The type of PBMC involved in HCV-specific production of Treg-associated cytokines was analyzed by multicolor fluorescence-activated cell sorting (FACS) (Fig. 4) in two patients (A and B) with whom Treg cytokine blockade increased PBMC IFNγ by ELISpot (Fig. 1A): 35 to 120 (patient A) and 55 to 105 SFC/106 PBMC (patient B). FACS analysis showed that in response to HCV stimulation, TGFβ was produced by CD8 T cells of patient A (Fig. 4A), and by both CD8 and CD4 T cells as well as IFNγ by CD8, but minimal IL-10 (rare CD8 cells only) from patient B (Fig. 4B). Interestingly, the T-cell population producing TGFβ was distinct from the IFNγ-producing population (Fig. 4C). Control data from other subpopulations of PBMC were also analyzed: no HCV-specific TGFβ responses were observed in CD4/CD8 double-negative or double-positive T cells, nor, significantly, in CD3-negative populations (not shown). T cells were also the likely source of the studied effect of intrahepatic HCV-specific Treg-associated cytokines (Fig. 1B), as only IHL and irradiated EBV-transformed B cells were present. These observations are in accord with our previously published results demonstrating that HCV-Core specific T cells can secrete TGFβ.25

Figure 4.

Frequencies of T cells producing TGFβ, IFNγ, and IL-10 in response to HCV. ICS multicolor FACS analysis was performed on PBMC from two subjects A (4A) and B (4B and 4C). In all, 150,000 events were acquired, of which 50,000 to 100,000 were gated viable cells. The staining rates for all isotype controls were low (<0.02%) (not shown). Gating on CD8+CD3+ lymphocytes and CD4+CD3+ lymphocytes, in response to HCV stimulation, showed production of TGFβ by CD8+ T cells in patient A (4A) and by both CD8+ and CD4+ T cells, IFNγ from by CD8+ T cells and little IL-10 from patient B (4B). The T-cell population producing TGFβ was distinct from the T-cell population producing IFNγ (C). Stimulation with PMA/I (phorbol myristate acetate/ionomycin) was used as technical control.

Relation of HCV-Specific Cytokines Produced by PBMC to Liver Histology.

Cytokines produced by PBMC in response to HCV and control CEF peptide pools were studied in relation to liver histology of matched liver biopsies. No direct association was found between any cytokines produced in response to HCV and liver fibrosis progression rate (P > 0.13) (not shown). However, there was significant inverse correlation between HCV-specific TGFβ and liver inflammation grade (R = −0.63; P = 0.008) (Fig. 5). Interestingly, HCV-specific TGFβ also significantly inversely correlated with liver fibrosis stage (R = −0.46; P = 0.05) (Fig. 5), although correlation with inflammation was more significant. Because grading and staging scores for liver biopsies are not continuous variables, we also analyzed the relation to liver histology using TGFβ median values, considering high grade and stage as >1 (not shown). In accordance with the inverse correlation results above, median HCV-specific TGFβ was significantly higher in subjects with lower grade and stage (P = 0.009 and P = 0.05, respectively). Furthermore, the index combining inflammation and fibrosis scores, HAI, strongly correlated with HCV-specific TGFβ (R = −0.65; P = 0.006) (Fig. 5). Of note, HCV-specific TGFβ did not correlate with peripheral ALT elevations (R = 0.06; P = 0.79) (Fig. 5). Intriguingly, HCV-specific IL-17 secretion was also inversely correlated with liver fibrosis stage (R = −0.55; P = 0.02), but not with liver inflammation grade (not shown). HCV-specific IL-17 also significantly inversely correlated with HAI (R = −0.62; P = 0.01) (not shown). No such relations were observed in response to CEF control (not shown) (P > 0.12). Similarly, no significant correlation was observed between liver histology and other studied cytokines, including HCV-specific IL-10 (P > 0.2). T cells from both slow and rapid progressors secreted high levels of IL-6 (median, range: 155 pg/mL, 4-1,113) and IL-1β (1,569 pg/mL, 42-11,373) in response to HCV peptides (not shown).

Figure 5.

HCV-specific TGFβ correlates with histological liver disease severity. TGFβ secreted by PBMC in response to HCV-Core peptides significantly inversely correlated with liver inflammation grade, fibrosis stage, and histological activity index (HAI = stage + grade), but not with serum ALT. TGFβ results are expressed as total responses to HCV-Core peptides (summed for three pools) (Spearman rank test).

Effect of Supernatants from HCV-Stimulated IHL on Fibrogenic Activation of HSCs.

To further explore potential effects of IHL regulatory activity, we tested the effects of IHL stimulation in response to HCV peptides on human HSC by transfer of conditioned supernatants. This was tested with IHL from five SP and five RP. In ELISpot assays, blocking TGFβ increased the intrahepatic HCV-specific IFNγ response in all tested SP, but not in any RP. HSC significantly increased expression of putatively fibrolytic transcript for MMP-1 upon culture with HCV-stimulated IHL supernatants from SP but not RP (Fig. 6). However, profibrotic COL1A1 gene expression significantly increased in response to IHL supernatant from one rapid progressor (RP-1) and remained unaffected by the remaining IHL supernatants.

