Tolerogenic maturation of liver sinusoidal endothelial cells promotes B7-homolog 1-dependent CD8+ T cell tolerance

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Liver sinusoidal endothelial cells (LSEC) are unique organ-resident antigen-presenting cells capable of cross-presentation and subsequent tolerization of naïve CD8+ T cells. We investigated the molecular mechanisms underlying this tolerance induction in naive CD8+ T cells. MHC class I–restricted antigen presentation by LSEC led to initial stimulation of naïve CD8+ T cells, which up-regulated CD69, CD25, CD44, and programmed death (PD)-1 and proliferated similar to dendritic cell (DC)–activated CD8+ T cells. Importantly, cognate interaction with naïve CD8+ T cells triggered increased expression of co-inhibitory B7-H1 but not co-stimulatory CD80/86 molecules exclusively on LSEC but not DC. This matured phenotype of B7-H1high CD80/86low was critical for induction of CD8+ T cell tolerance by LSEC: B7-H1–deficient LSEC, that failed to interact with PD-1 on stimulated T cells, were incapable of inducing CD8+ T cell tolerance. Moreover, increased costimulation via CD28 interfered with tolerance induction, indicating that the noninducible low expression levels of CD80/86 on LSEC supported B7-H1–dependent tolerance induction. LSEC-tolerized CD8+ T cells had a distinctive phenotype from naïve and activated T cells with CD25low, CD44high, CD62Lhigh. They also expressed the homeostatic cytokine receptors CD127, CD122, and high levels of Bcl-2, indicating survival rather than deletion of tolerant CD8+ T cells. On adoptive transfer into congenic animals, tolerized CD8+ T cells failed to show specific cytotoxicity in vivo. Conclusion: Cognate interaction of LSEC with naïve CD8+ T cells elicits a unique tolerogenic maturation of LSEC and permissiveness of T cells for tolerogenic signals, demonstrating that LSEC-induced tolerance is an active and dynamic process. (HEPATOLOGY 2007.)

Peripheral immune tolerance not only prevents development of autoimmunity by controlling autoreactive T cells that escaped thymic elimination but also limits T cell responses during infections with microorganisms.1 The mechanisms of peripheral immune tolerance also control development of hepatic immunity and may play an important role in elimination or persistence of viral infection of the liver.2 The mechanisms underlying peripheral CD8+ T cell tolerance include inhibition by regulatory T cells, cell intrinsic regulation of T cell function, or a combination of both.3, 4 Antigen-presenting cells (APC), in particular dendritic cells (DC), are instrumental for induction of peripheral tolerance.1, 5, 6 Clonal deletion or functional inactivation of T cells can be established through members of the B7/CD28 superfamily, such as PD-1 and CLTA-4, that transmit inhibitory signals.7–10 Accordingly, B7-H1 and B7-DC, the ligands for PD-1, inhibit activated T cells,11, 12 and stimulate naïve T cells.13, 14 B7-H1 is widely expressed, whereas B7-DC expression is restricted to DC and macrophages.10 The central role of B7-H1 is the regulation of antigen-specific immune responses, that is, its expression on immune and parenchymal cells determines induction of immune tolerance in the pancreas,15 and B7-H1 on nonparenchymal liver cells contributes to deletion of activated T cells preventing hepatic autoimmunity.16, 17

The presence of different DC-subtypes in different organ-specific microenvironments complicates DC-based immune regulation.18 Appropriate stimulation of DC by proinflammatory mediators or Toll-like receptor ligands induces functional maturation, characterized by increased expression of the co-stimulatory molecules CD80/86 or interleukin-12 (IL-12) leading to T cell immunity. However, incomplete or alternative activation of DC or macrophages via interleukin-10 (IL-10), transforming growth factor beta, or corticosteroids leads to development of a semi-mature state that promotes T cell tolerance.6, 19 Moreover, specialized organ-resident APC also contribute to peripheral immune tolerance, particularly in the liver.2, 20 We reported that liver sinusoidal endothelial cells (LSEC) are a unique population of organ-resident APC: LSEC combine scavenger activity with the capacity to (cross)present exogenous antigens on both major histocompatibility complex (MHC) II and MHC I molecules to CD4+ or CD8+ T cells, respectively.21, 22 LSEC initially stimulate naïve T cells but fail to support differentiation into effector cells, and induce T cell tolerance21, 22 toward antigens derived from the gastrointestinal tract23 and from apoptotic cells.24

Here, we report that LSEC undergo tolerogenic maturation, independent of exogenously added mediators, that necessitates cognate interaction with naive CD8+ T cells. This interaction promotes licensing of LSEC for delivery of B7-H1–dependent co-inhibitory signals and of LSEC-stimulated CD8+ T cells for receiving tolerogenic signals via programmed death 1 (PD-1). Our results demonstrate that induction of T cell tolerance by LSEC is an active process dependent on the dynamic balance between coinhibition and costimulation.

