fms-like tyrosine kinase 3 ligand
DC are believed to play important roles in the induction and regulation of immune responses in the liver, an organ implicated in peripheral tolerance. Since the liver is located downstream of the gut, it is constantly exposed to bacterial LPS. Our recent observations indicate that prior exposure to endotoxin modulates subsequent liver DC responses to this TLR4 ligand. In this study, we demonstrate that endotoxin modifies the capacity of mouse liver myeloid DC (MDC) activated by CpG (TLR9 ligand) to direct Th1-type responses. IL-12 production by liver MDC was significantly lower than that of spleen MDC following CpG or Imiquimod (R837; TLR7 ligand) activation in vitro. In addition, allogeneic T cells stimulated by CpG-activated liver MDC secreted significantly lower levels of IFN-γ than T cells stimulated with CpG-activated spleen MDC. A similar effect on liver DC was observed in response to in vivo CpG administration. This effect may be explained by exposure of the DC to endotoxin, because LPS attenuated IL-12 production by CpG-stimulated liver MDC, both in vitro and in vivo. Moreover, attenuation of the response to CpG was not observed in liver MDC from TLR4-mutant (C3H/HeJ) mice, in which TLR4 signaling is impaired. These data suggest that endotoxin-induced ‘cross-tolerance’ to TLR ligands in liver DC may contribute to down-regulation of hepatic immune responses.
Microbial-derived TLR ligands, such as LPS or CpG, induce the functional maturation of BM-derived DC and thereby promote adaptive Ag-specific immune responses 1, 2. Immature DC resident in peripheral tissues recognize invading pathogens via TLR; after exposure to microbial stimuli, DC undergo a complex process of maturation that results in their migration to secondary lymphoid organs and up-regulated expression of MHC and costimulatory molecules essential for efficient T cell activation 3. In addition, DC represent a critical source of IL-12, which drives Th1 responses 4.
Bacterial DNA contains immunostimulatory CpG motifs that interact with TLR9, expressed primarily by B cells and DC. This interaction leads to production/secretion of cytokines, chemokines and Ig, and skews the host immune reactivity in favor of Th1 responses (IL-12 and IFN-γ) 5. The striking immunostimulatory activity of CpG offers considerable therapeutic potential, including the use of CpG (alone or in combination with other agents) to improve resistance to cancer and infection and to deviate responses from Th2 to Th1 in allergic hypersensitivity 6. Recently, several clinical trials of CpG in hepatitis B virus infection have been reported 7, 8. However, CpG can also trigger deleterious autoimmune reactions and block the apoptotic death of activated lymphocytes 9. Several reports indicate that interaction of TLR ligands with DC likely plays a role in this process 10, 11. With respect to autoimmune liver disease, CpG is considered an adjuvant for the breakdown of self tolerance in mice 12, 13 and in patients with primary biliary cirrhosis 14, 15. These findings suggest that CpG might play a critical role in hepatic immune responses.
Endotoxin (LPS) tolerance, i.e. hyporesponsiveness to LPS stimulation after initial low-dose exposure, has been well documented in the laboratory 16 and clinic 17, 18. While underlying mechanisms have not been elucidated, monocytes/macrophages and DC play a role in this phenomenon 19, 20. Stimulation of macrophages with prototypic ligands for TLR2 or TLR9 also induces this state of hyporesponsiveness towards subsequent stimulation with the same ligand 21–23. Moreover, TLR4 or TLR9 ligands can substitute for each other, mediating ‘cross-tolerance’ in vitro and in vivo23–25. Interestingly, Dalpke et al.26 reported recently that pretreatment with LPS induced refractoriness (cross-tolerance) to CpG challenge in a mouse model of galactosamine-primed liver damage. In contrast to these findings, several groups have reported that selected TLR ligand combinations do not induce tolerance, or have synergistic effects on the production of inflammatory mediators 27–29. These observations suggest that the development of tolerance (or cross-tolerance) is dependent on the cell type, the respective TLR ligand and the experimental conditions.
