|Home | About | Journals | Submit | Contact Us | Français|
Hepatic stellate cells (HSCs) are critical for hepatic wound repair and tissue remodeling. They also produce cytokines and chemokines that may contribute to the maintenance of hepatic immune homeostasis and the inherent tolerogenicity of the liver. The functional relationship between HSCs and the professional migratory APCs in the liver, i.e. dendritic cells (DCs), has not been evaluated. Here, we report that murine liver DCs co-localize with HSCs in vivo under normal, steady-state conditions, and cluster with HSCs in vitro. In vitro, HSCs secrete high levels of DC chemoattractants, such as MIP1α and MCP-1, as well as cytokines that modulate DC activation, including TNFα, IL-6 and IL-1β. Culture of HSCs with conventional liver myeloid (m) DCs resulted in increased IL-6 and IL-10 secretion compared to that of either cell population alone. Co-culture also resulted in enhanced expression of co-stimulatory (CD80, CD86) and co-inhibitory (B7-H1) molecules on mDCs. HSC-induced mDC maturation required cell-cell contact and could be blocked, in part, by neutralizing MIP1α or MCP-1. HSC-induced mDC maturation was dependent on activation of STAT3 in mDCs and in part on HSC-secreted IL-6. Despite up-regulation of co-stimulatory molecules, mDCs conditioned by HSCs demonstrated impaired ability to induce allogeneic T cell proliferation, which was independent of B7-H1, but dependent upon HSC-induced STAT3 activation and subsequent up-regulation of IDO. In conclusion, by promoting IDO expression, HSCs may act as potent regulators of liver mDCs and function to maintain hepatic homeostasis and tolerogenicity.
Despite an elaborate in-house network of immune cells (NK cells, NKT cells, Kupffer cells [KCs], dendritic cells [DCs] and T cells) (1, 2), the liver exhibits remarkable tolerogenic properties, as evidenced by its retention of pathogens (e.g. the malaria parasite, and hepatitis B and C viruses), and its roles in oral and portal venous tolerance and tumor metastasis. This tolerogenic state is exemplified by the acceptance of liver allografts across MHC barriers without immunosuppressive therapy in animal models (3, 4), and the relative ease of acceptance of human liver transplants (5-8). However, acute or chronic liver graft rejection in a significant number of cases continues to be a major clinical challenge due to inadequate understanding of the mechanisms by which the hepatic immune system promotes and maintains tolerance.
Professional liver-resident APCs (DCs) constitute <1% of the non-parenchymal cell (NPC) population, yet they play an important role in regulation of ischemia-reperfusion injury (9, 10), liver transplant rejection (11) and hepatic fibrosis (12, 13). DCs acquire Ag in the local microenvironment and migrate to secondary lymphoid tissues, where they present processed Ag to T cells bearing Ag-specific receptors. DCs express co-stimulatory molecules, in particular CD80 and CD86, and other co-regulatory molecules, such as B7-homologue-1 (B7-H1; = programmed death ligand-1), and induce Ag-specific T cell responses that mediate allograft rejection or acceptance (14).
Both myeloid (m) DCs and non-conventional plasmacytoid (p) DCs (15, 16) are found in the hepatic microenvironment (2). They display lower levels of co-stimulatory molecules and have poor T cell allostimulatory capacity compared to their counterparts in blood and secondary lymphoid tissues (17-20). Mechanistically, low cell surface expression of CD80 and CD86 by liver DCs is associated with IL-6-driven STAT3 activity (21, 22) in the steady state. STAT3 drives the induction of co-inhibitory B7-H1 (23), as well as the immunoregulatory enzyme, IDO (24). Notably, IDO-mediated control of T cell proliferation may play a role in hepatic tolerance (25-27). While liver DCs have been shown to express IDO (28), there is a lack of understanding of mechanisms underlying IDO induction in liver DCs.
Hepatic stellate cells (HSCs), localized in the perisinusoidal space, constitute 8-10% of total liver cells. They make direct contact with hepatocytes and with cells of the sinusoids, including sinusoidal endothelial cells (SECs) and KCs, via their cytoplasmic extensions. In addition, HSCs display Ag-presenting and T cell co-stimulatory molecules, and produce various growth mediators and immune-modulating cytokines and chemokines (29-32). Thus, HSCs have potential to regulate the functions of many cell types, including hepatic immune cells. Indeed, HSCs can present lipid and peptide Ags to NKT and CD4+/CD8+ T cells respectively (33), induce T cell apoptosis via a B7-H1-dependent mechanism (34) and promote IL-2-dependent expansion of regulatory T cells (Tregs) (35). HSCs also enhance DC- and TGF-β-mediated expansion of Tregs, but block TGF-β-induced differentiation of Th17 cells (36). Activation of HSCs is associated with increased expression of B7-H4 that inhibits CD3/CD28-induced activation/proliferation of CD8+ T cells (37). Furthermore, HSCs inhibit splenic DC-induced proliferation of CD8+ T cells in a CD54-dependent manner (38), and induce apoptosis of conventional CD4+ T cells while expanding Tregs (39). However, whether HSCs affect the maturation and function of hepatic resident DCs directly is unknown.