Abbreviations

7-AAD, 7-actinomycinD; APC, antigen-presenting cells; B7-H1, B7-homolog 1; BMDC, Bone marrow-derived dendritic cell; CFSE, carboxy-fluorescein-succinimidyl-ester; DC, dendritic cells; ELISA, enzyme-linked immunosorbent assay; FBS, fetal bovine serum; IFN-γ, interferon gamma; IL, interleukin; LSEC, liver sinusoidal endothelial cells; MHC, major histocompatibility complex; OVA, ovalbumin; PD-1, programmed death-1; SEM, standard error of the mean; TCR, T cell receptor.

Materials and Methods

Mice.

C57BL/6 mice were obtained from Elevage Janvier (France). OT-I, DesTCR, CBA, and B7-H1−/− mice were bred in the central animal facility of the university hospital Bonn. B7-H1−/− mice were a gift from Dr. Lieping Chen. In vivo experiments were approved by the Animal Care Commission of Nord-Rhein-Westfalia.

Flow Cytometry.

Antibodies were purchased from BD Bioscience (Heidelberg, Germany) or eBioscience (San Diego, CA). 1 × 103 to 5 × 105 cells were stained with saturating concentrations of antibodies plus 10 μg/mL Fc block (clone 2.4G2) in fluorescence-activated cell sorting buffer [phosphate-buffered saline/1% bovine serum albumin/0.02%NaAz]. Acquisition and analysis was conducted on a FACSCalibur/CantoII (BD Bioscience) and FlowJo sofware (Tree Star Inc, Ashland, OR), respectively. To exclude dead cells, 7-AAD (Invitrogen, Paisley, UK) was added at a final concentration of 4 μg/mL.

Isolation and Culture of APC and CD8+ T Cells.

LSEC isolation livers were perfused with a 0.05% collagenase solution (Sigma-Aldrich, Munich, Germany), mechanically disrupted, and digested for 20 minutes at 37°C in Gey's balanced salt solution (GBSS)/collagenase, then filtered through a steel mesh and washed twice in GBSS. Nonparynchymal cells were separated by a 30% Nycodenz gradient. Afterward, LSEC were isolated by AutoMacs and ME9F1 beads (Miltenyi, Bergisch-Gladbach, Germany) according to manufacturers' instructions. Cells were seeded onto collagen-coated 24-well plates in Dulbecco's modified Eagle's medium (4500 mg/mL glucose) (Gibco) with 8% fetal bovine serum (FBS), 50 μM mercapto-ethanol, glutamine, and antibiotics. Cells adhered for 24 hours were washed and used 48 to 72 hours after preparation. Spleens were digested with a 0.05% collagenase solution for 20 minutes at 37°C. Splenic DC were isolated by AutoMacs using CD11c beads; CD8+ T cells, using CD8 beads. OT-I mice were depleted of natural killer T cells by injection of 200 μg anti-NK1.1 antibody (clone PK136) 3 days before T cell isolation. LSEC purity was checked by ME9F1-Alexa647 staining and flow cytometry. LSEC were routinely greater than 98% pure. CD8+ T cells were greater than 95% pure, and CD11c+ DC were greater than 80% pure. Bone marrow-derived dendritic cell (BMDC) were generated from C57BL/6 bone marrow cells by culture in granulocyte-macrophage colony-stimulating factor–containing medium for 6 days. On day 7, CD11c+ BMDC were purified by AutoMacs using CD11c beads. Purified naïve CD8+ T cells were seeded at 0.7 to 1 × 106 cells/well onto LSEC or cocultured with 50% CD11c+ DC. In OT-1 T cell cultures, 100 μg/mL ovalbumin (OVA) protein was added. Anti–B7-H1 (MIH5) and rat immunoglobulin G2a isotype control antibodies were added at 40 μg/mL. Anti-CD28 (37.51), anti–PD-1, hamster immunoglobulin G isotype control antibodies were added at 10 μg/mL. All T cells were cultured in Roswell Park Memorial Institute (RPMI) supplemented with 8% FBS, 50 μM mercapto-ethanol, glutamine, and antibiotics.