Figure 6.

Effect of supernatants from HCV-stimulated IHL on fibrogenic activation of HSC. Supernatants from IHL stimulated with media (control), HCV-Core pool-1, pool-2, and pool-3 peptides (1:20 dilutions of conditioned media) were incubated with cultures of human HSC (LX-2 cell line) for 24 hours. The HSC fibrogenic RNA response was assessed by qRT-PCR for COL1A1 and MMP-1 normalized to B2M, shown as relative fold change to media control. In comparison to media, HCV-stimulated supernatants from slow progressors (SP), but not rapid progressors (RP) induced expression of putatively fibrolytic MMP-1, but not of profibrogenic COL1A1. Stars indicate significant difference (P < 0.05).

Interestingly, treatment of supernatants above (SP-5) or IHL cultures (SP-6 and SP-7) with blocking anti-TGFβ mAb abrogated HSC MMP-1 expression (Fig. 7).

Figure 7.

Effect of treating IHL supernatants or IHL cultures with anti-TGFβ on HSC. Response of LX-2 human hepatic stellate cells to T-cell supernatants (SN) from three slow progressors in whom blocking TGFβ in IHL ELISpot assay increased HCV-specific IFNγ response. Either SN were harvested from IHL culture ±HCV stimulation then treated with anti-TGFβ mAb prior to coculture with LX-2 (SP-5) or IHL ±HCV were cultured with blocking Abs, then SN were harvested for coculture with LX-2 (SP-6 and SP-7). Treatment with anti-TGFβ Ab (in comparison with isotype control) abrogated induction of MMP-1 expression by HSC and also up-regulated COL1A1 (SP-7). Stars indicate significant difference (P < 0.05).

As several reports indicated that TGFβ produced by Treg could provide an effective mechanism of control of fibrosis progression in association with IL-10,32 IHL HCV-specific IL-10 production was measured in the remaining available IHL supernatants (eight patients) by ELISA. Significant amounts of IL-10 were observed in response to HCV-core stimulation (median, range: 365 pg/mL; 0-2,788).


Treg roles in HCV disease progression are not yet clearly established. The reasons could be that Treg are heterogeneous populations and unambiguous Treg markers remain elusive. Consequently, potential Treg subsets are certainly missed, because most Treg studies use phenotypic markers to identify Treg, even if identified cells are then studied functionally. In peripheral blood of subjects with chronic HCV infection, we previously detected TGFβ-mediated suppressive activity against HCV-specific effector function and identified a novel population of nonclassical human Tregs responsive to HCV that produced the Treg-associated cytokine TGFβ.25 In this report, we defined the relation of hepatic and peripheral HCV-specific T-cell-produced TGFβ to HCV-related liver disease.

Blocking Treg-associated cytokines increased effector HCV-specific T-cell responses in slow progressing subjects with chronic HCV infection. This suppressive function was detected in both peripheral and liver compartments, suggesting the presence of similar Treg activity in peripheral blood and liver, at least for certain types of Treg populations. The presence of various hepatic Treg populations have been suggested: CD4+CD25+Foxp3+ (by liver histological costaining assays)17 and CD8+IL-10+ Treg cells (systematic random cloning).23 However, it is not clear whether there are differences or similarities in Treg content and function between periphery and liver. Our finding of a strong correlation between HCV-specific PBMC and IHL IFNγ responses revealed upon Treg cytokine blockade support similarities in cytokine-mediated Treg activity between these compartments. In addition, the revealed PBMC HCV-specific effector responses actually correlated with IHL HCV-specific IFNγ responses assayed without Treg cytokine blockade. It would be ideal if these peripheral responses revealed upon use of Treg cytokine blockade reflect, at least in part, what is occurring in liver, because this would provide a robust surrogate, enabling follow-up longitudinal studies of T-cell immunity to HCV.

Importantly, increases of both PBMC and IHL HCV-specific IFNγ responses resulting from Treg-associated cytokine blockade were significantly greater in patients with slow liver fibrosis progression compared with subjects with rapid liver fibrosis progression, suggesting the presence of effective suppressive Treg-associated cytokine activity in slow, but not rapid progressors, and therefore involvement of this regulatory activity in protection from destructive inflammation and subsequent liver disease progression. Alternatively, it is possible that Treg-associated cytokine blockade in rapid progressors did not increase effector HCV-specific T-cell responses because T cells became anergic upon long-term immunosuppression. However, no differences in effector HCV-specific T-cell responses were observed between the groups without Treg cytokine blockade or in response to mitogen. The immunosuppressive effect appeared to be predominantly mediated by HCV-specific TGFβ rather than IL-10. This is consistent with our previous results with a different cohort of HCV subjects, where blocking TGFβ significantly increased IFNγ response to HCV, whereas IL-10 blockade did not have a significant effect,25 confirming the predominant involvement of TGFβ in this suppressive activity, rather than IL-10.