In this investigation, we observed direct association between HSCs and DCs in situ. In vitro co-culture experiments revealed that HSCs recruit and activate syngeneic liver mDCs, as evidenced by increased expression of CD80, CD86 and B7-H1. HSC-induced maturation of mDCs was contact-dependent and STAT3-mediated. Compared to control liver mDCs, mDCs conditioned by HSCs were inferior stimulators of allogeneic CD4+ and CD8+ T cell proliferation due to STAT3-dependent induction of IDO. These novel data reveal a previously undocumented role for HSCs in modulating the phenotype and function of liver mDCs, and provide insight into the mechanisms used by liver mDCs to undermine T cell responses.
All protocols were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC), in accordance with NIH guidelines. Eight- to twelve-week-old male C57BL/6 (B6), B6.129-Indotm1Alm/J (IDO−/−) or BALB/c mice (The Jackson Laboratory), IL-6−/− mice on a mixed B6.129 background and B6.129 control mice (bred at the University of Pittsburgh; a generous gift from Dr. A. Jake Demetris, Department of Pathology) and B7-H1−/− on a B6 background bred from pairs provided by Dr. Lieping Chen (Johns Hopkins University, Baltimore, MD) were provided free access to standard laboratory chow and water.
HSCs were isolated from B6 mice as described (39, 40). Briefly, livers were digested with collagenase- and protease-containing HBSS. Cells were separated from the digested livers, suspended in HBSS and the suspension filtered through nylon mesh. The filtrate was centrifuged 3-4 times at 50g for 1 min to remove hepatocytes and cell debris, and the supernatant then centrifuged at 1400g for 7 min. The pellet containing NPCs was suspended in HBSS, and after purification via Nycodenz gradient centrifugation, HSCs were suspended (1 × 106 cells/ml) in DMEM containing 10% FBS, 10% horse serum, 100 μg/ml penicillin and 100 U/ml streptomycin. Cells were seeded on gelatin (0.1% in PBS)-coated plates at a density of 0.5×106/cm2 and 20 min later, loosely adherent HSCs were harvested and re-seeded in new 6-well plates. The medium was renewed after 3h and then on alternate days. Cell purity, as determined by desmin and glial fibrillary acidic protein (GFAP) staining, was consistently >95%. The cells were used on day 7 of culture, when nearly 90% cells expressed α-smooth muscle actin (α-sma) (a marker of HSC activation), while retaining significant retinoid content (Supplemental Figure 1).
DCs were isolated from B6 (WT or IDO−/−) mice injected for 10 days with the endogenous DC poietin fms-like tyrosine kinase 3 ligand (Flt3L) (Amgen, Seattle, WA) (10 μg/mouse/day, i.p.) as described (17, 19). Following collagenase digestion of the liver, hepatocytes were removed from the cell suspension by low speed centrifugation. NPCs were then enriched by centrifugation on a 30% (w/v) Histodenz (Sigma-Aldrich) gradient, and depleted of NK1.1+ cells using anti-NK1.1 PE Ab, followed by incubation with anti-PE labeled magnetic beads (Miltenyi). pDCs were separated by positive selection using magnetically-labeled anti-PDCA-1 Ab (Miltenyi). mDCs were selected from the NK1.1−PDCA-1− elutant with magnetically-labeled anti-CD11c (Miltenyi). The purity of pDCs and mDCs was consistently >90%.
Liver cryostat sections of untreated or Flt3L-treated mice were stained using rabbit anti-desmin Ab (Abcam) and hamster anti-mouse CD11c Ab (BD PharMingen). Goat anti-rabbit Alexa 488 (green) and goat anti-hamster Cy3 (red) Abs were used along with Phalloidin Alexa-647 (Invitrogen). Nuclei were stained with Hoescht’s stain (Sigma). To examine in vitro interactions, 2.5 × 105 mDCs, labeled with CFSE (Celltracker green, Invitrogen) per the manufacturer’s instructions, were added to chamber slides containing HSCs (1.25 × 104). After overnight incubation in 500 μl RPMI-1640, the co-cultures were fixed with 2% paraformaldehyde in PBS for 1h. The cells were then permeabilized and stained for polymerized actin (rhodamine phalloidin, red; Invitrogen) and DNA (Hoechst’s stain, blue; Sigma). Images were acquired using an inverted Fluoview 1000 confocal microscope (Olympus, Peabody, MA). Stacks through the cells were reconstructed to produce 3D images.