Apoptosis Detection in CD8+ T Cells.

Naive DesTCR CD8+ T cells were cocultured with LSEC or DC, and at the indicated time caspase 3/7 activity was determined with an apoptosis detection kit (Immunochemistry Technologies, Bloomington, MN). In brief, cells were incubated with the active caspase 3/7-binding FAM-DEVD-FMK-peptide for 30 minutes at 37°C. Cells were counterstained with CD8α-alexa647 and incubated with 4 μg/mL 7-actinomycinD (7-AAD), before acquiring the samples by flow cytometry.

T Cell Restimulation In Vitro.

After 5 days of coculture with LSEC or DC, viable T cells were harvested by density gradient centrifugation. Cells were washed 3 times and 1 × 103 cells/well were seeded into 96-well plates, coated with 10 μg/mL anti-CD3 (145.2C11) or with anti-CD3 and anti-CD28 (31.51, 10 μg/mL). Alternatively, 1 × 103 T cells were incubated with 5 × 102 CD11c+ splenic DC in the presence or absence of 100 ng/mL lipopolysaccharide. After overnight incubation, IL-2 and interferon (IFN) concentration in culture supernatants were assessed by enzyme-linked immunosorbent assay (ELISA).

In Vitro Kill Assay.

To test the cytotoxic activity of OT-1 T cells, 5 × 106 RMA cells were either loaded with 1 M SIINFEKL-peptide or left untreated for 30 minutes in phosphate-buffered saline. Subsequently, SIINFEKL-peptide loaded cells were labeled with 1 μM carboxy-fluorescein-succinimidyl-ester (CFSE) and control cells with 0.1 μM CFSE for 10 minutes at 37°C. FBS was added to a final concentration of 5%, and cells were washed 2 times in RPMI/8% FBS. Cytotoxic activity of DesTCR CD8+ T cells was tested using H-2Kb RMA cells (CFSEhigh) and RMA-s cells (lacking surface H-2Kb) (CFSElow). CFSEhigh and CFSElow labeled cells were mixed at a ratio of 1:1, and 8 × 103 mixed target cells were incubated alone or with 1 × 105 T cells (effector-to-target ratio = 25). After 4 to 6 hours, the ratio of the surviving CFSEhigh and CFSElow cell populations were assessed by flow cytometry: In vitro kill was calculated as follows: % specific kill = 100 [100* (CFSEhigh / CFSElow)probe/(CFSEhigh / CFSElow)control].

In Vivo Kill Assay.

At day −2, 5 × 105 to 1 × 106in vitro generated activated or tolerant CD8+ T cells were injected intravenously into C57BL/6 mice. At day −1, 5 × 105 SIINFEKL-loaded CD11c+ BMDC were injected intravenously. At day 0, B6 spleen cells were isolated and either loaded with 1 μM SIINFEKL peptide or left untreated for 30 minutes in phosphate-buffered saline. SIINFEKL-peptide loaded spleen cells were labeled with 1 μM CFSE and untreated cells with 0.1 μM CFSE for 10 minutes at 37°C. CFSEhigh and CFSElow labeled spleen cells were mixed at a ratio of 1:1, and 2 × 107 total cells were injected intravenously. B6 mice that did not receive T cells or BMDC served as controls. After 16 to 18 hours, the spleen, inguinal lymph node, and liver lymphocytes were harvested. Liver lymphocytes were isolated by a 40%/80% Percoll gradient (Amersham, GE Healthcare, Munich, Germany). The ratio of the surviving CFSEhigh and CFSElow spleen cells was assessed by flow cytometry: In vivo kill was calculated as follows: % specific kill: 100 − [100*(CFSEhigh/CFSElow)exp mice/(CFSEhigh/CFSElow)control mice].

Statistical Analysis.

The Student t test was used for analysis. Data are depicted as the mean ± standard error of the mean (SEM) and P-values <0.05 were considered significant: *P < 0.05, **P < 0.01, ***P < 0.001.

Results

Induction of CD8+ T Cell Tolerance by LSEC Is Preceded by a Phase of Initial Stimulation.

