Potential conflict of interest: Nothing to report.
CD4+CD25+ regulatory T cells (Treg) are important mediators of peripheral immune tolerance; however, whether Treg participate also in hepatic immune tolerance is not clear. Therefore, we tested the potential of Treg to suppress stimulation of CD4+ T cells by liver sinusoidal endothelial cells (LSEC), Kupffer cells (KC), or hepatocytes. In the absence of Treg, all 3 types of liver cells could stimulate CD4+ T cell proliferation; in the presence of Treg, however, CD4+ T cell proliferation was suppressed. Interaction with KC even stimulated the expansion of the Treg population; LSEC or hepatocytes, in contrast, could not induce proliferation of Treg. Because liver inflammation can be induced by infection, we tested the potential of liver cells to modify Treg suppressor activity in the presence of microbial signals. In the presence of immune-stimulatory CpG-oligonucleotides, LSEC, KC, and hepatocytes could indeed overcome Treg-mediated suppression; in the presence of lipopolysaccharide (LPS), however, only KC and hepatocytes, but not LSEC, could overcome Treg suppressor activity. Hepatocytes from mice with deficient toll-like receptor-4 signaling failed to abrogate Treg suppression in response to LPS, indicating that overcoming Treg suppressor activity was indeed a response of the liver cell and not of the Treg. In conclusion, Treg can suppress CD4+ T cell stimulation by liver cells. However, in response to microbial signals, the liver cells can overcome the suppressive activity of Treg. Thus, liver cells may facilitate the transition from hepatic immune tolerance to hepatic inflammation by controlling Treg suppressor activity. (HEPATOLOGY 2005;42:193–199.)
Under physiological conditions, the liver is basically free of inflammation and the resident antigen-presenting cells (APC) of the liver are believed to actively participate in the maintenance of hepatic immune tolerance.1 Indeed, the most prevalent professional APC in liver, liver sinusoidal endothelial cells (LSEC) and Kupffer cells (KC),2, 3 have been shown to actively induce immune tolerance.2, 4, 5 In addition, hepatocytes, which can upregulate expression of class II molecules of the major histocompatibility complex (MHC II) during hepatitis6 and may then serve as APC for CD4+ T lymphocytes,7 also have been implicated in the induction of immune tolerance.8 Nevertheless, rapid infiltration of the liver with inflammatory cells is induced during infection. However, the mechanisms that mediate the transition from hepatic immune tolerance to hepatic inflammation are not clear.
CD4+CD25+ regulatory T cells (Treg) are important mediators of peripheral immune tolerance: Treg can suppress the activation of autoreactive CD4+ and CD8+ T cells9–11 and also control immune responses during infection.12, 13 Whether Treg are involved in the maintenance of hepatic immune tolerance is not clear; however, recent reports indicate that Treg indeed may suppress hepatic immunity to autoantigen14 or virus.15 Thus, it is likely that local suppressive Treg activity needs to be attenuated before liver inflammation can be induced.
It has been reported recently that splenic dendritic cells have the capacity to regulate the suppressive activity of Treg16: normally, dendritic cells stimulate Treg-mediated suppression; however, in response to microbial stimuli, like lipopolysaccharide (LPS) or immune-stimulatory oligonucleotide sequences that contain CpG motifs (CpG-ODN), dendritic cells overcome Treg-mediated suppression. Abrogation of the Treg suppressive activity seemed to be, at least in part, due to the toll-like receptor–mediated secretion of interleukin-6 by dendritic cells16 Thus, LPS or CpG-ODN can be considered danger signals, which prevent Treg-mediated suppression stimulated by dendritic cells.
The aim of the current study was to investigate the role of LSEC, KC, and hepatocytes in the stimulation and control of Treg. The data presented here indicate that LSEC, KC, or hepatocytes can stimulate the suppressive activity of Treg. However, in the presence of microbial stimuli, LSEC, KC, or hepatocytes could overcome Treg suppressive activity. Thus, by regulating Treg activity, liver APC may facilitate the transition from hepatic immune tolerance to hepatic inflammation.