With respect to mouse DC, An et al.30 found that TLR9 mRNA expression in BM-derived (cultured) DC was up-regulated following LPS stimulation, and that TNF-α production was increased by LPS + CpG stimulation. Similar synergistic effects of TLR4 and TLR9 ligation on IL-12 production have been reported for mouse DC 31. However, these cultured DC may differ significantly from those that differentiate in vivo32. In addition, DC subsets, Ag dose, stage of DC maturation, and microenvironmental signals determine the outcome of immune responses 33.
Several groups, including ours 34–36, have reported that liver DC are less immunostimulatory than spleen DC. Because the liver is located downstream of the gut, it is constantly exposed to endotoxin. We have observed recently that liver DC express comparatively low levels of TLR4 mRNA and poorer ability than lymphoid tissue DC to activate allogeneic T cells in response to LPS 34. In the present study, we have investigated the influence of endotoxin on responses triggered by the TLR9 ligand, CpG, in liver and spleen DC.
Liver CD11c+ cells exhibit less allogeneic T cell stimulatory activity than spleen CD11c+ cells
As reported previously by us 34 and others 35, 36, freshly isolated normal C57BL/10 (B10) mouse bulk liver CD11c+ cells exhibited substantially less normal T cell allostimulatory capacity and induced lower IFN-γ production by C3H/HeJ (C3H) T cells compared to normal B10 spleen CD11c+ cells isolated concomitantly from fms-like tyrosine kinase 3 ligand (Flt3L)-treated mice (Fig. 1A, and data not shown). However, evidence has accumulated that mouse CD11c+ cells constitute a heterogenous population. Thus, as reported elsewhere 35–38, further analysis revealed that bulk liver CD11c+ cells comprised several different populations. As shown in Fig. 1B, the proportions of B220+CD11c+ cells (putative plasmacytoid DC, pDC) and DX-5+CD11c+ cells (putative NK cells) were significantly higher in the liver (two- to threefold and three- to fourfold, respectively) than in the spleen.
Thus, in order to compare the same predominant population of CD11c+ cells (myeloid DC, MDC), these latter cells were excluded and the function of CD11c+ cells re-examined. As shown in Fig. 1C, the allostimulatory capacity of liver and spleen CD11c+B220–DX-5– cells (MDC) in 72-h MLR was similar. In addition, IFN-γ production (intracellular and secreted) by allogeneic T cells induced by freshly isolated liver MDC was not different from that induced by spleen MDC (Fig. 1D). However, basal IL-12p40 production by unstimulated liver MDC was lower than that of spleen MDC (Fig. 1E). IL-12p70 was not detected without cell activation.
CpG-stimulated liver MDC produce comparatively low levels of IL-12
We have shown previously 34 that, compared with spleen DC, mouse liver DC express lower TLR4 mRNA and inferior capacity to activate normal allogeneic T cells following LPS stimulation. By contrast, TLR9 expression is not significantly different between freshly isolated liver and spleen MDC of Flt3L-treated mice (Fig. 2A). To determine whether liver and spleen DC responded differently to the TLR9 ligand CpG, CD11c+ cells from both tissues were cultured with CpG (1 μg/mL) overnight (18 h). When bulk CD11c+ cells were cultured with CpG, liver CD11c+ cells exhibited significantly less T cell allostimulatory capacity than CpG-stimulated spleen CD11c+ cells of Flt3L-treated animals (Fig. 2B). However, the stimulatory capacity of purified liver and spleen MDC did not differ significantly (Fig. 2C), consistent with the finding that cell surface expression of MHC class II (IAb) and costimulatory molecules (CD40, CD80, and CD86) was similar on each population (Fig. 2D, and data not shown). Similar results were obtained when lower concentrations of CpG (0.1 and 0.01 μg/mL) were used (data not shown).
Although the T cell allostimulatory capacities of CpG-stimulated liver and spleen MDC were similar, IL-12 production (both intracellular IL-12p40 and IL-12p70 secretion) by liver MDC was significantly lower (p<0.05) than that of spleen MDC in response to CpG. IL-12p40 staining between liver and spleen B220 pDC was similar. (Fig. 2E).