We have previously shown that LSEC induce CD8+ T cell tolerance in vitro and in vivo.21, 23, 24 Here we investigated the molecular mechanisms underlying stimulation and tolerance induction in naïve CD8 T cells by antigen-presenting LSEC. Antigen-presenting LSEC initially stimulated T cells to increase expression of CD25, CD44, and PD-1 and to down-regulate CD62L within 24 hours (Fig. 1A), indistinguishable from DC stimulation of T cells. Proliferation and expansion of CD8+ T cells during the first days of stimulation followed, similar to T cell stimulation by DC (Fig. 1B and C). DC-stimulated T cells continued to expand after day 3. However, LSEC-stimulated T cells did not expand at days 4/5 after priming. Overall, LSEC-stimulated T cells increased more than 2-fold (Fig. 1C). This difference in expansion was not attributable to increased apoptosis in LSEC-stimulated T cells, as we found similar amounts of apoptotic T cells in LSEC and DC cocultures at all times tested (Fig. 1D). Surviving T cells expressed high levels of Bcl-2 (Fig. 1E), indicating that, unlike deletional CD8+ T cell tolerance induced by immature DC,25 LSEC do not cause deletion of naïve CD8+ T cells.

Figure 1.

Naive CD8+ T cells are stimulated during antigen-specific interaction with LSEC. (A) Naive OVA-specific OT-1 CD8+ T cells recognizing H2KbSIINFEKL were co-cultured with LSEC or DC in the presence (black, open histograms) or absence (gray, filled histograms) of 100 μg/mL OVA protein. After 24 hours, cells were harvested and stained for CD25, CD44, CD62L, and PD-1. Data shown were gated on 7-AAD negative, CD8-positive cells. (B) Naive CFSE-labeled DesTCR CD8+ T cells were co-cultured with C57Bl/6 LSEC or DC and proliferation was assessed after 72 hours compared with naive DesTCR CD8+ T cells cultured without APC. Similar results have been obtained with OT-1 T cells (data not shown). (C) Naive CFSE-labeled DesTCR T cells were cocultured with LSEC or DC, and expansion and viability of CD8+ T cells was measured from day 1 to 5 by flow cytometry. Cells were harvested at the indicated time-points, stained with 7-AAD to exclude dead cells, and samples were spiked with a set amount of calibration beads. Total CD8+ T cells were calculated as follows: (beadsTotal * cellsSample)/ beadsSample. (D) Naïve DesTCR T cells were cocultured with LSEC or DC. At days 1 through 5, cells were stained with FAM-DEVD-FMK peptide, CD8α, and 7-AAD. Dot plots depict representative day 4 data and CD8α gated cells. Numbers represent the percentage of FAM-DEVD-FMK+ cells in all CD8α+ cells. (E) Naive CFSE-labeled DesTCR T cells were cocultured with LSEC (black line) or DC (dashed line). At day 5, T cells were harvested and stained intracellularly for Bcl-2. Isotype control (filled gray histogram). Values (C) are depicted ±SEM. Statistical significance was calculated using a Student t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

To demonstrate that stimulation of naive CD8+ T cells strictly depended on T cell receptor (TCR)-MHC-I recognition, we co-incubated LSEC with 2 populations of naive CD8+ T cells bearing different transgenic TCRs, so that LSEC presented cognate antigen only to DesTCR T cells but not to OT-1 T cells. Only DesTCR but not OT-1 cells increased cell size, indicating T cell stimulation (Fig. 2A). Furthermore, only DesTCR transgenic T cells up-regulated CD25, CD69, and PD-1 and down-regulated CD62L (Fig. 2B). In contrast, OT-1 cells showed no phenotypic characteristics of stimulation in the absence of antigen (Fig. 2A,B), but were stimulated on addition of OVA to LSEC cultures (data not shown). In summary, these data indicate that the tolerization of CD8+ T cells by LSEC is strictly antigen dependent and is preceded by stimulation, proliferation, and clonal expansion of T cells.

Figure 2.

Antigen-specific cognate interaction is required for LSEC-mediated CD8+ T cell stimulation. Naïve DesTCR CD8+ T cells were mixed with naïve OT-1 CD8+ T cells at a ratio 1:1 and co-cultured on C57Bl/6-LSEC. After 24 hours, total CD8+ T cells were harvested and (A) OT-1 (I) or DesTCR (II) T cell size was determined by forward scatter. Alternatively (B), the expression level of the activation markers CD25, CD44, CD62L, and PD-1 on either OVA-specific OT-1 (gray dashed line) or DesTCR (black solid line) CD8+ T cells was determined. Filled gray histograms; isotype control staining.

Antigen-Specific Interaction with CD8+T Cells Increases Expression of the Co-inhibitory B7-H1 Molecule on LSEC But Not on DC.

