Activation of hepatic stellate cells after phagocytosis of lymphocytes: A novel pathway of fibrogenesis


  • Potential conflict of interest: Nothing to report.


Increased CD8-T lymphocytes and reduced natural killer (NK) cells contribute to hepatic fibrosis. We have characterized pathways regulating the interactions of human hepatic stellate cells (HSCs) with specific lymphocyte subsets in vivo and in vitro. Fluorescence-activated cell sorting (FACS) was used to characterize human peripheral blood lymphocytes (PBLs) and intrahepatic lymphocytes (IHLs) obtained from healthy controls and from patients with either hepatitis B virus (HBV) or hepatitis C virus (HCV) with advanced fibrosis. Liver sections were analyzed by immunohistochemistry and confocal microscopy. To investigate in vitro interactions, PBLs from healthy controls or patients with HCV cirrhosis were co-cultured with an immortalized human HSC line (LX2 cells) or with primary HSCs. Significant alterations in lymphocyte distribution were identified in IHLs but not PBLs. The hepatic CD4/CD8 ratio and NK cells were significantly reduced in HBV/HCV patients. Expression of alpha-smooth muscle actin and infiltration of CD4, CD8, and NK cells were readily apparent in liver sections from patients with cirrhosis but not in healthy controls. Lymphocytes from each subset were in proximity to HSCs primarily within the periportal regions, and some were directly attached or engulfed. In culture, HSC activation was stimulated by HCV-derived CD8-subsets but attenuated by NK cells. Confocal microscopy identified lymphocyte phagocytosis within HSCs that was completely prevented by blocking intracellular adhesion molecule 1 (ICAM-1) and integrin molecules, or by irradiation of HSCs. LX2 knockdown of either Cdc42 or Rac1 [members of the Rho-guanosine triphosphatase (GTPase) family] prevented both phagocytosis and the activation of HSC by HCV-derived lymphocytes. Conclusion: The CD4/CD8 ratio and NK cells are significantly decreased in livers with advanced human fibrosis. Moreover, disease-associated but not healthy lymphocytes are engulfed by cultured HSCs, which is mediated by the Rac1 and Cdc42 pathways. Ingestion of lymphocytes by HSCs in hepatic fibrosis is a novel and potentially important pathway regulating the impact of lymphocytes on the course of hepatic fibrosis. (HEPATOLOGY 2008.)

Hepatic fibrosis associated with inflammatory cell infiltration is a prominent feature of persistent infection by hepatitis B virus (HBV) and hepatitis C virus (HCV). The hepatic stellate cell (HSC) has assumed a central role in this response after its activation by inflammatory cytokines and mediators.1–5 The cell-mediated immune response after viral hepatitis reflects the activity of CD4+ helper T and CD8+ cytotoxic T lymphocytes. CD4+ T cells are activated through the interaction of major histocompatibility complex (MHC)-II with antigen-presenting cells, Kupffer cells, dendritic cells, and macrophages. CD8+ T cells are MHC-I restricted and are a major mechanism of cytotoxic clearance of infected cells.6 CD4+ T cells secrete cytokines such as tumor necrosis factor alpha, interferon gamma (IFN-γ), and interleukin 2. These cytokines are responsible for the activation of macrophages, Kupffer cells, and natural killer (NK) cells, leading to phagocytosis and nonspecific lysis of infected cells. Viral clearance during HCV infection is made possible through vigorous HCV-specific CD4+ and CD8+ T cell responses.7 It is thought that CD4+ T cell activation and priming are required for CD8+ T cells' ability to achieve viral clearance through cytotoxic effects.8, 9 These processes can also lead to activation of HSCs in response to injury, which in the absence of viral clearance is followed by matrix deposition, fibrosis, and eventually cirrhosis.10, 11 HCV-specific CD4 activity is correlated with histological fibrosis and portal tract inflammation.12–14

A role of HSCs in the inflammatory response to viral infection has been established based on their capacity to present antigen and modulate lymphocyte behavior.15, 16 Although lymphocytes directly interact with HSCs by adhesion,14 the pathways mediating interaction of lymphocytes and HSCs are not well understood. Moreover, HSC are phagocytic based on their internalization of latex particles, bacteria16 as well as apoptotic bodies.17 In murine models, fibrosis is mainly mediated by direct activation of HSCs by CD8 lymphocytes,18 and adoptively transferred CD8 cells are fibrogenic in naïve mice. In addition, NK cells display anti-fibrotic activity by killing activated HSCs that have lost the self-recognition marker, MHC class I.19–21 In this study, we have investigated the morphological and functional interactions between human HSCs and lymphocyte subsets. Our results indicate that HSCs not only physically interact with lymphocytes but may contribute to lymphocyte clearance by cellular ingestion.


α-SMA, alpha-smooth muscle actin; DiOC, dioctadecyloxacarbocyanine perchlorate; FACS, fluorescence-activated cell sorting; FBS, fetal bovine serum; FITC, fluorescein isothiocyanate; GTPase; guanosine triphosphatase; HBV, hepatitis B virus; HCV, hepatitis C virus; HSC, hepatic stellate cell; ICAM-1, intracellular adhesion molecule 1; IFN-γ, interferon gamma; IHL, intrahepatic lymphocyte; MHC, major histocompatibility complex; mRNA, messenger RNA; NK, natural killer; PBL, peripheral blood lymphocyte; PBS, phosphate-buffered saline; PE, phycoerythrin; PI, propidium iodide; siRNA, small interfering RNA.

Patients and Methods

Study Population.