CpG-stimulated liver MDC induce comparatively low levels of IFN-γ production
Consistent with these findings, allogeneic (C3H) T cells stimulated by CpG-activated B10 liver MDC produced significantly lower IFN-γ levels compared to T cells stimulated with spleen CpG-activated MDC (p<0.05; Fig. 2F). By contrast, IL-4 and IL-10 secretion did not differ between T cells stimulated by CpG-activated liver and spleen MDC (data not shown). CpG-activated liver MDC isolated from normal (non-Flt3L-treated) mice displayed similar, markedly reduced ability to induce IFN-γ in allogeneic T cells compared with normal CpG-activated spleen MDC (Fig. 3A). The inferior ability of liver MDC to produce IL-12 in comparison to spleen MDC was observed over a range of CpG concentrations. Similar differences were observed when liver MDC and spleen MDC were stimulated with a range of TLR7 ligand (R837) concentrations (Fig. 3B).
Attenuation of the response of liver MDC to CpG requires intact TLR4 signaling
To ascertain whether intact TLR4 signaling was required for attenuation of the LPS response by liver MDC, we examined IL-12p40 production by liver and spleen MDC isolated from TLR4-mutant mice (C3H). Similar levels of IL-12 production were observed (Fig. 4A), supporting the concept that intact TLR4 signaling is required for attenuation of the CpG response. Moreover, liver and spleen MDC from the TLR4-mutant mice induced similar levels of IFN-γ in allogeneic T cells (Fig. 4B).
Liver MDC produce less IL-12 in response to CpG stimulation in vivo
To determine the response of liver and spleen DC to CpG-containing oligonucleotides in vivo, 100 μg of CpG were injected i.v. into Flt3L pretreated B10 mice. Two, 6, and 16 h after CpG injection, the cells were harvested and purified. At all time points, freshly isolated, bulk CD11c+ cells from liver and spleen of CpG-stimulated mice exhibited higher normal T cell allostimulatory capacity compared with bulk CD11c+ cells from control (PBS-injected), Flt3L-mobilized mice. In addition, the allostimulatory capacity of CpG-stimulated liver bulk CD11c+ cells was significantly lower (p<0.05) than that of spleen bulk CD11c+ cells (Fig. 5A, and data not shown). Since we considered it possible that liver DC activated in vivo might migrate to secondary lymphoid tissue and be replaced by new, immigrant cells at later time points after CpG injection, we compared the functions of DC isolated 2 h after CpG injection.
At 2 h, total bulk spleen and liver CD11c+ cell numbers did not differ significantly between CpG-treated and control groups, but differences in DC subset composition were observed. Although the spleen contained a similar percentage (compared to control Flt3L-mobilized controls) of B220+CD11c+ cells after CpG injection, the liver exhibited a higher incidence of these cells (Fig. 5B). Thus, we compared the T cell allostimulatory capacities of purified MDC. Consistent with our observations on in vitro CpG-stimulated MDC (Fig. 2), liver MDC from CpG-injected mice exhibited similar allostimulatory capacity compared to spleen MDC (Fig. 5C). However, IL-12p40 production by liver MDC from CpG-injected mice was lower than that of spleen MDC from the same animals (Fig. 5D).
LPS attenuates CpG-induced IL-12 production by liver MDC
Because the liver is constantly exposed to bacterial LPS from the gut, we hypothesized that LPS might influence the liver MDC response to CpG. Purified, freshly isolated liver and spleen MDC from Flt3L-treated mice were cultured with CpG (1 μg/mL), with or without LPS for 18 h. Levels of cell surface costimulatory (i.e. CD80 and CD86) and MHC class II (IAb) molecule expression and naive T cell allostimulatory capacity of CpG-stimulated MDC were not enhanced by adding LPS to the cultures (data not shown). Moreover, CpG-stimulated IL-12p70 production by spleen MDC in response to CpG was not significantly different, with or without LPS (Fig. 6A). By contrast, IL-12p70 production by liver MDC was reduced significantly in the presence of LPS (Fig. 6A).
We further investigated the influence of LPS on MDC in vivo. LPS was injected i.v. and 2 h later, liver and spleen MDC were isolated and purified. The cells were then cultured with CpG for 18 h. IL-12p70 production by liver MDC from LPS-injected mice was significantly lower than that of spleen MDC from the same animals (Fig. 6B). In addition, IFN-γ production by T cells stimulated by CpG-activated liver MDC from LPS-injected mice was significantly lower than that of T cells stimulated by the corresponding spleen MDC (Fig. 6C). IL-4 and IL-10 production was not significantly different between T cell populations stimulated by these liver or spleen MDC (data not shown).