We further analyzed the nature of LSEC stimulation of naive CD8+T cells following antigen recognition. LSEC constitutively expressed the adhesion molecules CD54 and CD106 as well as the co-signaling molecules CD80/8626 and B7-H1 (Fig. 3A and 4A). Twenty-four hours after incubation of LSEC with naïve CD8+T cells, there was a 2-fold to 4-fold increase in expression CD54 and CD106 when CD8+ T cells recognized antigen presented by LSEC (Fig. 3A). Antigen recognition by T cells, however, did not increase expression levels of CD80 and CD86 on LSEC (Fig. 3A). In contrast, B7-H1 expression was more than 10-fold upregulated on antigen-presenting LSEC (Fig. 3A). A comparable increase in expression of these molecules on DC was not observed during antigen-specific interaction with T cells (Fig. 3A). Collectively, these data indicate that tolerization of naive CD8+ T cells by LSEC is an active process in which LSEC dynamically alter their phenotype to increase adhesion and co-inhibitory signaling.

Figure 3.

LSEC-derived co-inhibitory B7-H1 signaling is required for tolerization of naive CD8+ T cells. OT-1 CD8+ T cells were co-cultured with C57Bl/6 LSEC or DC in the presence or absence of 100 μg/mL OVA protein. (A) After 24 hours, cells were harvested and ME9F1+ LSEC and CD11c+ DC were stained for CD80, CD86, B7-H1, CD106, and CD54. X-fold increase was calculated as follows: mean fluorescence intensity with OVA/ mean fluorescence intensity without OVA. (B, C, D) Naive DesTCR CD8+ T cells were cocultured with C57Bl/6 LSEC or CD11c+ splenic DC for 5 days with 40 μg/mL anti-B7-H1 or rat isotype control antibody or (E) with B7-H1−/− LSEC. At day 5, viable T cells were harvested and restimulated with plate-bound anti-CD3. After 24 hours, IL-2 (B) and IFN-γ (C and E) production was determined by ELISA or (D) specific cytotoxic activity was measured by an in vitro kill assay. Percent specific kill was calculated as follows: 100 − [100*(CFSEhigh/CFSElow)probe/ (CFSEhigh/CFSElow)control]. (F) T cell expansion during co-culture was determined as described in Fig. 1C. (G) CFSE proliferation profiles of DesTCR T cells co-cultured with C57Bl/6 LSEC (dark line), B7-H1−/− LSEC (dashed line) or T cells alone (filled gray) were determined at 72 hours of culture. Values (B-F) are depicted ±SEM. Statistical significance was calculated using a Student t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

Figure 4.

Supplemental signaling through CD28 prevents antigen-specific tolerization of naive CD8+ T cells by LSEC. (A) Naive OT-1 CD8+ T cells were cocultured with LSEC or DC in the presence or absence of 100 μg/mL OVA protein. After 24 hours of T cell/APC co-culture, the expression levels of B7-H1 on 7-AAD–negative, ME9F1+-gated LSEC or CD11c+-gated DC were determined. Filled gray line: isotype control. APC plus OVA: black solid line: APC without OVA: gray dashed line. Similar results were obtained with naive DesTCR T cells cultured on H-2Kb+ compared with H-2Kk+ LSEC. (B) Absolute levels (mean fluorescence intensity, MFI) of CD80 and CD86 expression on nonstimulated LSEC and DC. (C) Naive DesTCR CD8+ T cells were cocultured with H-2Kb+LSEC or CD11c+ splenic DC for 5 days in the presence or absence of 10 μg/mL anti-CD28 or hamster isotype control antibody. At day 5, viable T cells were harvested and restimulated on plate-bound anti-CD3. After 24 hours, IL-2 production was determined by ELISA. (D) Naive DesTCR CD8+ T cells were cocultured with B6 or B7-H1−/− DC for 5 days and restimulated on plate-bound anti-CD3. After 24 hours, IFN-γ production was determined by ELISA. Values are depicted ±SEM. Statistical significance was calculated using a Student t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

B7-H1 Signalling Is Required for Induction of CD8+ T Cell Tolerance by LSEC.