Freshly-isolated DCs or DCs harvested after overnight culture with or without HSCs (attached to the culture plate) were stained in Cell Staining Buffer [CSB-PBS containing 0.5% bovine serum albumin]. Non-specific binding was blocked with CSB-PBS containing 2% goat serum and anti-CD16/32 (BD Bioscience), then the cells were labeled with the following Ab panels for flow cytometric analysis: anti-CD11c Pacific Blue, anti-B220 PeCy7, anti-CD80 APC, anti-CD86 PECy5, anti-B7-H1 PE and anti-I-Ab FITC Abs or anti-CD11c Alexa Fluor700, anti-B220 PE Cy7, anti-CD86 PeCy5, anti-I-Ab PE, anti-CD80 APC, anti-CD54 FITC and rabbit anti-desmin with goat anti-rabbit Pacific Blue or anti-CD11c Pacific Blue, anti-CD80 APC, anti-B7-H1 PECy5, anti-CD86 FITC and anti-IL-6R PE (BD Bioscience). In some experiments, cells were stained with anti-CD54 FITC and a primary rabbit anti-mouse desmin with a secondary goat anti-rabbit Pacific Blue. For intracellular IDO staining, cells were fixed with 2.0% paraformaldehyde, permeabilized with CSB containing 0.1% saponin, then stained with a primary rat anti-mouse IDO Ab (Biolegend) followed by a secondary goat anti-rat FITC (Cell Signaling). In some experiments intracellular IL-6 staining (PE labeled anti-IL-6 Ab; Biolegend) was also performed. A minimum of 20,000 live events were acquired on either a BD LSR II or a BD Fortessa (BD Bioscience). Data were analyzed using FlowJo v7.5.5 (Treestar, Inc).
DCs (1 × 106) were treated with STAT3 Inhibitor VII (250 nM) (Calbiochem) for 1-2h, washed with PBS, then cultured with either IL-6−/− HSCs, B6.129 control HSCs or B6 HSCs (5 × 104) overnight. DCs were collected and analyzed for expression of surface markers, IDO or used in CFSE-dilution assays.
Bulk CD4+ and CD8+ T cells were isolated from BALB/c spleens by negative immunomagnetic selection. RBC-depleted splenocytes were labeled with purified rat anti-mouse-I-Ab, -CD16/32, -B220, -Gr-1, -Ter119 mAbs (all from eBioscience), then depleted using magnetically-conjugated anti-rat IgG (Dynal Beads, Invitrogen). T cells were labeled with CFSE (Invitrogen) per the manufacturer’s instructions, then co-cultured with either liver DCs harvested from liver DC/HSC co-cultures, or with DCs that had been cultured alone in parallel overnight. T cells (1 × 105) were stimulated with DCs (1 × 104) for 4 days, harvested, then stained with anti-CD4 APC and anti-CD8 Pacific Blue (BD Biosciences). In some experiments, an IDO inhibitor, 1-methyltryptophan (1-MT; Sigma; 200 μM) or vehicle control (0.1N NaOH) was added at the start of the culture. A minimum of 20,000 live events were acquired on either a BD LSR II or a BD Fortessa. Data were analyzed using FlowJo v7.5.5.
Cell-free supernatants were collected from overnight cultures of DCs (1 × 106), HSCs (5 × 105) or HSCs+DCs in complete RPMI-1640 medium containing 10% FCS (Hyclone). Cytokines/chemokines in the supernatants were quantified using a Luminex™ 100 IS apparatus (Luminex, Austin, TX) and the BioSource International (San Diego, CA) Mouse 20-plex® Luminex™ beadset. In some experiments, IL-6 levels were quantified in cell-free supernatants by ELISA (Biolegend).
CD11c+ liver mDCs recovered following overnight culture (without or with HSCs) were washed twice with ice-cold PBS, and lysed in RIPA lysis buffer (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) on ice for 20 min. Proteins in the cell lysates (5.0 μg) were resolved by 12% PAGE, transferred onto polyvinylidene difluoride membrane and blocked with 5% skim milk (in PBS containing 0.1% Tween-20) for 2h. Total SOCS3 was probed using primary rabbit-polyclonal anti-SOCS3-Ab (a kind gift from Dr. Giorgio Raimondi, Thomas Starzl Transplantation Institute, University of Pittsburgh) overnight at 4°C and donkey anti-rabbit-HRP Ab (GE Healthcare UK Limited, Buckinghamshire, UK), and detected using ECL Western blotting detection reagent (GE Healthcare). Densitometric analysis was carried out using ImageJ software. Expression of total SOCS3 was normalized with β-actin as internal control.