Immune responses in, or elicited by the liver can result in tolerance rather than immunity. Hepatic tolerance was demonstrated initially by the acceptance of liver allografts across MHC barriers, without immunosuppressive therapy 39, 40. In addition, the liver appears to play an important role in the induction of oral tolerance, as well as in the development/persistence of certain viral infections and cancer 41–43. This comparative immune privilege might reflect, at least in part, functional differences in hepatic compared with secondary lymphoid tissue APC, including liver DC, Kupffer cells, or liver sinusoidal-endothelial cells 41, 42.
Numerous factors induce the maturation of DC, including (1) pathogen-associated molecular patterns, such as bacterial LPS or CpG-containing oligonucleotides and double-stranded RNA, and (2) the balance between local pro- (TNF-α, IL-1β, IL-6) and anti-inflammatory cytokines (IL-10, TGF-β) and prostaglandins 44. The liver exhibits features of a tolerogenic microenvironment that is rich in IL-10 and TGF-β 43, 45. In addition, the flow of blood from the intestines to the liver results in continuous exposure of hepatic leukocytes, endothelial cells, and other cells to LPS. We have reported recently that highly purified liver DC express low levels of TLR4 and induce comparatively low proliferative responses in naive allogeneic T cells following exposure to LPS 34. This suggests that other signals are required to promote the (full) maturation of liver DC and optimal priming of T cells for efficient effector function. Here we have compared the Th1-polarizing capacity of liver and spleen DC in response to the TLR9 ligand, CpG.
Recently, several groups have shown that liver CD11c+ cells comprise distinct subpopulations 35–38. As shown in this study, and consistent with previous reports using normal (not growth factor-mobilized) mice 35–37, the mouse liver contains more B220+ and DX-5+ CD11c+ cells than spleen in Flt3L-treated animals. When we compared the normal T cell stimulatory capacity of DC from which these populations were removed, the remaining cells (MDC) induced similar allogeneic T cell proliferation and cytokine production. This indicates that liver and spleen MDC have similar capacity to activate T cells, at least in primary allogeneic MLR.
The T cell allostimulatory capacity of purified liver and spleen MDC following CpG stimulation was similar, both in vitro and in vivo. This finding differs from our observation concerning LPS-activated liver MDC 34. The reason for this is not clear, but transcription profiling has revealed that different subsets of genes are regulated by distinct TLR ligands 46. For example, CpG and LPS each mobilize nuclear factor κB, but gene microarray analysis indicates that CpG affects only a subset of the genes activated by LPS 46. In contrast to their ability to up-regulate MHC and costimulatory molecule expression and their capacity to induce naive allogeneic T cell proliferation, the Th1-polarizing capacity of purified MDC was significantly different between liver and spleen MDC stimulated with CpG, both in vitro and in vivo. This could reflect different thresholds for the induction of different responses. Thus, CpG provides a signal to liver MDC sufficient to up-regulate surface costimulatory molecules, but not to promote Th1 polarization as effectively as spleen MDC.
Consistent with previous studies 47, mouse splenic DC were activated in vivo by exposure to CpG. Responses of liver DC populations to CpG in vivo have not previously been investigated. The incidence of CD11c+B220+ DC (most of which are pDC 48) in the liver was significantly higher than in spleen. Similar results were obtained using normal control (not Flt3L-treated) mice (data not shown). Jomantaite et al.36 have reported that murine CMV infection selectively increases B220+ DC in the liver. Since murine CMV activates DC via TLR9 ligation, this observation is consistent with our finding that the incidence of B220+ DC in the liver is increased after CpG injection (Fig. 3B). The reason(s) why B220+ cells are increased in the liver in response to CpG has not been clarified. However, we have shown that while CC chemokine receptor 7 gene expression by liver MDC and pDC is enhanced, the migratory response of pDC to the CC chemokine receptor 7 ligand CCL19 is lower than that of MDC in an in vitro system 49. Thus, we suggest that after CpG administration, many activated MDC migrate to secondary lymphoid tissue, and that consequently, the proportion of pDC is increased.