Blood and liver samples of patients referred for liver biopsy because of chronic HBV and HCV were obtained with informed consent after the approval of the Hadassah Hospital Ethics Committee. All HCV patients were positive for serum HCV antibodies (Abbot) and HCV RNA (tested by HCV amplicor, Roche). Similarly, HBV patients were positive for serum hepatitis B surface antigen (Abbot) and HBV-DNA (HBV Monitor, Roche). Only patients with advanced but compensated fibrosis were included in this cohort. Advanced fibrosis was clinically established by the presence of splenomegaly, thrombocytopenia, and irregular liver echotexture, and confirmed by liver biopsy demonstrating Metavir F3 or F4,22 as assessed by a single pathologist. No patient had evidence of HBV/HCV, HBV/human immunodeficiency virus or HCV/human immunodeficiency virus co-infection. Control liver biopsy specimens from HBV/HCV-negative patients were obtained during hepatobiliary surgery for either resection of hemangioma or benign tumors, in which histologically normal liver tissue surrounding the resected lesion was used.

Lymphocyte Isolation.

Human peripheral blood lymphocytes (PBLs) were collected in heparinized tubes from patients and healthy volunteers after the guidelines approved by the Hadassah Hospital Ethics Committee. Mononuclear cells were isolated by centrifugation over Ficoll-Hypaque (Pharmacia) according to Boyum.23 After three washes in saline, cells were resuspended in medium. For freezing, Roswell Park Memorial Institute 1640 medium supplemented with 50% heat-inactivated fetal bovine serum (FBS; Gibco) and 20% dimethyl sulfoxide was used, and cells were stored at −70°C. The culture medium for PBLs was Roswell Park Memorial Institute 1640 with 10% FBS. Human NK, CD4, and CD8 cells were isolated from PBLs using a magnetic cell sorting kit (Miltenyi Biotec) according to manufacturer's instructions. For intrahepatic lymphocytes (IHL),24 liver tissue obtained from needle biopsies or surgery was washed twice in 10 mL phosphate-buffered saline (PBS) to remove remaining PBLs. Tissue was cut into 1-mm pieces and incubated in 2 mL Roswell Park Memorial Institute 1640 (Gibco, Great Britain) for 15 minutes with collagenase type IV (Worthington, Lakewood, NJ) and deoxyribonuclease I (Worthington) at 37°C. After incubation, the tissue was run through a 1-mL pipette and then filtered through a cell strainer fluorescence-activated cell sorting (FACS) tube (Becton Dickinson, Mountain View, CA). Cells were washed and centrifuged at 92 g for 5 minutes at 4°C. The pellet was resuspended in FACS buffer for flow cytometry analysis. The cell yield was between 100,000 to 200,000 cells per biopsy.

In Vitro Analysis of Human Lymphocyte Interactions with HSCs.

The human hepatic stellate cell line LX2 was used to examine interactions of lymphocytes with HSC in culture. This cell line has been extensively validated in a number of studies that establish its relevance to primary HSCs.25 For validation of results, primary human HSCs were used.

Primary Human HSC Isolation and Culture.

HSCs were isolated as previously described.26 Isolated HSCs were cultured in M199 containing 10% FBS, changed 24 hours after plating and every 3 to 4 days thereafter. When the cultures reached confluence, they were trypsinized (0.05% trypsin/0.53 mM ethylenediaminetetra-acetic acid) and passaged at a ratio of 1:3. Subsequent passages were performed every 7 to 10 days.

Activation of HSCs by PBLs.

Healthy or cirrhotic HCV-derived PBLs (106 cells) were isolated from eight donors in each group. Isolated PBLs were then co-cultured individually with LX2 cells or primary isolated HSCs in 18-mm dishes 1% FBS (Atlantic Biologicals) medium. Triplicate cultures were performed from each PBL donor. HSCs (105 cells) were previously cultured 3 days (up to subcomplete confluence) before PBL constitution. Either 106 mixed PBL cells or separate subsets (CD4, CD8, or NK cell) were used in the co-culture. After 24 hours of co-culture activation of HSC medium with lymphocytes was removed, and cells were washed, harvested by cell scraping, and analyzed for alpha0smooth muscle actin (α-SMA) Western blotting and messenger RNA (mRNA) expression.


Confocal imaging was used to investigate phagocytosis of lymphocytes by HSCs using mixed PBLs either derived from HCV-infected patients with cirrhosis or from healthy volunteers. HSC/PBL co-cultures were incubated for 0, 6, 12, 24, and 72 hours. Circular glass slides (Marienfeld, Laboratory Glassware, Germany) placed in the bottom of 12-well flasks (Nunc Brand Products, Denmark) were used in each co-culture. Expressions of CD4, CD8, NK, and α-SMA markers were analyzed at each time point by immunocytochemistry. Thirty-six images from different fields were acquired from each co-culture dish.

Phagocytosis was further assessed using confocal microscopy of 1-μm sections. In addition, PBLs from three different donors were preincubated with 3,3′-dioctadecyloxacarbocyanine perchlorate (DiOC; Sigma). DiOC, a cyanine dye, was dissolved in dimethylsulfoxide (2 mg/mL; Sigma) as described previously to stain intracellular lipids green.27 After overnight incubation with agitation, DiOC-stained cells were washed several times and then cultured with HSCs.

To assess the existence of ligand/receptor-mediated phagocytosis, confocal assessment was repeated in different modifications of HSC/PBL co-cultures. These included preincubation of HSCs with blocking antibodies at a final concentration of 100 μg/mL for 30 minutes and then cultured with HCV-derived PBLs. HSCs were preincubated with either mouse anti-human class I (HLA-ABC Antigen, DakoCytomation), class II (HLA-DP, DQ, DR Antigen, DakoCytomation), intercellular adhesion molecule (ICAM-1) (CD54, ICAM-1, DakoCytomation), the type I membrane glycoprotein adhesion molecule “Integrin alphaV” (polyclonal Intigren, CHEMICON International), MIC-A or MIC-B (BioLegend, 6D4), or control antibodies. For T-cell receptor blocking, mouse anti-human T-cell receptor antibodies were incubated with lymphocytes before co-culture under similar conditions. In separate experiments, either HSCs or PBL cells were irradiated to obtain apoptosis,16, 28 before co-culture to distinguish whether HSCs are responsible for phagocytosis of lymphocytes. HSCs were washed twice with PBS, trypsinized, and irradiated with 6000 rads, a dose shown to inhibit proliferation without affecting cell viability or membrane protein expression.15, 16 Experiments with irradiated nonproliferating PBLs were performed using 3000 rads.15, 16

Fluorescent-Activated Cell Sorting Analysis (FACS).