T-cell secretion of TGFβ in response to HCV has been described for CD4+CD25+22 and CD8+CD25-Foxp3−25 Tregs in subjects with CHC and an antiinflammatory role for TGFβ during chronic HCV infection has been suggested.22 Interestingly, in HCV-HIV coinfection, high levels of plasma TGFβ, but not CD4+Foxp3+ cells, is associated with low levels of liver fibrosis.33 Here, we provide a plausible mechanistic explanation for this observation, because we studied HCV-specific TGFβ T-cell production in relation to liver inflammation and fibrosis. Not only do we confirm an inverse correlation of HCV-specific TGFβ (not IL-10) with histological liver inflammation, but we also found significant inverse correlation with liver fibrosis stage and progression. Together, these findings support the hypothesis that locally HCV-specific T-cell-produced TGFβ may play a role in controlling the chronic inflammatory response, and consequently may even have an antifibrotic role, thereby attenuating hepatic scarring in chronic HCV infection. Intriguingly, our data also suggest involvement of IL-17 as antifibrotic in this setting, because we found a strong inverse correlation of liver fibrosis stage with HCV-specific IL-17. Other cytokines that we found in substantial amounts in HCV-stimulated supernatants, IL-1β and IL-6, might be concurrently expressed in vivo and influence T-cell lineage commitment. Together, these observations underline the complexity of the system, because IL-17 is a proinflammatory cytokine and TGFβ, generally assumed to be antiinflammatory, can become proinflammatory in combination with other cytokines such as IL-1 and IL-6.26, 34 In this context the ability of TGFβ-producing Treg to readily lose Foxp3 and acquire IL-17 expression in Th17-polarizing conditions has been described.34, 35

Because of the lack of definitive surface marker(s) for TGFβ-producing Treg, and therefore, currently an absence of methods allowing their depletion, direct demonstration of the effect of their elimination was not possible. Because establishing pure cocultures of the “Treg” of interest with human HSC is not currently practical, we tested the effect of IHL supernatants in response to HCV peptides on human HSC. Interestingly, factor(s) produced in supernatants in response to HCV from slow progressors, in whom TGFβ blockade increased IHL effector IFNγ response, had an antifibrotic effect on human HSC, and treating these supernatants with anti-TGFβ antibodies abrogated fibrolytic gene expression by HSC. Conversely, no such effect was observed with rapid progressors for whom TGFβ blockade had no such enhancing effect. Such apparently paradoxical observations for TGFβ, most often considered a profibrogenic cytokine, could be explained because TGFβ is a multifunctional cytokine that modulates its function depending on the cell type producing it and other factors present with it or induced by it. In this regard, regulatory IL-10 might also be involved. In a fibrosis mouse model, for example, TGFβ gene therapy lead to the appearance of cells producing IL-10.32 TGFβ “gene therapy” ameliorated fibrosis in wildtype, but not IL-10-deficient mice, and induced Smad4, which then binds to and activates the IL-10 promoter. In our study, we observed a significant production of IL-10 by IHL in response to HCV and peripheral HCV-specific IL-10 data did not exclude its participation in suppressive activity. Although too preliminary to formally conclude, it is possible that when TGFβ is locally produced by HCV-specific Treg it induces substantial amounts of other cytokines, including IL-10, that participate to counterbalance the profibrogenic effect of TGFβ produced by other surrounding hepatic cells. Whether IL-10 and/or other factors contribute to explain the antifibrotic effect of Treg TGFβ will be the object of our future studies.

In conclusion, these results suggest that TGFβ produced locally by Tregs suppresses, rather than enhances, hepatic fibrogenesis. The data also suggest that suppression is in part in concert with other regulatory factor(s) secreted by intrahepatic lymphocytes in response to HCV. Tregs have been associated with HCV persistence in chronic HCV infection.13, 14, 36 However, they may play a more beneficial antiinflammatory role by locally protecting against surrounding tissue damage. Failure to develop appropriate effector and regulatory HCV-specific T-cell responses presumably serves to drive HCV-related liver fibrosis. A better understanding of the opposing effects and roles of different T cell subsets could provide novel tools allowing maintenance of such a beneficial balance, suggesting novel therapeutic approaches to prevent HCV-mediated liver disease progression.

Author Contributions

Data acquisition, analysis: S.L., L.V., I.A.N., Y.P. Providing samples, clinical consulting: N.H.A., D.S., M.J.K. Funding/material support: M.J.K., M.A.E., N.A. Data interpretation, revision: N.A., D.S., M.A.E. Study concept, design, draft: N.A.