Potential conflict of interest: Nothing to report.
The Wnt/β-catenin pathway has been known to play a role in induction of immune tolerance, but its role in the induction and maintenance of natural killer T (NKT) cell anergy is unknown. We found that activation of the Wnt pathways in the liver microenvironment is important for induction of NKT cell anergy. We identified a number of stimuli triggering Wnt/β-catenin pathway activation, including exogenous NKT cell activator, glycolipid α-GalCer, and endogenous prostaglandin E2 (PGE2). Glycolipid α-GalCer treatment of mice induced the expression of wnt3a and wnt5a in the liver and subsequently resulted in a liver microenvironment that induced NKT cell anergy to α-GalCer restimulation. We also found that circulating PGE2 carried by nanoparticles is stable, and that these nanoparticles are A33+. A33+ is a marker of intestinal epithelial cells, which suggests that the nanoparticles are derived from the intestine. Mice treated with PGE2 associated with intestinal mucus-derived exosome-like nanoparticles (IDENs) induced NKT cell anergy. PGE2 treatment leads to activation of the Wnt/β-catenin pathway by inactivation of glycogen synthase kinase 3β of NKT cells. IDEN-associated PGE2 also induces NKT cell anergy through modification of the ability of dendritic cells to induce interleukin-12 and interferon-β in the context of both glycolipid presentation and Toll-like receptor–mediated pathways. Conclusion: These findings demonstrate that IDEN-associated PGE2 serves as an endogenous immune modulator between the liver and intestines and maintains liver NKT cell homeostasis. This finding has implications for development of NKT cell–based immunotherapies. (HEPATOLOGY 2013)
Unlike T cells, natural killer T (NKT) cells respond to lipid-based antigens including self and foreign glycolipid and phospholipid antigens1 presented by CD1d-restricted antigen-presenting cells (APCs). Among these lipid-based antigens, alpha-galactosylceramide (α-GalCer) is a synthetic glycosphingolipid derived from the marine sponge, Agelas mauritianus, and is commonly used in mice and human NKT studies as a potent activator of NKT cells in vivo or in vitro.2 A single injection of the exogenous α-GalCer leads to NKT cell activation followed, by long-term anergy, thereby limiting its therapeutic use.3 A number of potential endogenous glycolipids derived from dietary metabolic products and lipids derived from some intestinal bacteria migrate constantly into the liver,4-6 and these lipids can activate liver NKT cells in vitro.7 It is, therefore, remarkable that liver NKT cells are normally quiescent even though they are constantly exposed to intestinal-derived products. The molecular mechanisms that underlie induction of liver NKT cell anergy regulated by either exogenous α-GalCer or endogenous lipids are largely unknown.
The gut communicates extensively with the liver8 through a number of gut-derived molecules that are constantly migrating into the liver. Prostaglandin E2 (PGE2) and Wnt ligands are enriched in the gut, and whether they migrate into the liver and subsequently contribute to induction of liver NKT anergy has not been fully investigated. Both PGE29 and Wnt10 regulated pathways are known to play a crucial role in immune tolerance; however, a direct link between these two key pathways remains to be identified, although recent studies have proposed involvement of the Wnt pathway in regulating T cells11,12 and dendritic cell (DC)10 activation.
In this study, we demonstrate that preconditioning of the Wnt pathway for activation in the liver microenvironment is essential for induction of NKT cell anergy regardless of whether the NKT cells are naïve or educated. We further identified a previously undescribed mechanism in which the carriage of PGE2 by intestinal mucus-derived exosome-like nanoparticles (IDENs) into the liver created an environment in which activation of the Wnt/β-catenin pathway is induced.
NKT cells were enriched via negative magnetic sorting (Miltenyi Biotec) using anti-CD11b, B220, CD8α, Gr-1, CD62L, and CD11c antibodies. Enriched NKT cells (5 × 106 per mouse) were then injected intravenously into irradiated nonobese diabetic (NOD)–severe combined immunodeficient (SCID) mice. In some cases, NK1.1+CD5+ surface stained cells (NKT) were sorted using a FACSVantage. Sorted NKT cells were 85%-90% pure as determined by tetramer staining. To determine the effects of the liver microenvironment created by Wnt signaling on liver NKT cells, Tcf/LEF1-reporter mice as recipients were treated with α-GalCer (3 μg; Avanti Polar Lipids, Inc., Birmingham, AL) or lithium chloride (LiCl) (200 mg/kg; Sigma) every 3 days for 12 days. Recipients were then irradiated (750 rads) before intravenously administering enriched NKT cells (10 × 106 per mouse) from C57BL/6 CD45.1+ mice. Twenty-four hours after cell transfer, the mice were injected intravenously with α-GalCer (5 μg/mouse).