Data were analyzed by Student’s paired ‘t’-test using Excel (Microsoft). A ‘p’-value of <0.05 was considered significant.
Liver DCs localize predominantly in the portal areas of normal liver tissue (41, 42). However, in addition to portal myofibroblasts (Fig. 1A), confocal imaging of normal B6 mouse liver identified CD11c+ cells in close proximity to HSCs (Fig. 1B). This spatial relationship clearly indicated close physical association between HSCs and DCs, and should be considered to be highly significant since DCs constitute <1% of liver cell population and HSCs about 8-10%. HSC-DC interaction increased markedly following mobilization of DCs into the liver by treatment of mice with Flt3L (Figs. 1C, 1D). Overnight co-culture of HSCs with freshly-isolated syngeneic liver CD11c+PDCA1−NK1.1− conventional mDCs resulted in accumulation of mDCs around HSCs (Figs. 1E, 1F), mimicking their in vivo interaction. These observations suggest that interactions between HSCs and DCs may have functional consequences.
HSCs express ICAM-1 (CD54) (38, 43) and desmin (29, 30) (Supplemental Fig. 2A). Liver mDCs cultured with HSCs up-regulated CD54 expression, but displayed low levels of desmin, verifying that the non-adherent CD11c+ cells from HSC-DC co-cultures were DCs, not HSCs (Supplemental Fig. 2A). Furthermore, mDCs, compared to HSCs, were much smaller (Figs. 1E, 1F and Supplemental Fig. 2B), providing an additional means to distinguish DCs from contaminating HSCs by flow cytometry.
HSCs produce numerous chemokines and cytokines that may influence the immunological activity of liver DCs. To assess whether interaction between these two cell types might influence each other’s cytokine and chemokine production in the hepatic microenvironment, supernatants were collected at 24h from mDC or HSC mono-cultures or HSC+mDC co-cultures, and analyzed for specific cytokines and chemokines by Luminex. HSCs cultured alone secreted high levels of MCP-1, CXCL1 and MIP1α (CCL3) (Fig. 2A), suggesting their possible involvement in recruitment of monocytes and neutrophils in the liver. HSCs may also play a direct role in innate immunity, since CXCL1 and MIP1α possess anti-bacterial (44) and anti-viral activity (45), respectively. Co-culture of HSCs with mDCs did not alter the levels of MCP-1, CXCL1 and MIP1α production (Fig. 2A). HSCs secreted IFNγ-induced protein 10 (IP-10), that recruits lymphocytes, neutrophils and monocytes, while mDCs secreted low levels of monokine induced by IFNγ (MIG=CXCL9), that selectively recruits activated DCs; secretion of both these chemokines increased markedly when mDCs and HSCs were cultured together. These data suggest that interaction between mDCs and HSCs may lead to enhanced recruitment of immune cells into the hepatic microenvironment and that HSCs may induce DC chemotaxis.
HSCs and mDCs, cultured alone overnight, produced the pro-inflammatory cytokines TNFα and IL-1β, respectively, and both cell types produced low levels of IL-1α (Fig. 2B). Co-culture of HSCs with mDCs did not alter the levels of TNF-α, IL-1β or IL-1α significantly compared to the levels of these cytokines detected in cultures of either HSCs or mDCs alone. Neither cell type produced detectable levels of IFNγ (data not shown). HSCs produced IL-10 (Fig. 2B), which is known to suppress DC maturation (46), and its production was unchanged when mDCs were cultured with HSCs (Fig. 2B). Secretion of IL-6, an important regulator of DC differentiation (47) and liver DC activation (21), was increased significantly in co-cultures of liver mDCs and HSCs (Fig. 2B). Secretion of IL-6 by mDCs (Fig. 2) as well as its intracellular expression without or with HSC-conditioning (Fig. 4A lower panel) was relatively very low. Thus, HSCs are the primary source of secreted IL-6 in the co-culture. Cytokines typically associated with T cell differentiation/proliferation (i.e. IL-2, IL-4, IL-5, IL-12p70, IL-13, and IL-17) were not detected in supernatants from either mDCs or HSCs cultured alone or together (data not shown). These data show that interaction between HSCs and mDCs alters the cytokine milieu, with implications for the activation state or function of liver mDCs.
Up-regulation of CD54 on APCs correlates with their activation (48), and mDCs co-cultured with HSCs up-regulated CD54 (Supplemental Fig. 2). Therefore, we evaluated expression of conventional activation markers,-MHC class II (I-Ab), CD80, CD86 and B7-H1 on liver mDCs cultured with HSCs. As shown in Figs. 3A and 3B, co-culture of mDCs with HSCs increased the incidence of mDCs expressing CD80, CD86 and B7-H1, as well as the level of expression of these molecules determined by mean fluorescence intensity (MFI) (Fig. 3C). These effects of HSCs on mDCs could be titrated by altering the HSC:mDC ratio in the co-cultures (data not shown).