We found that LSEC-stimulated naive CD8+ T cells up-regulated PD-1, which contributes to the induction of peripheral CD8+ T cells tolerance by resting DC in vivo.8 Reciprocally LSEC showed increased expression levels of B7-H1. To investigate whether B7-H1 signaling is involved in CD8+ T cell tolerance induction by LSEC, naïve CD8+T cells were cocultured with LSEC or DC in the presence or absence of blocking anti-B7-H1 or PD-1 antibodies. After 5 days of coculture with antigen-presenting LSEC CD8+ T cells did not produce IL-2 or IFN-γ on restimulation or showed cytotoxicity toward antigen-expressing target cells (Fig. 3B-D). The addition of blocking anti-B7-H1 (Fig. 3B-D) as well as anti–PD-1 (not shown), but not isotype control antibodies, abrogated tolerance induction, because T cells produced both IL-2 and IFN-γ and regained cytotoxic activity (Fig. 3B-D), showing that tolerance induction was B7-H1 dependent. Because B7-H1 can be expressed on activated T cells,27 we determined whether LSEC-derived or T cell–derived B7-H1 signaling induced tolerization of CD8+ T cells by coculture of naive CD8+ T cells with B7-H1–deficient antigen-presenting LSEC. Under these conditions, T cells produced high levels of interferon gamma (IFN-γ) on restimulation comparable to cytokine levels produced by T cells activated by DC (Fig. 3E). Thus, B7-H1 signals derived from LSEC were sufficient and necessary for induction of CD8+ T cell tolerance through interaction with PD-1 expressed on CD8+ T cells.

B7-H1 signaling was not only important for the induction of CD8+ T cell tolerance, but was also important for survival of tolerized CD8+ T cells. During culture on B7-H1–deficient LSEC, CD8+ T cell expansion was severely reduced (Fig. 3F). This was not because of inability of CD8+ T cell to proliferate because T cells stimulated by B7-H1–deficient LSEC exhibited a similar proliferation profile to those stimulated by C57Bl/6 LSEC (Fig. 3G). Taken together, these results show that B7-H1 signaling by antigen-presenting LSEC is required for both T cell tolerance induction and survival of tolerized T cells.

The Balance Between Costimulatory/Inhibitory Signalling Is Important for Induction of CD8+T Cell Tolerance by LSEC.

Both LSEC and DC expressed B7-H1 (Fig. 4A) and induced PD-1 expression on CD8+ T cells (Fig. 1A), raising the question of why only LSEC induced CD8+ T cell tolerance. Compared with LSEC, DC expressed higher levels of CD80 and CD86 (Fig. 4B). Moreover, the expression levels of CD80/CD86 on either DC or LSEC hardly changed during antigen-specific interaction with T cells (Fig. 3A and 4A). This suggested that costimulation is preset to high levels on DC and to low levels on LSEC. To test whether lack of CD28 co-stimulation allows B7-H1–mediated induction of CD8+ T cell tolerance, naive CD8+ T cells were cocultured with LSEC in the presence of stimulating anti-CD28 antibody. CD8+T cell tolerance induction was only overcome in the presence of anti-CD28 (Fig. 4C). CD28 costimulation, however, did not stimulate CD8+T cells to produce cytokine levels similar to those of T cells activated by DC, suggesting that B7-H1 signaling continuously restricted CD28 costimulation. Also, B7-H1−/− DC showed enhanced T cell effector activity compared with C57Bl/6 DC (Fig. 4D), confirming this hypothesis. Collectively, this indicates that the balance between costimulatory and coinhibitory signals given by LSEC dictates induction of CD8+ T cell tolerance.

Distinct Phenotype of LSEC-Induced Tolerant CD8+ T Cells.

We found that LSEC-tolerized T cells lost CD25 expression but highly expressed both CD44 and CD62L (Fig. 5A). In contrast, activated CD8+ T cells expressed CD25 and CD44 but down-regulated CD62L (Fig. 5A). Thus, LSEC-tolerized CD8+ T cells are distinct from naïve CD25negCD44lowCD62Lhigh T cells and activated CD25highCD44highCD62Llow T cells. Furthermore, LSEC-tolerized T cells expressed both the IL7R (CD127) and the IL-2/IL-15R (CD122) (Fig. 5A), suggesting that they can receive survival signals from IL-15 and IL-7. Taken together, the phenotype of LSEC-tolerized CD8+T cells (CD25lowCD44highCD62LhighCD122highCD127high) is different from activated and naïve T cells.

Figure 5.