All mice were generated, bred, and kept under specific pathogen-free conditions in the animal facilities at the Johannes Gutenberg-University, Mainz. Only male mice of BALB/c background were used for the experiments at the age of 8 to 12 weeks. BALB/c-LPSd mice, which express the LPSd-allele from the LPS-hyporesponsive strain C3H/HeJ,17 were provided by Chris Galanos (Max Planck Institut für Immunobiologie, Freiburg, Germany).
Isolation of Hepatocytes.
Hepatocytes were isolated from mouse livers as described.7 Briefly, after perfusion of the portal vein with 0.05% collagenase (NB8, Serva, Heidelberg, Germany), the livers were mechanically dissected and centrifuged (30g). To remove residual non-hepatocytes from the hepatocyte preparation, the cells were incubated with monoclonal antibodies from supernatants of the hybridoma 3.155, F4/80.1.15, B17 anti-CD13, RA3-3A1/6.1, M1/184.108.40.206.HL, and MAR18.5 and rabbit complement for 20 minutes at 37°C. Hepatocytes were resuspended in Dulbecco's modified Eagle medium (Invitrogen, Karlsruhe, Germany) supplemented with 10% fetal calf serum, 1% glutamine, 1% penicilline/streptomycin, and cultured in 96-well Primaria flat-bottom plates (Falcon, Becton Dickinson, Heidelberg, Germany) at a density of 2 × 104 or in collagen-R (Serva)–coated 24-well culture plates (Greiner, Solingen, Germany) at 1 × 105. Hepatocytes were cultured overnight, and experiments were performed on the next day.
Isolation of Kupffer Cells (KC) and Liver Sinusoidal Endothelial Cells (LSEC).
LSEC and KC were isolated as previously described.18 Briefly, livers were perfused through the portal vein with 0.05% collagenase A (Roche, Mannheim, Germany) in a calcium-free phosphate buffer. Livers were mechanically disrupted and incubated with 0.05% collagenase IV (Sigma, Taufkirchen, Germany) in Grey's balanced saline solution for 30 minutes at 37°C with constant rotation (240 rpm). The cell suspension was passed through a cell sieve (stainless steel 200 μmol/L) to remove debris. The fraction of nonparenchymal cells was recoverd by density centrifugation on a 30% metrizamide gradient (Nycomed, Oslo, Norway). Further separation of LSEC and KC was achieved by counterflow centrifugal elutriation (JE-6B rotor with standard elutriation chamber and J2-MC centrifuge; Beckman, München, Germany): Rotor speed was kept at 2,500 rpm, and cell fractions were collected by increasing counter flow (LSEC: 23 mL/min; KC: 56 mL/min). Elutriated cells were washed in phosphate-buffered saline and cultured in Dulbecco's modified Eagle medium (with 10% fetal calf serum; 1% glutamine; 1% penicillin/streptomycin) on 96-well Primaria flat-bottom plates (Falcon, Heidelberg, Germany) at 2 × 104 or on collagen-R–coated 24-well culture plates (Greiner Bio-One, Frickenhausen, Germany) at 1 × 105. LSEC and KC were kept in culture for 3 days before experiments were performed.
Preparation of T Cell Populations (CD4+ T Cells and CD25+ Treg).
T cells were isolated from spleens using magnetic cell separation (Miltenyi Biotech, Bergisch-Gladbach, Germany) according to the manufacture's instructions. CD25+ Treg cells were labeled by sequential incubation with biotin-conjugated anti-CD25 antibody (7D4bio, Pharmingen, Heidelberg, Germany), Streptavidin-PE (Dianova, Hamburg, Germany) and anti-PE MicroBeads (Miltenyi) and recovered by positive selection on LS columns (Miltenyi). CD4+ T cells were isolated from remaining spleen cells (without CD25+ T cells) by sequential labeling with biotin-conjugated anti-CD4 antibody (H129.19, Pharmingen) and strepavidin-MicroBeads (Miltenyi) and positive selection on LS columns. To exclude the influence of residual accessory cells in the T cell preparations, both cell populations (Treg and CD4+ T cells) were additionally depleted of CD8+ T cells (Dynabeads Mouse CD8, Dynal Biotech, Oslo, Norway), B cells (Dynabeads Mouse pan B [B220]) and macrophages (Dynabeads M-450 Epoxy-conjugated with anti–Mac-1α antibody (supernatant of hybridoma M1/220.127.116.11.2). Purity of resulting cell populations was determined by FACS analysis (FACScan, Becton Dickinson, Franklin Lakes, NJ) and was routinely above 98%.