Recently, Dalpke et al.26 observed the induction of ‘cross-tolerance’ to TLR4/9 both in vitro using a macrophage cell line, and in vivo using the galactosamine-induced liver injury model. In addition, Sanchez-Campillio et al.50 demonstrated that LPS down-regulated TLR9 mRNA in a murine hepatocyte cell line and further showed that LPS injection of mice down-regulated TLR9 mRNA in the liver. These findings suggest that pretreatment with LPS (mimicking the normal physiological situation in the liver) affects the response to CpG in the liver. Notably, the interaction between TLR4/9 ligands in liver DC has not been investigated. Herein, we have shown that LPS attenuates IL-12p70 production by liver DC in response to CpG, both in vitro and in vivo. This attenuation appears to be dependent on TLR4 signaling, as liver and spleen MDC from TLR4 signaling-deficient C3H mice produced similar levels of IL-12.
Our observation of attenuated liver MDC responses to CpG differs from previous reports using mouse BM-derived (cultured) DC 30, 31. The reasons for this discrepancy have not been clarified, but there are several possible explanations. First, TLR expression and the response to TLR ligands by cultured DC may differ significantly from those of DC that differentiate in vivo. For example, TLR4 mRNA is expressed at high levels in mouse BM-derived MDC, but at low levels in splenic MDC 33. This differential expression is reflected in their different abilities to produce IL-12p70 in response to LPS 33. Furthermore, TLR4 expression is also different between human circulating MDC and cultured (monocyte-derived) DC 1. Second, since the liver is exposed continuously to LPS, and local synthesis of IL-10 and TGF-β is prevalent in the liver 43, 45, these features of the liver microenvironment may affect the response to TLR ligands.
In our experiments, we used CpG-B to activate DC and to elicit cytokine production. At least three structurally distinct classes of synthetic CpG have been described 51, 52. We compared directly the in vitro and in vivo effects of CpG-B and CpG-C (CpG 2395) and obtained very similar results (data not shown). Thus, the effects observed are not CpG-B-specific.
The immunostimulatory activity of CpG has prompted interest in its potential clinical use in a diverse array of conditions. Evidence from phase I and II clinical trials has indicated that CpG are safe and that in some cases, they boost the immunostimulatory activity of co-administrated vaccines 6. However, as shown in this study, CpG-stimulated liver MDC clearly exhibit less Th1-polarizing ability than similarly stimulated spleen MDC. This suggests that further study of the type, dose, duration, site, and mode of CpG delivery is merited, especially in patients with liver disease.
Materials and methods
Male B10 (H2b) and C3H/HeJ (C3H; H2k) mice (8–12 wk of age) were purchased from The Jackson Laboratory (Bar Harbor, ME). C3H mice exhibit a point mutation in the third exon of the TLR4 receptor gene that results in a defective (low) response to LPS. The mice were maintained in the specific pathogen-free facility of University of Pittsburgh School of Medicine in accordance with institutional guidelines. Experiments were conducted in accordance with the National Institutes of Health Guide for Case and Use of Laboratory Animals and under an Institutional Animal Care and Use Committee-approved protocol. Mice received Purina rodent chow (Ralston Purina, St. Louis, MO) and tap water ad libitum.
RPMI 1640 supplemented with 10% v/v heat-inactivated fetal calf serum (Atlanta Biologicals, Lawrenceville, GA), non-essential amino acids, L-glutamine, sodium pyruvate, penicillin-streptomycin and 2-ME (all from Life Technologies, Gaithersburg, MD) was used as the culture medium. Chinese hamster ovary cell-derived recombinant human Flt3L was provided by Amgen (Seattle, WA). LPS (ultra pure, Escherichia coli K12) was purchased from InvivoGen (San Diego, CA). CpG 1826 (CpG-B) was purchased from Coley Pharmaceutical Group (Wellesley, MA) and Imiquimod (R837) from InvivoGen. In some experiments, mice received CpG (100 μg), LPS (1 μg) or PBS i.v. via the lateral tail vein.