Briefly, isolated lymphocytes were adjusted to 106/mL in staining buffer (in saline containing 1% bovine albumin; Biological Industries, Israel). Fifty microliters cell suspension was incubated with antibody on ice for 30 minutes, washed with staining buffer, and fixed with 2% paraformaldehyde. Fc receptors were blocked by incubation with 1% human plasma for 15 minutes on ice. Blocked lymphocytes were then mixed with either fluorescein isothiocyanate (FITC), phycoerythrin (PE), allophycocyanin, or peridinin-chlorophyll-α-protein conjugated anti-CD4, anti-CD8, anti-CD16, and anti-CD45 antibodies, respectively (IQ Products, Groningen, Netherlands; antibodies were diluted 1:40) or isotype controls for 20 minutes on ice, and were washed with FACS buffer. CD45 was used as a global marker for leukocytes. Intracellular staining of lymphocytes with anti-IFN-γ and transforming growth factor beta (TGF-β) antibodies (FITC and PE conjugated, respectively) was performed according to the manufacturer's protocol (BD Biosciences). For apoptosis measurements of lymphocytes inside the HSCs, propidium iodide (PI) staining of fragmented DNA and phosphatidylserine staining by annexin V conjugated to FITC (R&D Systems, Minneapolis, MN) were used according to the manufacturer's instruction. Lymphocyte apoptosis was therefore defined as CD45+, annexin-V(+) but PI(−). FACS data were acquired using a FACS Caliber Flow Cytometer (Becton Dickinson); the data were analyzed using the CellQuest 3.3 software. For all analyses, lymphocyte gating was performed using the forward scatter versus side scatter plot.


Liver biopsies or cell culture slides were incubated overnight at room temperature with isotonic PBS, 10% sucrose, and 4% formaldehyde solution. Then they were frozen at −80°C for storage, and 7-μm-thick frozen sections were prepared using Cryostat (Leica CM 3000). Cells were permeabilized using the 0.2% Triton (Sigma). To block nonspecific background staining, 1% bovine serum albumin was used for 20 minutes. After PBS washing, slides were incubated with primary antibodies for 45 minutes at room temperature in the dark. Primary antibodies used in liver biopsies were human FITC-anti-CD4 (at a dilution of 1:100), anti-CD8 (at a dilution of 1:70), and anti-CD16 (at a dilution of 1:100) markers (IQ Products, Groningen, Netherlands). Primary antibodies used in cell culture slides were human FITC-anti-CD4, PE-anti-CD8, PE-anti-CD16, FITC, or PE-anti-CD45 (at a dilution of 1:50), FITC-anti-Rac1, and FITC-anti-Cdc42 (at a dilution of 1:50, Santa Cruz Biotechnology, INC.) markers. For identification of HSCs: α−SMA (DAKO, cat# M0851) primary antibody (at a dilution of 1:150) conjugated to Cy-5 (at a dilution of 1:40) as the secondary antibody (Jackson Immunoresearch) were used. The Cy-5 secondary antibody was then added, preceded and followed by three PBS washes. To preserve staining, sections were stacked and covered with Fluoromount-G (Southern Biotechnology Associates). Sections were then stored at 4°C while awaiting analysis.29, 30

Confocal Microscopy and Image Capture.

A Zeiss LSM 410 Confocal laser scanning system (Zeiss, Germany) attached to a Zeiss Axiovert 135 M microscope was used. The fluorescence images were collected by employing plan-apochromat Zeiss, 40 × 1.4 lens. The system was equipped with an argon laser (488 nm excitation line) for green fluorescence and two helium-neon lasers (543 nm and 633 nm lines) for red fluorescence. Triple-labeled specimens were excited with three lasers and monitored simultaneously using triple detectors and filter-block combinations. The excitation powers and emission filters were tuned to keep the overlap from each channel at a minimum. In each experiment, laser intensity, background level, contrast, aperture, and electronic zoom size were collected at the same level. Fifty images were collected from each specimen and converted to tiff format and processed using Zeiss LSM Image Browser software. Image processing was performed using Adobe Photoshop software (Adobe Systems UK, Uxbridge, and Middlesex, UK) and ImagePro Plus programs (Media Cybernatics, USA).31, 32

α-SMA Immunoblot.

Immunoblot analysis of α-SMA in cultured HSC protein extracts was performed as previously described.18, 19, 21

Real-Time Polymerase Chain Reaction Analysis.

Washed LX2 cells were harvested by a scraper for RNA extraction and complementary DNA transcription as previously described.18, 19 The complementary DNA product was used for real-time polymerase chain reaction as previously described.18, 19 β-Actin served as internal controls and H2O served as a negative control. Primers used for the β-actin were as follows:



For SMA:



HSC Transfection by Small Interfering RNA.

Small interfering RNA (siRNA) were diluted in 100:l in medium without serum to a concentration minimum of 5 nM and maximum of 25 nM. Three microliters HiPerFect transfection reagent (Qiagene) was added to the diluted siRNA and mixed well. The mixture was incubated at room temperature for 10 minutes to allow the formation of transfection complexes. The complex was then added to 105 cells seeded the day before on 24-well plates. The efficiency of transfection was performed using positive and negative siRNA silencing controls provided in Qiagen siRNA human/mouse starter kit. siRNA transfection was validated by immunofluorescence and the gene silencing confirmed by real-time polymerase chain reaction.

siRNA Silencing of Rac1/Cdc42.