Proliferation assays were performed in 96-well flat-bottom culture plates. T cells were added to isolated LSEC, KC, or hepatocytes (2 × 104/well) either seperately (1 × 105 CD4+ T cells or 1 × 105 Treg) or in co-culture (1 × 105 CD4+ T cells plus 1 × 105 Treg) and stimulated with soluble anti-CD3 antibody (1 μg/mL; Pharmingen). Alternatively, splenic dendritic cells (2 × 104/well), isolated as described,16 were used as APC for Treg expanded by 3 days of stimulation in KC culture. Because the T cells in this assay had been depleted of residual accessory cells (see above), the assayed proliferation of the T cells under these conditions depended entirely on co-stimulatory signals delivered by the liver APC. LPS (Escherichia coli 055:B5, Sigma) or CpG-ODN (5′-tcc atg acg ttc ctg atg ct-3′, phosphothionate, Roth, Karlsruhe, Germany) were added at the start of the cultures as indicated. In another set of experiments, recombinant murine interleukin-6 (50 ng/mL) was added.
After 4 days, 3H-thymidine was added to the culture, and after and additional 18 hours, 3H thymidine uptake was assessed by β-scintillation counting. The suppressed proliferation of CD4+ T cells (in the presence of Treg) minus proliferation of Treg alone (background) is given as percentage of the non-suppressed proliferation of CD4+ T cells (in the absence of CD25+ Treg):
When indicated, purified Treg were incubated with 2.5 μmol/L CFSE for 10 minutes at 37°C. The CFSE-labeled Treg (1 × 105) were activated by anti-CD3 antibody and unlabeled KC for 5 days. Proliferation of labeled Treg was assessed by FACS analysis at days 0, 3, and 4 of co-culture.
LPS (100 ng/mL) or CpG-ODN (300 pmol/mL) were added to cultures of LSEC, KC, or hepatocytes, each at a density of 1 × 105 in a 24-well culture plate. After 24 hours of incubation, culture supernatants were removed and the amount of secreted interleukin-6 was assessed by sandwich ELISA according to the manufacturer's instructions (Pharmingen).
The data are given as mean of triplicate wells ± standard deviation of the mean. Statistical significance of differences between data sets was tested by Student t test. A P value less than .05 was considered significant.
Liver APC Can Stimulate the Suppressive Activity of CD4+CD25+ T Regulatory Cells.
To study the capacity of liver APC to stimulate Treg suppressor activity, we determined the proliferative response of CD4+ T cells and Treg, both in separate culture and in co-culture, after stimulation by LSEC, KC, or hepatocytes and agonistic soluble anti-CD3 antibody. Under these conditions, stimulation of both CD4+ T cells and Treg depended entirely on the co-stimulatory signals delivered by the hepatic APC. Indeed, T cell proliferation was undetectable in the absence of APC (not shown). Proliferation counts of APC alone were negligible; therefore, no pre-treatment to inhibit proliferation of APC was necessary. We found that LSEC or KC as well as hepatocytes induced the proliferation of CD4+ T cells in the absence of Tregs (Fig. 1, white bars). Hepatocytes or LSEC did not induce the proliferation of Treg; KC, in contrast, seemed to induce Treg proliferation (Fig. 1, gray bars). When both CD4+ T cells and Treg, co-cultured at a 1:1 ratio, were stimulated by LSEC, KC, or hepatocytes, T cell proliferation was suppressed (Fig. 1, black bars): LSEC, KC, and hepatocytes induced Treg-mediated suppression to below 5% of CD4+ T cell proliferation in the absence of Tregs. These findings indicate that liver APC can stimulate the suppressive activity of Treg.
KC Induce the Proliferation of Tregs.