Since non-conventional pDCs account for approximately 20% of the total mouse liver DC population (49), the ability of HSCs to modulate the activation of liver pDCs was also tested. Liver pDCs cultured overnight with HSCs failed to up-regulate cell surface CD80 and CD86 expression (data not shown). However, co-culture of HSCs with splenic mDCs resulted in up-regulation of DC maturation markers, similar to the effect of HSCs on liver mDCs (data not shown). Collectively, these data show that HSCs selectively modulate mDCs.
Since HSCs and DCs associate in vitro and in vivo (Fig. 1) and others have reported a need for cell-cell contact in order for DCs to activate HSCs (12), we examined the ability of HSCs to promote the phenotypic maturation of mDCs as a result of their physical association. When cell-cell contact was prevented using a transwell system, the ability of HSCs to up-regulate the expression of CD80, CD86 and B7-H1 on mDCs was markedly inhibited (Figs. 3A, 3D).
HSCs secrete MCP and MIP1α (Fig. 2), which are potent DC chemoattractants (50). When co-cultured in the presence of MCP- or MIP1α-neutralizing Abs, the capacity of HSCs to promote CD80, CD86 and B7-H1 expression on liver mDCs was reduced, anti-MIP1α being modestly, though not statistically, more effective than anti-MCP (Fig. 3E). The influence of these chemokines was not additive or synergistic, as indicated by no further change in the expression of co-regulatory molecules when both anti-MCP and anti-MIP1α were added to the co-cultures. Collectively, these data suggest that the ability of HSCs to promote CD80, CD86 and B7-H1 expression on mDCs requires direct cell contact and that this contact may be mediated, in part, by HSC-secreted chemokines.
Unlike other hematopoietic cells, DCs are resistant to Fas/FasL-induced apoptosis (51, 52). Indeed, Fas/FasL interactions can induce DC maturation (51, 53), increase IL-1β secretion (53) and inhibit DC apoptosis via NFκB activation (53). Since HSCs express Fas (CD95) (39, 54), and direct contact is required for HSCs to promote mDC maturation (Figs. 3A, 3D), we hypothesized that Fas/FasL interaction might mediate the up-regulation of CD80, CD86 and B7-H1 on mDCs co-cultured with HSCs. Flow analysis of mDCs cultured overnight with HSCs revealed only modest up-regulation of surface FasL (CD95L) (Supplemental Fig. 3A), and addition of neutralizing anti-CD95L mAb at the start of HSC/mDC co-cultures caused no significant change in expression of co-regulatory molecules (Supplemental Fig. 3B).
Previous studies suggest that IL-6 impairs the differentiation of DCs (47), including liver DC maturation (21). On the other hand, IL-6, in concert with IL-1β and TNFα, can act as a potent activator of DCs (55). Since high levels of IL-6, TNFα and IL-1β were observed in HSC-mDC co-cultures (Fig. 2B), we investigated the role of IL-6 as either an inducer or inhibitor of HSC-mediated mDC maturation. To this end, we co-cultured mDCs with HSCs isolated from either WT (B6.129) or IL-6−/− mice, and found that HSCs were the primary source of secreted IL-6 (Fig. 4A upper panel). Intracellular staining confirmed that liver mDCs express very low level of IL-6, which was not altered by co-culture with HSCs (Fig. 4A, lower panel), and our previous work has demonstrated high level intracellular IL-6 expression by HSCs (39). Fewer mDCs conditioned with IL-6−/− HSCs up-regulated CD80, CD86 and B7-H1 (Fig. 4B). IL-6−/− HSCs induced a lower level of expression of these markers, when compared to that of WT HSCs (Figs. 4B, 4D). These data suggest that IL-6 plays a role in HSC-mediated maturation of liver mDCs. Intracellular signaling coupled to DC activation involves transcriptional regulation via STAT1, STAT3 and NFκB (56). IL-6 signals through STAT3, and STAT3 activation is reported to be critical for IL-6-driven B7-H1 up-regulation on DCs exposed to low levels of TLR4 stimulation (23). To delineate transcriptional events that may be involved in HSC-induced mDC activation, mDCs were pretreated with a STAT3 inhibitor prior to culture with HSCs. STAT3 inhibition markedly reduced the expression of CD80, CD86 and B7-H1 on WT HSC-conditioned mDCs (Figs. 4C, 4D) suggesting a requirement for STAT3 activation in HSC-induced mDC maturation.