LX2 cells or primary HSCs were transfected either with commercial Rac1 or Cdc42 siRNAs using RNA interference human/mouse starter kit (HiPerFect Transfection Reagent, QIAGEN) according to manufacturer instructions: complementary DNA sequences of human Cdc42 and Rac1: 5′-GUG UCG GCA UCA UAC UAA AdTdT-3′ (Cdc42#1) and 5′-CAG CAA UGC AGA CAA UUA AdTdT-3′ (Cdc42#4); 5′-GGU UGG UAU UAU CAG GAA AdTdT-3′ (Rac1#1) and 5′-GAC AUA ACA UUG UAC UGU AdTdT-3′ (Rac1#3). siRNA transfected HSC were then co-cultured with cirrhotic HCV-derived lymphocytes. For controls (available in the same kit), HSCs were transfected with control-siRNA before co-culture. Nonsilencing control siRNA (Alexa Fluor 488 Labeled, QIAGEN) were used for negative control. As a positive control, the specific positive siRNA (MAPK1) was transfected into HSCs but without co-culture with lymphocytes. Six triplicates of co-culture from six different patients were used in each of the first two groups. Activation of HSCs was assessed after 24 hours of co-culture. Medium containing free lymphocytes was removed, and HSCs were harvested by scraping for α−SMA Western blotting assessment.

Statistical Methods.

Results are illustrated as mean ± standard deviation. For statistically significant differences; paired and unpaired Student t test and analysis of variance were used.


Altered Distribution of Lymphocyte Subsets in Chronic HCV and HBV.

Accumulating data implicate several different lymphocyte subsets in hepatic inflammation, but their relative abundance and importance are uncertain.33–36 To clarify this issue, the composition of PBL and IHL from six control cases was compared with 25 HCV and seven HBV patients with advanced fibrosis (Table 1). Significant differences were mainly confined to IHL and not PBL (Fig. 1). Intrahepatic CD4 cells were 34% ± 7.5% of total lymphocytes in healthy controls but only 22.6% ± 6.6% in HCV patients (P = 0.004) and 22% ± 5.3% among HBV patients (P = 0.04). Differences in CD8 cells between healthy controls and patients with either HBV or HCV were not significant (Fig. 1, upper panel). However, the CD4/CD8 ratio was significantly decreased from 0.88% ± 0.17% in healthy donors to 0.53% ± 0.12% (P = 0.007) in HCV and 0.44 ± 0.12 (P = 0.02) in HBV cases. As previously reported,18, 19, 21 significant NK alterations in the current study were confined to IHLs (Fig. 1). Specifically, intrahepatic NK cells were significantly decreased (P = 0.04) from 24.9% ± 2.8% of healthy lymphocytes to 17.7% ± 5% and 19.3% ± 4.1% in HCV-infected and HBV-infected individuals, respectively.

Table 1. Patient Characteristics
  • *


Age* (years)38.7 ± 543.2 ± 1435.6 ± 10.6
ALT* (units)23.3 ± 5105 ± 11456.5 ± 81.8
Albumin*>33 g/L 42.8 ± 5.235.4 ± 3.934.1 ± 2.6
Platelets*<80 × 109/L 230 ± 65120 ± 42118 ± 34
F3 scoreMETAVIR scale None204
F4 scoreMETAVIR scale None53
HCV status:  
Genotype 1 21 
Genotype 2 1 
Genotype 3 2 
Genotype 4 1 
HBV status:   
HBeAg+  5
HBV loadCopies/mL <100,000
Figure 1.

Lymphocyte alterations in fibrotic patients. FACS analysis of isolated intrahepatic and peripheral blood lymphocytes from healthy donors (black bars), HCV (open bars), and HBV (gray bars) fibrotic patients are illustrated (see Patients and Methods). Significant differences were confined to intrahepatic lymphocytes, in which CD4 decreased, leading to an increase of CD4/CD8 ratio (Fig. 1); NK cells decreased in both fibrotic groups.

HSCs and Lymphocytes Interact Directly In Vivo.

Lymphocytes were assessed in situ in liver sections from HCV patients with cirrhosis using confocal microscopy. Scattered, weakly positive α-SMA–positive cells were apparent in control livers (Fig. 2A) but became prominent in fibrotic livers (Fig. 2B-D). Immunostaining for lymphocyte subsets was negative in normal livers, suggesting that lymphocytes do not adhere to liver parenchyma in the absence of injury (Fig. 2A). In contrast, fibrotic livers were positively stained for CD4, CD8, and NK (CD16) cells (Fig. 2B-D, respectively). Lymphocytes were located only in direct proximity to the α-SMA–positive cells along fibrotic septa and not elsewhere. (The classic green color of FITC fluorescence was remapped to blue for easier viewing.) Therefore, FITC-conjugated subsets with the classic blue color became pink when merged with the red Cy-5 α−SMA (Fig. 2B-D). Similar results were also obtained in HBV cirrhotic livers (data not shown).

Figure 2.

Direct lymphocyte–hepatic stellate cell adhesion in situ: Liver biopsies from fibrotic HCV patients (B-D) and healthy controls (A) were stained with primary antibodies. Cy-5 conjugated α-SMA in red was used to stain HSCs, and FITC-CD45 served as a common leukocyte marker. Confocal laser scanning microscopy was used to visualize stained sections as described in Patients and Methods. HSCs (red arrows) were stained as small red cells adjacent to hepatocytes. Stained lymphocytes (white arrows) in blue/purple were only located in a direct proximity to the α-SMA–positive cells along fibrotic septa but not elsewhere. FITC-conjugated CD45 with the classic blue color became pink when merged with the red Cy-5 α-SMA.

Immune Cell Interactions Contribute to Fibrosis.

Alterations of human lymphocyte subsets associated with hepatic fibrosis were similar to those seen in rodent fibrosis models.18–21 To evaluate the functional impact of interactions of lymphocytes with HSCs, a co-culture system was employed, combining mixed PBLs, isolated CD4, CD8, or NK cells with LX2 cells for 48 hours, comparing the effects of cells from normal donors with those from donors with HCV.