To confirm the finding that KC have the capacity to induce proliferation of Treg, we stimulated CFSE-labeled Treg with unlabeled KC and anti-CD3 antibody and followed Treg expansion by FACS-analysis of CFSE-labeled cells. We found that a fraction of the Treg population underwent up to 2 divisions within 4 days of stimulation by KC (Fig. 2A), whereas another fraction did not proliferate. However, after 4 days of stimulation on KC, a population of dead cells, marked by low CFSE intensity, became apparent. Cell death of this population was confirmed by propidium iodide staining (not shown). To confirm that the surviving proliferated CD25+ T cells were indeed Treg and not activated conventional CD4+ T cells, the suppressor activity of the expanded cells was tested in an assay of CD4+ T cell stimulation induced by splenic dendritic cells (Fig. 2B). We found that the expanded CD25+ T cell population retained suppressive activity, which was, however, lower than the suppression by freshly isolated CD25+ T cells (suppression of CD4 T cell proliferation to 40% by expanded Tregs vs. 15% by fresh Tregs). Nevertheless, KC appear to be unique among the liver cells in their potential to induce the proliferation of Treg.
Liver APC, in Response to Microbial Stimuli, Can Overcome Treg Suppressor Activity.
Because it was reported that splenic dendritic cells, after stimulation with microbial signals, can overcome the suppressive activity of Treg,16 we tested whether liver APC incubated with LPS (100 ng/mL) or CpG-ODN (300 pmol/mL), like splenic dendritic cells, may control Treg suppressor activity. The capacity of hepatocytes to stimulate CD4+ T cells was not significantly modulated by addition of LPS or CpG-ODN (Fig. 3A). However, hepatocytes incubated with LPS or CpG-ODN strongly abrogated the suppressive activity of Treg (Fig. 3A): T cell proliferation in the presence of LPS or CpG-ODN was restored from below 5% (no microbial stimulus) to more than 50% of unsuppressed proliferation.
When LSEC were used as APC, CD4+ T cell proliferation was enhanced in the presence of CpG-ODN but not of LPS (Fig. 3B). LSEC incubated with CpG-ODN could partially overcome Treg-mediated suppression, from below 5% to some 20% of the unsuppressed proliferation (Fig. 3B). However, LSEC incubated with LPS (Fig. 3B) were not proficient to overcome suppression.
When KC were used as APC, the proliferation of Treg was induced in both the presence and absence of LPS or CpG-ODN (Fig. 3C). KC, like hepatocytes, could overcome Treg-mediated suppression in response to LPS or CpG-ODN (Fig. 3C).
Thus, liver APC, in response to microbial stimuli, were able to control the suppressive activity of Treg. CpG-ODN induced abrogation of Treg activity by KC, hepatocytes, and to a lesser extent by LSEC; LPS, in contrast, did induce abrogation of Treg activity by hepatocytes and KC, but not by LSEC.
The Toll-like Receptor Pathway of Hepatocytes Is Required to Overcome the Suppressive Activity of Treg.
LPS and CpG-ODN signal through ligation of specific cellular receptors, toll-like receptor (TLR)-4 for LPS and TLR-9 for CpG-ODN. To confirm that LPS- or CPG-ODN–mediated control of the Treg suppressive activity required activation of the toll pathway in hepatocytes, rather than in Treg, we used hepatocytes from BALB/c-LPSd mice, which express the LPSd-allele responsible for deficient TLR-4 signaling in C3H/HeJ mice.17 BALB/c-LPSd hepatocytes were co-cultured with CD4+ T cells and Treg of TLR-signaling proficient BALB/c wild-type mice, and the suppressive activity was determined in response to LPS; hepatocytes from TLR-signaling proficient BALB/c wild-type mice were used as control. As shown in Fig. 4, hepatocytes with deficient TLR-4 signaling (white bars), after incubation with LPS, could not overcome the suppressive activity of Treg; wild-type hepatocytes with proficient TLR-4 signaling (black bars), in contrast, partially restored T cell proliferation after LPS incubation. Addition of CpG-ODN to both TLR-4 signaling proficient or deficient hepatocytes also restored CD4 T cell proliferation. Therefore, abrogation of the Treg suppressive activity required the activation of the toll-like receptor pathway in liver cells, not in Treg.
The Role of Interleukin-6 in Control of Treg Suppressor Activity.