Details of other methods used in this study are described in the Supporting Information.
Wnt Activation Creates a Microenvironment that Causes NKT Cell Anergy to α-GalCer Stimulation In Vitro and In Vivo.
We first tested whether activation of Wnt/β-catenin modulates the activity of liver NKT cells. Sorted liver NKT cells that were transfected with constitutively activated β-catenin (Ctnnb1) exhibited a reduction in α-GalCer tetramer-stimulated NKT cell proliferation (Fig. 1A) and production of interferon (IFN)-γ and interleukin (IL)-4 (Fig. 1B). Because of this result, we tested whether the wnt/β-catenin pathway was activated when mice are treated with α-GalCer. We found that a single injection of α-GalCer caused an increase in β-catenin/Tcf/LEF1 signaling throughout the liver of mice, as indicated by β-galactosidase activity. Multiple injections of α-GalCer resulted in much stronger β-catenin/Tcf/LEF1 signaling than a single injection (Fig. 2A). The α-GalCer treatment modulated the levels of expression of genes encoding multiple components of the Wnt pathway, including LEF1, Tcf7, Wnt5a, Wnt3a, Wnt7a, Wnt7b, and Wnt10a, as determined by quantitative real-time polymerase chain reaction (RT-PCR) (Supporting Fig. 1). The results from both enzyme-linked immunosorbent assay (ELISA) (Fig. 2B, left panel) and western blot analysis (Fig. 2B, right panel) indicate that a significant increase in the levels of Wnt5a and Wnt3a proteins and β-catenin protein was observed during liver recovery following α-GalCer restimulation. Since the α-GalCer stimulation or α-GalCer/α-GalCer restimulation did not affect β-catenin transcription (Supporting Fig. 1), the α-GalCer stimulation most likely influences the levels of β-catenin through nontranscriptional mechanisms. To determine whether activation of Wnt signaling occurs after α-GalCer stimulation in vivo, hepatocytes were isolated and cultured with liver leukocytes. Repeated addition of α-GalCer resulted in higher expression of Wnt5a and Wnt3a (Fig. 2C). In addition, treatment of a stably transfected, Tcf-driven green fluorescent protein (GFP) NKT hybridoma with an α-GalCer tetramer led to an increase in the numbers of GFP+ cells, with restimulation leading to a further increase in the numbers of GFP+ cells (Fig. 2D). Consistent with the fluorescence-activated cell sorting (FACS) analysis data, treatment of the NKT hybridoma with α-GalCer tetramer resulted in enhancement of phosphorylation of β-catenin and glycogen synthase kinase 3β (GSK3β) (Fig. 2E) as well as induction of expression of the genes encoding Wnt5a and Axin2 (Fig. 2F). Unlike the induction of expression of the genes encoding Wnt5a and Axin2, restimulation was required for induction of the genes encoding Cbl-b, Grail, and Itch (Fig. 2F), which are known to be associated with the development of the anergic state after stable expression of β-catenin in T cells.13–15 Furthermore, knockdown of LEF1 in the NKT hybridoma that led to a partial reversing of the α-GalCer restimulation did not elicit IL-2 production (Supporting Fig. 2), suggesting that LEF1 is a critical transcriptional factor that regulates α-GalCer mediated anergy of NKT cells.