Figure 3A shows α-SMA (upper lane) and β-Actin (second lane) expression as assessed by western blotting from cultured protein extracts of these co-cultures. Both CD8 and to a lesser extent CD4 cells (middle panel), as well as mixed HCV lymphocytes (upper panel) from HCV-infected patients could activate HSCs, as manifested by increased α-SMA expression in the LX2 cells. In contrast, NK cells had minimal effects (middle panel). Moreover, neither mixed nor isolated subsets from healthy controls had any effect on LX2 HSCs (Fig. 3A, lower panel). Expression of α−SMA mRNA (Fig. 3B) corresponded to western blot results. Specifically, α-SMA mRNA expression was increased twofold ± 0.8-fold in mixed HCV-derived lymphocytes co-cultured with LX2 cells (Fig. 3B, upper panel), compared with LX2 cells cultured alone. The expression in mixed cultures with CD4 cells was 1.6 ± 0.9, CD8 was 3 ± 0.4, and in NK cells 0.9 ± 0.7. Significant differences were seen when CD8-LX2 co-cultures were compared with co-cultures containing mixed (P = 0.04), CD4 (P = 0.01), and NK (P = 0.004) subsets. Primary isolated HSC revealed the same pattern seen with LX2 cells (Fig. 3B, lower panel). Similar to results in mouse fibrosis models,18–20 NK cells attenuated HSC activation compared with mixed lymphocytes (P = 0.03), reinforcing the relevance of these earlier mouse studies to human disease. Intracellular cytokine analysis of cultured lymphocytes was studied by FACS using healthy or HCV-derived PBL (Fig. 3C). The findings further support the effect of each subset on HSCs: compared with healthy PBLs, HCV-derived NK cells showed a significant decrease of TGF-β secretion from 2.3% ± 1.2% to 1.5% ± 0.8% of NK cells (P = 0.05) and a significant increase (P = 0.03) of IFN-γ secretion from 0.2% ± 0.1% to 0.5% ± 0.4% of NK cells (Fig. 3C, left upper and lower panels). Decreased TGF-β and increased IFN-γ secretion by NK cells are consistent with an antifibrotic effect. HCV-derived CD8 cells, however, showed a significant increase of TGF-β secretion from 9.2% ± 3.6% to 46.1% ± 19.7% (P < 0.0001) and a significant decrease (P = 0.009) of IFN-γ secretion from 5% ± 5.3% to 0.6% ± 0.5% of CD8 cells (Fig. 3C, middle upper and lower panels). Increased TGF-β and decreased IFN-γ secretion from CD8 cells are compatible with their profibrotic effect. TGF-β was also increased from 122 ± 15.2 (P = 0.04) in the case of CD4 subsets; however, IFN-γ secretion was similar in both healthy and HCV-derived CD4 lymphocytes (Fig. 3C, left upper and left lower panels), indicating a milder profibrogenic effect as compared with CD8 cells.

Figure 3.

The profibrogenic and antifibrogenic effect of lymphocyte subsets in culture: LX2 cells were co-cultured 24 hours with lymphocytes from either HCV patients or healthy controls, either as mixed or isolated subsets. (A) Results from two individuals of each group. Protein extracts were evaluated for α-SMA (upper lanes) as a marker for HSC activation and compared with β-actin (second lanes). The upper panel demonstrates the activation of LX2 cells after co-culture with mixed HCV-derived PBLs. The middle panel indicates that HCV CD8 cells, and to a lesser extent CD4 cells as well as mixed HCV lymphocytes, activate HSCs, as manifested by a more intense expression of α-SMA protein as assessed by western blot. NK cells barely activate HSCs. Healthy lymphocytes, in the lower part of the figure, fail to activate HSCs. Results from mRNA analysis for expression of α-SMA (B) correspond to results from western blot (LX2 in the upper panel and primary isolated HSCs in the lower one). Experiments were repeated three times with nearly identical results. Moreover, although this figure derived from two HCV and two healthy donors, results are reproducible when the experiment is repeated with lymphocytes from other cases (data not shown). Intracellular cytokine analysis of cultured lymphocytes was studied by FACS using healthy or HCV-derived PBLs (C). Results show decreased TGF-β and increased IFN-γ secretion by NK cells, decreased TGF-β from CD4 and CD8 cells with increased IFN-γ secretion only in the CD8 subsets.

Very similar results to Fig. 3A-C were also obtained using primary HSCs instead of LX2 cells (data not shown).

Direct Contact/Attachment Is Followed by Phagocytosis of Lymphocytes by HSCs.

We previously demonstrated direct contact of HCV-derived NK cells with cultured human HSCs.19 To extend these findings, LX2 cells were co-cultured with mixed PBLs from HCV-infected donors with cirrhosis and healthy donors for 0, 6, 12, 24, and 72 hours (as detailed). LX2 cells were stained red by Cy-5 antibody conjugated to anti-α−SMA (Fig. 4, red arrows). Figure 4 demonstrates representative images of interactions (for 6 hours' co culture) of the three different lymphocyte subsets (white arrows) in proximity to HSCs (red arrows). This is including FITC-conjugated CD4 (Fig. 4A, blue), PE-CD8 (Fig. 4B, green), as well as PE-CD16 (NK) HCV-derived cells (Fig. -4C, green). These interactions were not present when LX2 cells were cultured with healthy lymphocytes (Fig. 4D). Thus, direct interaction between PBL and LX2 cells could contribute to the functional effects of lymphocytes from HCV patients on HSCs. Similar to findings in LX2 cells, HCV-derived lymphocyte subsets (Fig. 5A -C) but not healthy PBLs (Fig. 5D) were also engulfed within primary HSCs.

Figure 4.

Direct contact ends as phagocytosis of lymphocytes by HSC: LX2 cells were co-cultured with HCV or healthy lymphocytes for 6 hours. LX2 cells were stained red by the Cy-5 conjugated to α-SMA (red arrows) in direct proximity to HCV lymphocytes (white arrows) for FITC-conjugated CD4 (A; blue), PE-CD8 (B; green), as well as PE-CD16 (NK) HCV-derived cells (C; green). This interaction was not demonstrated using healthy PBL controls (D).