In vivo function and phenotype of tolerant CD8+ T cells. Naïve DesTCR CD8+ T cells were cocultured with C57Bl/6 LSEC or CD11c+ splenic DC. After 5 days, viable cells were (A) harvested and stained for CD25, CD44, CD62L, CD122, CD127, and Bcl-2. Filled gray line: isotype control; black solid line: specific stain of tolerant CD8+ T cells; gray dashed line: specific stain of activated CD8+ T cells. (B) Naive DesTCR CD8+ T cells were co-cultured with H-2Kb+LSEC or CD11c+ splenic DC. At day 5 viable T cells were harvested and restimulated on plate-bound anti-CD3 with or without anti-CD28. Alternatively T cells were restimulated with immature or lipopolysaccharide-matured CD11c+ DC. After 24 hours, IL-2 production was determined by ELISA. (C) OT-1 CD8+ T cells were co-cultured in vitro with H2b LSEC or DC in the presence of 100 μg/mL OVA protein. After 5 days, viable tolerant or activated CD8+ T cells were harvested and 4 to 5 × 105 cells were adoptively transferred intravenously into C57BL/6 mice. Mice were injected 1 day later intravenously with 5 × 105 SIINFEKL-loaded BMDC for restimulation of T cells in vivo followed after 1 day by injection of 107 of SIINFEKL-loaded CFSEhigh-labeled spleen cells mixed with 107 unloaded CFSElow-labeled spleen cells. After overnight in vivo incubation, cells were harvested from liver, spleen, and lymph node, and CFSE-stained cells were quantified by flow cytometry. Percent specific kill was calculated as described in Fig. 3. Values (B) are depicted ±SEM. Statistical significance was calculated using a Student t test (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001).

Interestingly, neither additional anti-CD28 signaling nor antigen presented by immature or lipopolysaccharide-matured CD11c+ DC could reactivate tolerant CD8+ T cells in vitro (Fig. 5B), indicating that the tolerant phenotype is extremely stable. To determine the functionality of LSEC-tolerized CD8+ T cells in vivo, tolerized or activated CD8+ T cells were transferred into congenic animals. LSEC-tolerized CD8+ T cells did not show in vivo cytotoxicity, whereas DC-activated CD8+ T cells efficiently killed specific target cells in all organs investigated (Fig. 5C), indicating that LSEC-induced CD8+ T cell tolerance is also stable in vivo.

Discussion

Elucidating the cellular and molecular mechanisms of tolerance induction in the liver is critical to our understanding of autoimmune liver disease and chronic hepatitis caused by persistent infection. Elimination of activated T cells and active tolerization of naive T cells have been reported to contribute to the tolerogenic function of the liver.21, 28–30

APC, in particular mature DC, can prime adaptive immune responses31 by provision of signal 1 (TCR), signal 2 (CD28), and signal 3 (IL-12) but induce self-tolerance when immature.25 LSEC, a unique liver-resident APC population, can also induce CD8+ T cell tolerance toward cross-presented exogenous antigens.21–24 We now show that CD8+ T cell tolerance induction by LSEC is a B7-H1–dependent nondeletional process that involves mutual stimulation of both LSEC and CD8+ T cells as a licensing step preceding the induction of T cell tolerance.

Antigen-specific initial activation of naive CD8+ T cells by tolerogenic LSEC was as efficient as by immunogenic splenic DC (Fig. 1). Proliferation and expansion of CD8+ T cells during the first 72 hours did also not differ after activation by LSEC or DC, demonstrating that tolerance was preceded by an initial stimulation step as reported in other experimental systems.32 At later points, however, LSEC-stimulated T cells did not expand to the same extent as DC-stimulated T cells. The high Bcl-2 expression levels in LSEC-stimulated CD8+ T cells likely allowed them to proliferate and expand, as deletion by apoptosis of CD8+ T cells is inhibited by Bcl-2.33 This excludes clonal deletion as the mechanism by which LSEC achieve CD8+ T cell tolerance, differentiating them from resting DC and hepatocytes, which induce deletional T cell tolerance.8, 28

We found that LSEC were stimulated during initial antigen-specific interaction detected by rapid up-regulation of CD54 and CD106. In vivo, both liver-expressed CD54 and CD106 are involved in hepatic retention of activated CD8+ T cells.34 Both adhesion molecules may stabilize the antigen-dependent interaction of LSEC with naive CD8+ T cells supporting T cell stimulation.35