Activated DCs are known to induce activation/proliferation of CD4+ and CD8+ T cells. Since HSCs up-regulated maturation markers (CD80, CD86 and B7-H1) on mDCs, we investigated the influence of HSC-conditioned mDCs on naive allogeneic T cell proliferation. Either liver mDCs or HSCs cultured alone were used as controls. HSCs alone failed to induce the proliferation of CD4+ or CD8+ T cells in CFSE-MLR, while liver mDCs induced the proliferation of both CD4+ and CD8+ T cells (Fig. 5A). Conditioning mDCs with HSCs strongly inhibited their ability to induce allogeneic CD4+ and CD8+ T cell proliferation when compared to control mDCs (Figs. 5A, 5C).
The expression of B7-H1, a potent co-inhibitory molecule (57) was up-regulated markedly on HSC-conditioned liver mDCs (Fig. 3A), and driven in part by HSC-derived IL-6 and STAT3 activation in mDCs (Fig. 4). We hypothesized that this mechanism might account for the diminished allostimulatory capacity of HSC-conditioned mDCs. To test this, mDCs conditioned by HSCs from WT or IL-6−/− mice were used as stimulators of naive allogeneic T cells. IL-6−/− HSC-conditioned mDCs were consistently better stimulators of CD8+ allogeneic T cells, but exhibited equivalent ability to stimulate CD4+ T cells compared to WT HSC-conditioned mDCs (data not shown). However, inhibition of STAT3 activation in liver mDCs prior to culture with HSCs restored their ability to stimulate both CD8+ and CD4+ T cell proliferation (Figs. 5B, 5C). Interestingly, inhibition of STAT-3 in mDCs cultured alone increased their ability to induce T cell proliferation, which was enhanced when mDCs were conditioned by HSCs.
Given that HSCs cause pronounced up-regulation of B7-H1 on liver mDCs, we postulated that conditioning B7-H1−/− mDCs with HSCs would enhance their ability to induce proliferation of T cells. However, B7-H1−/− liver mDCs, conditioned by HSCs, also exhibited reduced ability to stimulate proliferation of allogeneic T cells (Supplemental Fig. 4A), as seen with WT mDCs (Fig. 5A). Collectively, these data suggest that a distinct mechanism, driven by STAT3 activation in mDCs, is responsible for HSC-induced impaired ability of liver mDCs to stimulate T cell proliferation.
We have described that liver mDCs, matured phenotypically in an IL-6-dependent manner, are unable to induce robust proliferation of T cells. STAT3 regulates IDO transcription (24). Interestingly, in the absence of suppressor of cytokine signaling (SOCS)-3, IL-6 can activate DCs while causing simultaneous up-regulation of IDO activity, and thus induces an activated, yet tolerogenic DC phenotype (58, 59). We hypothesized that the apparent discordant effects of HSC conditioning on liver mDC phenotype and function might result from increases in STAT3-driven IDO activity. In support of this, expression of IDO, as determined by flow cytometry, was significantly (1.7 fold) higher in mDCs conditioned by HSCs (Figs. 6A, 6B). The HSC-induced increase in IDO was contact-dependent (data not shown) and blocked in mDCs treated with a STAT3 inhibitor prior to co-culture with HSCs (Figs. 6A, 6B).
SOCS3 has been shown to regulate ubiquitination-mediated proteosomal degradation of IDO in DCs (59). We found that expression of total SOCS3 in mDCs was significantly down-regulated by HSCs (Figs. 6C, 6D), suggesting that the increased IDO expression in mDCs is up-regulated by HSCs by inhibiting SOCS3. To determine if the reduced T cell allostimulatory function of HSC-conditioned mDCs was due to increased IDO activity, an IDO inhibitor (1-MT) was added to CFSE-dilution assays, together with either control or HSC-conditioned mDCs. IDO blockade enhanced the ability of control hepatic mDCs to stimulate proliferation of CD4+ and CD8+ allogeneic T cells (Fig 7A). Blocking IDO activity in HSC-conditioned mDCs restored their ability to induce proliferation of CD4+ T cells to the levels seen with control mDCs, and also enhanced the ability of HSC-conditioned mDCs to induce proliferation of CD8+ T cells (Fig. 7B). To further confirm that the suppressive effect of HSC-conditioned mDCs was indeed due to up-regulated IDO, we used mDCs isolated from IDO−/− mice. The ability of HSC-conditioned mDCs to suppress T cell proliferation was not apparent in IDO−/− mDCs (Fig. 7A, 7B). In addition, the allostimulatory potential of HSC-conditioned mDCs was markedly increased in a CFSE-MLR with excess (500 μM) L-tryptophan (60) (data not shown). This increase in allostimulatory capacity mirrors the increased expression of co-stimulatory molecules (Fig. 3) seen on HSC-conditioned mDCs, which may be masked by induction of IDO.