Isolation of DC
CD11c+ cells were isolated from spleens and livers of normal animals or mice given the DC poietin Flt3L (10 μg/mouse/day i.p., for 10 days). Bulk DC were enriched by density centrifugation using Nycodenz (Sigma, St. Louis, MO) and further purified using anti-mouse CD11c-coated magnetic beads (Miltenyi Biotec, Auburn, CA) by MACS®, as described in detail 34, 49 For MDC purification, cells harvested after density centrifugation were incubated with biotin-conjugated rat anti-mouse B220/CD45R and CD49b and depleted by negative selection. CD11c+ cells were then positively selected using MACS®. The purity of MDC (CD11c+B220–DX-5–) was consistently >95%.
Cells were incubated with 10% v/v normal goat serum for 20 min to avoid non-specific Ab binding. They were then incubated for 30 min with FITC-, PE-, or Cy-Chrome-conjugated mAb to detect cell surface expression of CD11c (clone HL3), B220/CD45R (RA3-6B2), IAb β-chain (25-9-17), CD40 (3/23), CD80 (16-10A1), CD86 (GL1) or CD49b/pan-NK cells (DX5). These mAb and appropriate Ig isotype controls were obtained from BD PharMingen (San Diego, CA). To detect intracellular TLR9 expression, the cells were permeabilized with 0.1% saponin, then incubated with anti-mouse TLR9 mAb (Hycult Biotechnologies, Uden, The Netherlands) followed by streptavidin-PE (BD PharMingen), as described 53. Flow cytometric analysis was performed using a FACSCalibur (BD PharMingen), and results were expressed as % positive cells and mean fluorescence intensity.
Intracellular cytokine staining
Bulk CD11c+ cells or MDC were treated with brefeldin A (GolgiPlugTM, 1 μL/mL; BD PharMingen) for 5 h, then labeled with FITC- or Cy-Chrome-conjugated mAb and fixed in 4% paraformaldehyde. The cells were permeabilized with 0.1% saponin, then incubated with PE-conjugated anti-IL-12p40/p70 (C15.6) or rat IgG (BD PharMingen) for 30 min. T cells recovered from 72-h MLR were restimulated with plate-bound anti-CD3ϵ (10 μg/mL, clone 17A2) and 2 μg/mL soluble anti-CD28 mAb (37.51) (each from BD PharMingen) for 5 h at 37°C, in the presence of brefeldin (GolgiPlugTM, 1 μL/mL). After cell surface staining with FITC-CD4 and Cy-Chrome-CD3 mAb, the cells were fixed, permeabilized with 0.1% saponin, then stained for intracellular IFN-γ with PE-conjugated anti-IFN-γ (XMG 1.2) or isotype control Ig (BD PharMingen).
DC were cultured (106/mL) with graded doses of CpG and/or LPS for 18 h. Levels of IL-12p70 in culture supernatants were determined by ELISA using commercial kits (R&D Systems, Minneapolis, MN) and following the manufacturer's instructions. The sensitivity limit was 2.5 pg/mL. IFN-γ, IL-10, and IL-4 secretion in 72-h MLR culture supernatants (DC: 2×104; T cells: 2×105) was determined using ELISA kits (Biolegend Inc., San Diego, CA). The sensitivity limits were 4, 30, and 1 pg/mL, respectively.
T cell proliferation assay
Graded doses of B10 bulk CD11c+ cells or purified MDC were γ-irradiated, then used as stimulators in 72-h primary MLR with nylon wool-purified, allogeneic (C3H) T cells as responders in 96-well, round-bottomed culture plates (Corning Inc., Corning, NY). For the final 18 h of the culture, 1 μCi (3H) was added to each well, and 3H incorporation determined using a scintillation counter. Results are expressed as means ± 1 SD.
Data are expressed as means ± 1 SD. The significances of differences between means were determined by unpaired Student's t-test. A p value of <0.05 was considered significant.
This work was supported by National Institutes of Health grants DK 49475 and AI 60994 to A.W.T. and by an American Society of Transplantation International Fellowship to M.A. The authors thank Amgen (Seattle, WA) for providing Flt3L, and Ms. Miriam Meade for excellent administrative support.