The induction of T cell immunity or tolerance is dependent on delivery of cosignaling.36 Importantly, cognate interaction did not alter expression of costimulatory CD80/86 on LSEC and DC but increased expression levels of the coinhibitory B7-H1 exclusively on LSEC. Predominant signaling through PD-1 and cytotoxic T lymphocyte-associated antigen 4 on T cells by resting DC leads to CD8+ T cell tolerance in vivo.8 On LSEC (B7-H1high CD80/86low), the PD-1 ligand B7-H1 was necessary and sufficient for induction of CD8+ T cell tolerance, but only together with low levels of costimulation. Additional CD28 signals overruled B7-H1–dependent CD8+ T cell tolerization by LSEC, indicating that the costimulatory/coinhibitory balance determines whether tolerance or immunity is induced.37 Although IL-12 is necessary for full activation of CD8 T cells, exogenously added IL-12 did not modify tolerance induction by LSEC (data not shown and Limmer et al.21). Because expression of the IL-12 receptor is dependent on CD28 costimulation,38 the low surface expression of CD80/CD86 on LSEC may ensure that CD8 T cells stimulated by LSEC do not become receptive for IL-12. Thus, the functional phenotype of LSEC favors co-inhibition over costimulation and determines their tolerogenic function.

Functional maturation of DC is controlled by incorporation of signals from the environment, such as for instance Toll-like receptors. In contrast, contact with immune-regulatory mediators (such as IL-10, transforming growth factor beta) can render DC tolerogenic.6, 39 Tolerogenic maturation of LSEC, in contrast, was independent of exogenous mediators and only required cognate interaction with naive CD8+ T cells. PD-1 and simultaneous B7-H1 up-regulation on T cells and LSEC, respectively, were not only instrumental for exchanging tolerogenic signals but also may represent a licensing step. Whereas glutaraldehyde-fixed LSEC were still able to present antigen and initially stimulate T cells (CD69 up-regulation), tolerance was not induced as T cells underwent incomplete stimulation and died (data not shown). Although we cannot exclude that LSEC express an uncharacterized molecule involved in mediating tolerance, the inability of both fixed and B7-H1–deficient LSEC to induce full tolerance strongly argues for a critical role of B7-H1 in CD8+ T cell tolerance induction. Moreover, simultaneous licensing of both LSEC and T cells may assure that only antigen-specific T cells are rendered tolerant.

Additionally, B7-H1–deficient LSEC did not support T cell expansion. This assigns a so far unrecognized role to B7-H1 in promoting survival of tolerant CD8+ T cells. In thymocytes, B7-H1–mediated signaling via PD-1, in combination with costimulation, down-regulates Bcl-2 and inhibits positive selection.40 Additionally, T cell apoptosis induction by hepatocytes and hepatic stellate cells was shown to be B7-H1 dependent.41, 42 The apparent contradictory promotion of CD8+ T cell survival by LSEC-derived B7-H1 signaling could be attributable to lack of costimulatory molecules on LSEC or the additional production of cytokines such as transforming growth factor beta by stellate cells.42 Hepatocytes induce B7-H1–dependent apoptosis in activated T cells. However, apoptosis induction in naïve T cells may be differently regulated. Furthermore, the binding of B7-H1 to another putative receptor present on activated CD8+ T cells43 could also explain the survival signals originating from B7-H1.

Phenotypically, LSEC-tolerized CD8+ T cells were clearly distinctive from naive and activated CD8+ T cells, being CD25lowCD44highCD62Lhigh. Furthermore, tolerant CD8+ T cells expressed the homeostatic cytokine receptors CD122 and CD127, indicating that IL-15 and IL-7 may promote survival of tolerant CD8+ T cells in vivo. Although tolerant T cells (CD44highCD62LhighCD122highCD127high) resemble central memory CD8+ T cells,44 on adoptive transfer LSEC-tolerized CD8+ T cells remained functionally tolerant. This combination of T cell phenotype and function resemble more the exhausted CD8+ T cells that are found during persistent viral infection.45 Therefore, LSEC may contribute to the generation of these functionally inert T cells in vivo. After release of large concentrations of viral antigens from infected hepatocytes—observed for hepatitis B and C virus infection—viral antigens may be cross-presented by LSEC to novel thymic emigrant T cells,46 skewing anti-viral immunity through generation of tolerant T cells that fail to eliminate virus-infected hepatocytes.

Our findings reveal a fundamental difference in immune regulation between DC and LSEC. DC incorporate signals from their microenvironment, leading to either functional maturation and immunity or tolerogenic maturation and tolerance, whereas tolerance induction by LSEC is a cell-autonomous maturational program after antigen-specific contact with T cells. These new insights on LSEC-induced immune tolerance induction will help us to deliberately manipulate the local immune response in the liver.

Acknowledgements

The authors acknowledge the technical assistance of F. Küpper and of E. Endl, A. Dolf and P. Wurst from the Flow Cytometry Core Facility at the Institute of Molecular Medicine and Experimental Immunology, University of Bonn, Germany.

Ancillary