A similar effect was seen when B7-H1−/− hepatic mDCs conditioned with HSCs were used to stimulate allogeneic T cell proliferation in the presence of 1-MT (Supplemental Fig. 4). These data emphasize the importance of IDO over B7-H1 in HSC-induced impaired ability of mDCs to induce T cell proliferation, and support our hypothesis that HSC-induced STAT3 activation increases induction of IDO in mDCs.
Very few studies have evaluated the functional relationship between liver DCs and other liver-resident NPCs, despite their physical association in vivo, and the well-established importance of the hepatic environment in the maintenance of tolerance (1, 2). Interactions between liver macrophages (KCs) and DCs are the best-defined, with KCs reported to capture circulating DCs from the blood utilizing an N-acetylgalactosamine-specific C-type lectin-like receptor (61, 62). Moreover, MIP1α-secreting KCs recruit DCs to inflammatory sites (63), and KCs can initiate transcriptional changes that promote liver DC tolerogenicity by secreting high levels of IL-10 (64). Thus, the role of KCs in promoting tolerance may be prominent in conditions with intact endothelium. However, cold preservation- and reperfusion-induced injury to the liver disrupts the endothelial barrier (65). This damage, together with sluggish blood flow due to hepatic vasoconstriction, allows direct and protracted interactions between cells in the sinusoids (such as liver DCs) and perisinusoidal HSCs that also produce IL-10. From these pathophysiologic events, it is clear that HSCs, that are similar in number to KCs in hepatic NPC populations, may also contribute to hepatic immune tolerance.
Since HSCs make contact with almost all liver cell types via cytoplasmic extensions (30) and express immune regulatory molecules (32), their influence on cells of the innate and adaptive immune systems has attracted increasing attention (33, 38, 66). Initially, it was reported that IFN-γ-stimulated mouse HSCs could induce apoptosis of activated allogeneic CD4+ and CD8+ T cells (34). Subsequent studies have shown that HSCs can promote IL-2-dependent expansion of Tregs (35) and control bacterial infection by activating NKT cells via a lipid Ag-presenting mechanism (33). These studies portray HSCs as regulators of T cell activation within the liver. However, whether HSCs affect the immunobiology of the liver’s professional migratory APCs (DCs) has not been investigated.
In this report, we show for the first time that HSCs interact physically with conventional liver mDCs, enhancing their expression of cell surface co-regulatory molecules (CD80, CD86 and B7-H1). These effects are cell-cell contact dependent and require HSC-derived soluble mediators (MCP-1, MIP1α and IL-6). Other factors secreted by HSCs, including endothelin-1 and prostaglandins (67-69) have been reported to modulate DC activation (70-72), but in the current study, pharmacological inhibition of endothelin receptors did not affect HSC-induced liver mDC maturation (data not shown). Furthermore, addition of indomethacin only modestly reduced CD80 and CD86 expression on mDCs cultured with HSCs (data not shown). HSCs also express high levels of Fas (54) and induce modest expression of FasL on mDCs, but neutralizing FasL in mDC-HSC cultures had only a minor effect on maturation marker expression by liver mDCs. Likely, the up-regulation of these markers on liver mDCs is due, at least in part, to the combined effects of multiple pro-inflammatory cytokines (TNFα, IL-6 and IL-1β) secreted by HSCs in co-culture with mDCs (Fig. 2).
We found that reducing the levels of IL-6 available to liver mDCs inhibited their expression of CD80, CD86 and B7-H1. Several groups have shown that liver DCs have comparatively poor T cell allostimulatory capacity compared with splenic or blood DCs, which may reflect the influence of anti-inflammatory mediators (e.g. IL-10 and TGF-β) produced in the liver (2). Our results demonstrate that HSCs can regulate the maturation and allostimulatory function of liver mDCs. In fact, compared with untreated liver mDCs, those co-cultured with HSCs were even poorer stimulators of allogeneic CD4+ and CD8+ T cell proliferation, despite their increased CD80 and CD86 expression. Since several groups have shown that expression of B7-H1 is an important regulator used by liver NPCs to control the activation/proliferation of CD8+ T cells (17, 73),we postulated that the impaired T cell allostimulatory function conferred by HSCs might be due to marked HSC-induced up-regulation of B7-H1 on liver mDCs. However, our data show that the impaired T cell stimulatory activity of HSC-conditioned liver mDCs was independent of their B7-H1 expression.