Figure 5.

Similar results of Fig. 4 were seen using primary HSCs instead of LX2 cells.

Based on the direct proximity of lymphocytes to HSC at early time points (Figs. 2, 4, and 5), we monitored the HSC/PBL co-culture longer than 6 hours. Phagocytosis of CD45 cells (as a pan leukocyte marker, conjugated to FITC in the case of co-culture with LX2 cells and to PE in the case of primary isolated HSC) by HSCs was found at 12, 24, and 72 hours of co-culture. HSCs (Fig. 6A -D; LX2 in Fig. 6A-B; as well as primary HSCs in Fig. 6C-D) were stained red by the Cy-5 conjugated to α−SMA (red arrows), and lymphocytes were stained blue (white arrows), phagocytosis was associated with α-SMA–positive encapsulation around ingested lymphocytes (Fig. 6B and D, green arrows). Furthermore, serial 1-μm section by confocal images demonstrated HSCs within the same image plane because both appear and disappear simultaneously, confirming the presence of phagocytosed lymphocytes (white arrows) within HSCs (red arrows; selective sections are presented in Fig. 7A -D with LX2 cells and Fig. 7G-J with primary isolated HSCs). To further demonstrate phagocytosis and not physical overlay of the two cell types, HCV-derived lymphocytes were preincubated with DiOC to stain intracellular lipid content, washed, and then cultured with HSCs. Fig. 7E-F with LX2 cells and Fig. 7K-M with primary isolated HSCs show lymphocytes inside HSCs with diffusion of lymphocyte-derived DiOC within the cytoplasm of HSCs.

Figure 6.

After 12, 24, and 72 hours of co-culture, all CD45+ cells underwent phagocytosis inside the HSC (A-B, with LX2 cells; C-D, with primary isolated HSCs), which were also demonstrated in each lymphocyte subset separately (data not shown). Moreover, phagocytosis was associated with α−SMA–positive encapsulation around ingested lymphocytes (green arrows, B and D). This experiment was repeated more than four times.

Figure 7.

Confocal microscopy documents phagocytosis: Serial 1-μm section images showed HSC and CD45+ cells appear and disappear at the same confocal channel, confirming phagocytosis (selective sections are presented in A-D with LX2 cells and in G-J with primary isolated HSCs). HCV-derived lymphocytes were preincubated with DiOC, washed, and then cultured with HSC cells. (E-F) LX2 cells and (K and L) primary isolated HSCs demonstrate lymphocytes inside the HSCs, with release of their DiOC into the HSC cytoplasm. The same was also demonstrated for each lymphocyte subset individually (data not shown). The experiment was repeated four times.

The apoptosis of lymphocytes was then evaluated using HCV-derived PBLs in culture for 48 hours either alone or with HSC (Fig. 8). In the case of co-culture with LX2 cells, apoptosis was evaluated only in the engulfed PBLs after washing the floating cells. Apoptotic PBLs (Fig. 8A), defined as CD45+, Annexin+, and PI(−), were significantly increased from 19% ± 4.5% in the case of PBL culture alone to 25.7% ± 1.7% of total CD45+ cells in the case of engulfed lymphocytes (P = 0.002). Interestingly, the dead lymphocytes (Fig. 8b), defined as CD45+, Annexin+, and PI(+), were also increased even more significantly from 7.7% ± 0.8% to 26.2% ± 2.2%, respectively (P < 0.0001). The data suggest a rapid killing of lymphocytes inside HSCs. Apoptosis of lymphocytes co-cultured with primary HSCs showed the same pattern (Fig. 8C-D).

Figure 8.

The apoptosis and death of lymphocytes increased after phagocytosis by HSCs: The apoptosis of lymphocytes was evaluated by FACS after HCV-derived PBL culture for 48 hours either alone or with HSCs (LX2 in panels A-B as well as primary isolated HSCs in panels C-D). Apoptotic PBLs (A), defined as CD45+ Annexin+ and PI(−), were significantly increased in the case of engulfed lymphocytes. Dyed lymphocytes (B) defined as CD45+ Annexin+ and PI(+) were also increased, suggesting a rapid killing of lymphocytes inside the HSCs.

We next explored whether phagocytosis of lymphocytes by HSCs was regulated by specific ligand–receptor interactions, focusing on several families of molecules. Phagocytic receptors are very diverse and include at least one member of each prototypical adhesion receptor family (integrin, cadherin, and so forth).37 Figure 9 demonstrates selected examples from each condition in which LX2 cells were stained red by the Cy-5 conjugated to α-SMA (red arrows), and lymphocytes were stained with PE-CD45 cells (white arrows). Compared with the phagocytosis of lymphocytes seen in untreated coculture (Fig. 9A), phagocytosis was not affected when HSCs were preblocked by specific antibodies to either MHC class I and class II molecules (Fig. 9B-C) or natural-killer group 2, member D (NKG2D) receptors for MHC class I chain-related gene A (MIC-A) and MHC class I chain-related gene B (MIC-B) ligands (data not shown). Similarly, phagocytosis was not affected when lymphocytes were preblocked by specific antibodies to T-cell receptor (Fig. 9F). In contrast, phagocytosis was blocked by preincubation with antibodies to either ICAM-1 or integrin alphaV before co culture (Fig. 9D-E). To confirm that LX2 cells are responsible for phagocytosis of lymphocytes rather than penetration of LX2 cells by lymphocytes, either lymphocytes (Fig. 9G) or LX2 cells (Fig. 9H) were irradiated to attenuate apoptosis in only one of the two cell types before co-culture. Only when HSCs (but not lymphocytes) were irradiated was phagocytosis completely blocked. This finding is consistent with results in which radiation has attenuated allogeneic T-lymphocyte proliferation when co-cultured with HSCs.16 Primary HSC followed similar responses as LX2 cells (Fig. 10).

Figure 9.