The mechanism by which dendritic cells seem to control Treg suppressor activity has been reported to involve, at least in part, secretion of interleukin-6 from dendritic cells.16 To learn whether downregulation of Treg suppressor activity in liver may likewise involve interleukin-6, we determined the secretion of interleukin-6, spontaneous or LPS- or CpG-ODN induced, by LSEC, KC, and hepatocytes (Fig. 5). KC spontaneously secreted interleukin-6 at considerable amounts (134 pg/mL; white bars); spontaneous interleukin-6 secretion by LSEC or hepatocytes was negligible (9 pg/mL and 46 pg/mL; white bars). LPS stimulation (gray bars) induced upregulation of interleukin-6 secretion by hepatocytes (192 pg/mL), LSEC (228 pg/mL), and strong upregulation by KC (1,594 pg/mL); CpG-ODN stimulation (black bars) induced only a minor up-regulation of interleukin-6 secretion in hepatocytes (70 pg/mL) or LSEC (30 pg/mL), and a moderate up-regulation of interleukin-6 secretion in KC (372 pg/mL). Thus, the potential of liver APC to secrete interleukin-6 is compatible with a role for interleukin-6 in controlling Treg suppressor activity. However, the levels of interleukin-6 secreted by liver APC did not correspond to their potential to control Treg suppressive activity (Fig. 3); for example, LSEC secreted much more interleukin-6 after LPS stimulation (228 pg/mL) than after CpG-ODN-stimulation (30 pg/mL). Nevertheless, LPS-stimulated LSEC, in contrast to CpG-ODN–stimulated LSEC, could not abrogate Treg suppressive activity (Fig. 3B).
Therefore, we tested whether addition of recombinant interleukin-6 to co-cultures of hepatocytes, CD4+ T cells, and Treg was sufficient to overcome Treg-suppressive activity in the absence of microbial signals. We found that Treg suppressive activity induced by hepatocytes (suppression to 5.4% of unsuppressed CD4+ T cell proliferation; Fig. 6A-B) was not altered by the presence of interleukin-6 (25-75 ng/mL) (suppression to 5 % of unsuppressed CD4+ T cell proliferation). Therefore, the relevance of interleukin-6, secreted by liver cells, as a mediator of Treg suppressor activity seems questionable.
The mechanisms that regulate the transition from hepatic immune tolerance to hepatic inflammation are not clear. Because Treg play an important role in the maintenance of immune tolerance in most, if not all, peripheral organs, it was conceivable that Treg also may be involved in the regulation of hepatic inflammatory activity. Functional testing of Treg suppressor activity is important, because no reliable phenotypic markers allow the distinction of Tregs from activated conventional CD4 T cells. Up to now, in vitro studies exploring the suppressive function of Treg used non-hepatic APC, mainly dendritic cells or irradiated spleen cells. Here we investigated the ability of liver APC to induce and control suppressive activity of Treg. KC and LSEC are the major populations of APC in the liver that express MHC class II constitutively and thus can present antigen to CD4+ T cells and Treg. Indeed, we found that KC and LSEC could stimulate Treg to suppress the proliferation of CD4+ T cells (Fig. 1). In inflammatory states of the liver, hepatocytes can upregulate MHC class II6 and consequently gain the ability to interact with CD4+ T cells and Treg too. In addition to LSEC and KC, we found that hepatocytes also induce suppressive activity of Treg (Fig. 1). Therefore, Treg stimulated by liver APC may be involved in the maintenance of hepatic immune tolerance. Our findings are compatible with recent reports suggesting a potential role for Treg in hepatic immune tolerance.14, 15 The frequency of intrahepatic CD25+ T cells in humans has been reported to be approximately 1% to 2%15; we find compatible frequencies of 2% to 4% of the intrahepatic lymphocytes and 15% to 20% of the intrahepatic CD4+ T cells in mouse (not shown). Thus, the numbers of intrahepatic CD25+ T cells seem sufficient to regulate hepatic immune responses.