Interestingly, tolerogenic/regulatory DCs may express high levels of CD80, CD86 and B7-H1, yet also produce IDO, a potent enzyme that suppresses T cell proliferation (74). Altering the intracellular adaptor protein, SOCS3 can change the signaling pattern of DCs, turning anti-inflammatory IL-6 into a DC-activating cytokine, but also inducing IDO and imparting a tolerogenic phenotype on the cells (58, 59). Recently we have identified a role for DNAX-activating protein of 12 kDa (DAP12), a molecule associated with IDO induction, in the regulation of liver mDC maturation (19). Interactions with HSCs may induce changes in these intracellular molecules (SOCS3 and DAP12) and alter IL-6 signaling to induce a tolerogenic DC phenotype. Although our results are consistent with the reported positive association between serum IL-6 concentration and IDO expression in a murine model of sepsis (75), exogenous IL-6 failed to up-regulate IDO expression in liver mDCs (results not shown). Furthermore, IL-10 also did not alter IDO expression in mDCs. We also considered a role for TGF-β, which is expressed at low levels in rat (76) and mouse (A. Dangi and C.R. Gandhi, unpublished observation) HSCs, in IDO increase in mDCs. However, incubation in the presence of anti-TGF-β Ab did not alter HSC-induced IDO expression in DCs (results not shown). These results indicate an alternate mechanism of HSC-induced IDO up-regulation in DCs. Nevertheless, considering the significant role of IDO induction in immune regulation in experimental hepatitis (77, 78), HSC-induced IDO in liver mDCs may be important in regulation of alloimmunity.
IL-6, which is produced by HSCs and further up-regulated in HSC-mDC coculture, is known to signal via STAT3 activation (21, 56), an important transcriptional pathway for maintaining DC immaturity, particularly in the liver (21, 22). However, we found that blocking STAT3 activation in liver mDCs reduced their CD80, CD86 and B7-H1 expression, suggesting that, in some circumstances, STAT3 can promote DC maturation. In support of this, we found previously that the majority of active STAT3 was expressed in CD80+ or CD86+ liver mDCs (19). Earlier studies have shown that, in a tumor-specific environment, myeloid APCs mature in a contact-dependent manner via STAT3 activation (79, 80). Interestingly, STAT3 activation was also associated with increased IDO expression in HSC-conditioned liver mDCs. IDO induction may occur directly, through STAT3 acetylation (24), or may be induced indirectly, by shifting NFκB signaling to a non-canonical pathway (81). Alternatively, STAT3 activation in mDCs may increase the expression of chemokine receptors involved in recruitment of DCs to HSCs, as has been reported for T cells (82). Although in our study, the effects of STAT3 were likely mediated through IL-6 or IL-10, neither cytokine caused IDO up-regulation in mDCs. These data suggest that contact-dependent interactions between HSCs and mDCs may be a mechanism of STAT3 activation and consequent IDO up-regulation in mDCs. Such direct contact is likely enhanced in transplantation due to cold ischemia/reperfusion-induced disruption of the endothelial barrier (65) and an important mechanism of graft tolerance.
While HSCs are quiescent under steady-state conditions, they become activated into a proliferating myofibroblast-like phenotype during liver injury. This process involves progressive loss of stored retinoids by HSCs, and their phenotypic differentiation, through transitionally-activated (α-sma expression and lower retinoid content) to fully-activated cells (high α-sma expression and very low or no retinoids). Activation of HSCs is associated with production of greater amounts of extracellular matrix (29, 30, 32) and synthesis and release of various cytokines and chemokines (31, 32, 83) that modulate immune cells. HSCs used in this study at 7 days of culture represent a transitionally-activated phenotype, that expresses the marker of rodent HSC activation (α-sma), but also contain significant residual retinoids (Supplemental Fig. 1). Such HSCs are found in the liver early after hepatic injury (84, 85), their activation being driven by free radicals and cytokines produced by resident macrophages and infiltrating neutrophils and other lymphocytes (32). Thus the modulation of liver mDC phenotype and function by transiently-activated HSCs that we have described herein may be an important mechanism that affects the course of liver transplantation.
In conclusion, our results further implicate HSCs as a cell type that may regulate hepatic immune reactivity. In addition to their direct inhibitory effect on naïve and activated T cells (34, 38, 39), and their capacity to expand naturally-occurring Tregs (39), HSCs undermine liver mDC function in a contact-dependent manner and promote their tolerogenic potential by increasing their IDO activity. These characteristics of HSCs may play an important role in hepatic immunobiology, ranging from transplant tolerance to fibrogenesis.
We thank Dr. Osamu Yoshida for his help in isolation/purification of liver dendritic cells, our colleagues Drs. David Geller, Noriko Murase and Jake Demetris for valuable discussion, and Ms. Miriam Freeman for administrative support.
This work was supported by National Institutes of Health (NIH) grant PO1AI81678 (AWT and CRG), a VA Merit Review Award (CRG), an American Society of Transplantation Basic Science Fellowship and an American Liver Foundation Roger Jenkins Fellowship (TLS), and by NIH T32 AI74490 (AWT).
Potential conflict of interest: The authors have no financial conflicts of interest. The content of this work does not represent the views of the Department of Veterans Affairs or the United States Government.