Phagocytosis is mediated by a ligand/receptor adhesion. Compared with the nonmanipulated co-culture and unstimulated phagocytosis (A), prevention of phagocytosis was achieved when the HSC-related ICAM1 and integrin alphaV were blocked before co-culture as described in Patients and Methods (D and E, respectively). Blocking of HSC-related class I as well as class II before co-culture with lymphocytes did not affect phagocytosis (B and C, respectively). Blocking of lymphocyte-related T9cell receptor before co-culture with lymphocytes did not affect phagocytosis (F). Only when HSC (H) but not lymphocytes (G) were irradiated was phagocytosis blocked. (I) Absence of phagocytosis when HSC were co-cultured with healthy lymphocytes.

Figure 10.

Primary HSCs displayed the same responses as LX2 cells in Fig. 9. The experiment was repeated four times.

The blocking of ICAM-1 or integrin alphaV before co-culture with HCV-derived PBLs (Fig. 9D-E) was accompanied by a decrease of HSC activation (Fig. 11). Specifically, expression of α−SMA assessed by western blotting was decreased after ICAM-1 or integrin alphaV blocking in the presence of equal β-actin (Fig. 11, lower panel). Expression of α−SMA mRNA (Fig. 11, upper panel) corresponded to the western blot results. Specifically, α-SMA mRNA expression was increased 2.8 ± 0.05-fold in mixed HCV-derived lymphocytes co-cultured with LX2 cells, compared with LX2 cells cultured alone. The expression significantly decreased to 1.9 ± 0.09 (P = 0.005) after ICAM-1 blocking, and to 1.6 ± 0.02 (P = 0.005) in integrin alphaV blocking. Significant differences were also seen when ICAM-1 blocking was compared with integrin alphaV blocking (P = 0.02). Primary HSCs responded similarly to LX2 cells after ICAM-1 or integrin alphaV blocking (data not shown).

Figure 11.

The blocking of ICAM-1 or integrin before co-culture with HCV-derived PBL functionally accompanied with a decrease of the HSC activation: The bands of α-SMA expression by western blotting were decreased after ICAM-1 or integrin blocking in the presence of equal β-actin (lower panel). Expression of α-SMA mRNA (Fig. 10, upper panel) corresponded to western blot results.

Rac1 & Cdc42 Are Involved in the Phagocytosis of Lymphocytes by HSC.

After the recognition that intercellular adhesion molecules contribute to the phagocytic response of HSCs, we sought to identify the underlying intracellular signaling pathways. Signaling cascades regulating cellular adhesion parallel those activated during phagocytosis.37 In particular, Rho-family guanosine triphosphatases (GTPases) are activated downstream of adhesion receptors and control the cytoskeletal changes that underlie adhesive events.38 Accordingly, we examined the potential roles of Cdc42 and Rac1 (as part of the Rho-family GTPases) by characterizing these two molecules with specific antibodies in the co-culture system (Fig. 12). Overlay of both molecules together with that of Cy-5 red α−SMA staining the HSC was apparent (Fig. 12), suggesting that Cdc42 and Rac1 recruitment contribute to HSC–lymphocyte adhesion and phagocytosis.

Figure 12.

Rac1 & Cdc42 are recruited to the phagocytosis contact area. Cdc42 and Rac1 recruitment in the phagocytosis and cell-to-cell adhesion area are demonstrated in HSC/HCV-derived PBL co-cultures (the left two panels showing co-culture with LX2 cells and the right two panels with primary HSCs). (B and E) FITC-stained Cdc42 and Rac1 (respectively) in blue; (C and F) PE-stained lymphocytes in green. (A and D) Overlay of both stains together with that of Cy-5 red α-SMA staining of LX2 cells (respectively). The experiment was repeated more than four times.

To uncover the functional importance of Rac 1 and Cdc42 to the adhesion and phagocytosis of lymphocytes by HSCs, we employed selective knockdown using specific siRNAs,38 confirmed by using confocal microscopy of fluorescent control siRNA (data not shown). Figure 13 documents Cdc42 down-regulation by specific siRNAs as assessed by western blotting (lower panel), which was associated with attenuated phagocytosis as demonstrated by confocal microscopy (data not shown because they are similar to Fig. 9D-E, H) associated with reduced HSC activation as assessed by α-SMA expression (upper panel). Results of the Rac1 siRNA silencing were identical to those using Cdc42 silencing (data not shown). Cdc42 and Rac 1 were also involved in the phagocytosis of PBLs by primary HSC (data not shown).

Figure 13.

Rac1 and Cdc42 are required for phagocytosis of lymphocytes by HSC. Silencing of either Cdc42-related or Rac1-related LX2 cells was achieved using specific siRNA transfection. Cdc42 down-regulation was confirmed by western blotting after the specific siRNA silencing compared with negative siRNA control (lower panel). As a result, there was decreased activation of HSC as assessed by α-SMA expression compared with negative siRNA control (upper panel). Results of the Rac1 siRNA silencing were in line with the Cdc42 effects (data not shown). This experiment was repeated four times.


During viral infections, cytotoxic CD8+ T cells are activated by specific peptides, and their response is enhanced by specific helper CD4+ T cells. Attack of activated T cells on HCV-infected hepatocytes may explain, at least in part, histological damage.12, 39–44 Our study has evaluated T-cell subset distribution in PBLs and intrahepatic T cells from patients with HBV-related and HCV-related liver fibrosis. Significant differences were confined to IHLs and not PBLs, suggesting that lymphocytes may be sequestered at the site of infection and that assessment of PBL may not be representative of local interactions in liver.45, 46 Indeed, alterations of the intrahepatic cytokine network may be the basis for T-cell alterations in liver injury.47

As others have previously reported in inflammatory liver injury,14, 48, we have found a decrease in the percentage of CD4 T cells in the fibrotic livers compared with the peripheral blood, leading to a significantly reduced CD4/CD8 ratio, even though there is no significant change in CD8 cell abundance.