Interestingly, Treg proliferated after being stimulated by KC, but not by LSEC or hepatocytes (Figs. 1 and 2). So far, dendritic cells, notably bone marrow–derived dendritic cells, have been observed to stimulate Treg expansion most efficiently.19 Our finding of a weak Treg proliferation induced by KC, being liver macrophages, is compatible with the observation that peritoneal macrophages are weak stimulators of Treg expansion.19 Dendritic cell–expanded Treg seem to have full suppressive potential.20 We find the CD25+ T cells, which had been expanded by KC, also to retain suppressive activity (Fig. 2B), although suppression was somewhat lower than that of freshly isolated Treg. The potential of KC to propagate Treg at least in vitro seems to be unique among liver cells; whether KC-induced Treg proliferation is of relevance in vivo is not clear.
However, to protect the host against invading pathogens, one must abgrogate immune suppression by Treg; Treg inhibit not only autoimmunity, but also immunity to pathogens.12, 13 Recently, it has been shown that splenic dendritic cells, in response to microbial stimuli, acquire the ability to overcome Treg-mediated suppression16: LPS induced the secretion of interleukin-6 by splenic dendritic cells; secreted interleukin-6 seemed to be necessary but not sufficient for abrogation of Treg suppressive activity. In contrast to other organs, the presence of gut-derived bacterial compounds, such as LPS, in the hepatic microenvironment and portal blood is physiological. Thus, LSEC and KC are constantly exposed to low amounts of LPS in the liver sinusoids; hepatocytes seem to have less access to physiological LPS, because most LPS is sequestered from portal blood by the LSEC and KC barrier. However, the presence of high amounts of LPS or CpG-ODN sequence motifs as used in our assays are not physiological and may be perceived by liver APC as the signature of ongoing infection. Indeed, we found that Treg activity could be controlled by liver APC after stimulation with CpG-ODN (Fig. 3); intraperitoneal application of CpG-ODN to mice did not change the intrahepatic frequency of Treg in vivo (not shown). Thus, in vivo, Treg activity in response to infection does not seem to be regulated by a change in Treg frequency, but rather by a change in function. Administration of CpG-ODN to mice can unmask otherwise controlled CD8+ T cell–mediated autoimmunity against hepatocytes, which was accounted for by upregulation of costimulatory molecules on liver APC induced by CpG-ODN.21 The capacity of liver APC to abrogate Treg-mediated suppression in response to CpG-ODN may represent an additional mechanism of inducing hepatic inflammation.
Interestingly, only KC or hepatocytes could overcome Treg suppressive activity in the presence of LPS, whereas LSEC did not abrogate Treg-mediated suppression in response to LPS (Fig. 3). The lack of Treg control by LPS-stimulated LSEC could not be accounted for by insensitivity to LPS, because LSEC did respond to LPS stimulation with a moderate upregulation of interleukin-6 secretion (Fig. 5).
Interleukin-6, secreted by APC in response to microbial stimuli, has been reported to be involved in overcoming Treg suppressor activity.16 LSEC, KC, and hepatocytes can secrete interleukin-6 in response to microbial stimuli (Fig. 5). However, the findings that the amounts of interleukin-6 secreted by liver APC did not correspond to their ability to control Treg suppressor activity, and that exogenous addition of interleukin-6 did not increase the capacity of hepatocytes to control Treg activity (Fig. 6), indicate that interleukin-6 is not of central importance for overcoming Treg suppressor activity. It rather appears that interleukin-6 may amplify the effects of another factor, which has not yet been identified.
Which molecules mediate Treg control by liver APC remains to be determined. Nevertheless, the toll-like receptor pathway of hepatocytes seemed to be required to overcome Treg activity: hepatocytes stimulated with LPS were only able to control Treg suppressive activity, if the hepatocytes had functional TLR-4 signaling (Fig. 4). Thus, the perception of microbial signatures and the decision to block the suppressive activity of Treg is at the level of the liver APC, rather than at the level of the Treg.
In conclusion, KC and LSEC can stimulate the suppressive activity of Treg under physiological conditions. This finding may explain the maintenance of hepatic immune tolerance in the absence of danger signals. In the presence of microbial signals, however, KC and LSEC can overcome Treg-mediated suppression and promote proliferation of CD4+ T cells. In addition, MHC class II–expressing hepatocytes in the inflamed liver also overcome Treg suppressive activity in response to microbial signals and thus may contribute to inflammatory immune responses. By controlling Treg suppressor activity, liver cells may thus facilitate the transition from hepatic immune tolerance to hepatic inflammation.