NK cells are cytotoxic cells of the innate immune system with inhibitory receptors recognizing class I MHC, and a variety of activating receptors including NKG2D.49 We and others have reported an anti-fibrotic effect of NK cells in vivo and in vitro.19, 20, 50 in which NK cells become stimulated in liver injury as the expression of inhibitory killing immunoglobulin receptors decrease, compared with the activation receptors. HSC activation, conversely, leads to reduced expression of class I molecules, leading NK cells to recognize HSCs as ‘non-self’ that provokes HSC killing.19 NK cells (CD16+) significantly decrease in fibrotic infected patients (Fig. 1), thereby reducing an important anti-fibrotic subset. In support of this possibility, Morishima et al.51 reported a decreased frequency of NK cells, similar to earlier studies in blood52–54 and liver.55, 56 NK cell function is reduced in patients with primary biliary cirrhosis57 and in HCV-infected individuals.58–60 Our current findings further support an anti-fibrotic role of NK cells, in part through direct interaction with HSC in vivo and in vitro (Fig. 4-7) as well as through altered cytokine expression that reduces their activation (Fig. 3C). Moreover, NK cell numbers are reduced in virally infected livers, thereby attenuating an important anti-fibrotic pathway (Fig. 1). Direct NK–HSC adhesion is accompanied by evidence of increased apoptosis of activated HSCs and liver NK cells in situ.19

Although lymphocyte infiltration is a prominent feature of viral hepatitis, their functional impact on HSC activation has not been well established in human disease, whereas animal models have suggested a direct fibrogenic activity.18, 19, 21 In this study, evidence of fibrogenic stimulation associated with a direct physical interaction and an enhanced fibrogenic cytokine profile is suggested by the co-culture results, wherein CD8+ T-cell subsets harbored the greatest fibrogenic activity, and NK cells decreased the activation of HSC (Fig. 3). The finding significantly expands earlier observations implicating CD8+ T cells in biliary fibrosis associated with murine graft-versus-host disease34 and in alcoholic cirrhosis.35

Phagocytosis of lymphocytes by HSCs is consistent with recent reports documenting the phagocytic capacity of HSCs towards macromolecules.16, 61, 62 Moreover, because engulfment of hepatocyte apoptotic bodies is fibrogenic,17 the same may hold true for phagocytosis of lymphocytes by HSCs. The capacity to internalize extracellular material has also been described in other fibrogenic cell types (i.e., mesangial cells) as well as in hepatic nonprofessional antigen presenting cells such as sinusoidal endothelial cells.62, 63 Our results are consistent with the findings of a recent study showing that activated HSCs can phagocytose apoptotic bodies.17 HSCs can phagocytose both apoptotic and nonapoptotic lymphocytes (Fig. 8), and increased apoptosis and death of lymphocytes inside HSC may be an inactivation mechanism of lymphocytes. The loss of lymphocyte engulfment by irradiated HSC implicates HSC phagocytosis in this process. The translocation of cytoplasmic lymphocyte contents into the HSC cytoplasm demonstrated by DiOC diffusion into the HSC cytoplasm provides additional further evidence and explains the different effects of each lymphocyte subset on HSC responses. Phagocytosis of activated lymphocytes by the HSCs was also found in vitro in liver injury because of graft-versus-host disease after bone marrow transplantation in the absence of viral infection (data not shown). Therefore, activated but not infected lymphocytes are responsible for in vitro HSC activation. We therefore suggest that phagocytosis of lymphocytes in the human in vivo setting is an additional inflammatory mechanism to activate HSC in addition to the recognized paracrine inflammatory pathways. The engulfment of apoptotic cells by phagocytes was reported to prevent the release of potentially toxic or immunogenic intracellular contents from the dying cells.65 Similar results have been found with fibroblasts and epithelial cells.63 Therefore, the phagocytosis of lymphocytes inside HSCs provokes the concentration of immunogenic intracellular contents within the cells. In this case, the prominent engulfed lymphocyte subsets will dictate the behavior of HSCs. We reported previously that apoptosis of LX2 cells occurs when they are co-cultured with NK cells,19 and here we demonstrate that those NK cells are IFN-γ secretors, and their phagocytosis leads to inactivation of LX2 cells (Fig. 3). Conversely, LX2 proliferation and activation are increased in the case of engulfed TGF-β–enriched CD8 subsets (Fig. 3). Because healthy lymphocytes do not undergo phagocytosis, their co-culture with LX2 cells did not alter the expression of SMA (Fig. 3A, lower panel).

We further explored the potential pathways specifically mediating phagocytosis. Phagocytosis was initiated among different HSC surface adhesion ligands/receptors that are involved with lymphocyte interaction, including CD11c, and ICAM-1.65 In contrast, neither MHC class I and II, nor NK cell receptor–ligand interactions are important for HSCs to display antigen presentation properties,16 whereas both class II and CD11c are essential for antigen presentation.66

Recent studies implicate cytoskeletal rearrangements mediated by Rho GTP-binding proteins.67 Rho proteins or their downstream effectors can be recruited to the adhesion structure, for example, Rac1 and Cdc42, and co-localize at cell–cell contacts.68 Indeed, Rac1 and Cdc42 silencing blocked phagocytosis and prevented HSC activation. Rac1/Cdc42 is not the only pathway involved in phagocytosis, and other mechanisms could be also involved. Phosphatidylserine and phosphatidylserine receptor interactions have been suggested as an alternative mechanism in HSCs. Phosphatidylserine on apoptotic cells promotes their uptake and induces anti-inflammatory responses in phagocytes, including TGF-β release.69 Interaction between phosphatidylserine and the phosphatidylserine receptor inhibits immune responses in vivo.70 This interaction is relevant to the LX2 cells, because they express the phosphatidylserine receptor.61

In summary, our results imply that human lymphocytes may mediate not only liver injury but also the fibrogenic response through direct, regulated interactions with hepatic stellate cells. The findings further expand a growing body of evidence that defines specific lymphocyte interactions in hepatic fibrogenesis.