To directly assess whether the liver microenvironment created by α-GalCer stimulation has functional significance in the induction of NKT cell anergy, we used an adoptive transfer approach in which NKT cells from naïve mice were transferred into γ-irradiated Tcf/LEF1-reporter mice that had been preinjected with α-GalCer, LiCl, or vehicle (phosphate-buffered saline [PBS]) as a control. The reconstituted mice were then injected with α-GalCer. Staining of liver sections from the Tcf/LEF1-reporter mice for β-galactosidase showed that Wnt signaling in the liver is indeed activated by α-GalCer or LiCl treatment (Fig. 3A). The production of IFN-γ and IL-4 was significantly lower in the recipient mice that had been pretreated with α-GalCer or LiCl than in PBS-treated animals (Fig. 3B). The α-GalCer or LiCl treatments had a similar effect on the production of IL-12 (Fig. 3B), suggesting that anergy of IL-12–producing cells, including DCs, is induced. To confirm whether the anergy of adoptive transferred NKT cells from naïve mice is induced, NKT cells were recovered from the liver of α-GalCer– or LiCl-treated recipient mice and cocultured with liver DCs from naïve mice. NKT cells reisolated from the livers of recipient mice pretreated with α-GalCer or LiCl were not responsive to treatment with α-GalCer ex vivo (Fig. 3C), whereas the NKT cells reisolated from the livers of recipient mice pretreated with vehicle control were responsive. To confirm these findings, we also adoptively transferred NKT cells recovered from vehicle-, α-GalCer–, or LiCi-injected mice to Rag1KO mice. The production of IFN-γ and IL-4 was significantly lower in the Rag1KO mice that received NKT cells from α-GalCer or LiCl-treated mice when compared with Rag1KO mice that received NKT cells from PBS-treated mice (Fig. 3D). Collectively, these results indicate that the liver microenvironment plays a key role in the induction of anergy in NKT cells, because irrespective of whether the NKT cells had been pretreated with α-GalCer, they lost their responsiveness to α-GalCer as long as a Wnt-enriched liver environment was established.
PGE2 Plays a Role in the Induction of Liver NKT Cell Anergy by Activation of Wnt/β-Catenin.
PGE2 is known to modulate Wnt/β-catenin activity in cancer cells. However, the role of PGE2 in terms of regulating NKT cell activation is not known. Treatment of wild-type NKT cells with PGE2 led to increased phosphorylation of β-catenin, as well as prompting phosphorylation of GSK3β in a time-dependent manner (Fig. 4A) and induction of the expression of the genes encoding the Wnt ligands Wnt5a and Axin2 (Fig. 4B). Interestingly, treatment with PGE2 together with α-GalCer had a synergistic effect on the induction of expression of the gene encoding Wnt 5a (Fig. 4B).
To further determine whether PGE2-mediated inactivation of GSK3β plays an essential role in induction of NKT cell anergy, NKT cells were treated with two GSK3β-specific inhibitors. Liver NKT cells were stimulated with α-GalCer in thiadiazolidinone, a non–adenosine triphosphate (ATP) competitive inhibitor of GSK3β that binds to the active site of GSK3β. The addition of the non-ATP competitive GSK3β inhibitor suppressed the production of IFN-γ and IL-4 in α-GalCer–stimulated NKT cells (Fig. 4C). Similar results were obtained upon treatment of NKT cells with another non-ATP competitive GSK3β inhibitor, LiCl (Fig. 4C). Treatment of the α-GalCer–stimulated NKT cells with Wnt3a or Wnt5a ligands resulted in effects similar to those obtained for treatment of NKT cells with GSK3β inhibitors (Fig. 4D). Reduced IFN-γ production was not due to an intrinsic defect, because there was no difference in the intracellular-stained IFN-γ in the PMA and ionomycin-stimulated NKT cells that had been treated with PGE2 or PGE2 vehicle (Supporting Fig. 3). Furthermore, blocking of the canonical Wnt activation pathways using IWR1 or IWP2 led to a partial reversal of PGE2-mediated inhibition of production of IFN-γ and IL-4 (Fig. 4E). The results from RT-PCR analysis also indicate that the PGE2-mediated reduction in IFN-γ and IL-4 correlated with the inhibition of expression of T-bet and GATA-3 (Fig. 4F), key transcription factors regulating IFN-γ and IL-4 expression and maturation of NKT cells.16,17 Furthermore, knockdown of LEF1 in the NKT hybridoma led to partial reversing of PGE2-mediated inhibition of IL-2 production (Supporting Fig. 4), suggesting that LEF1 is a critical transcriptional factor that regulates PGE2 mediated anergy of NKT cells. Collectively, PGE2 stimulation led to activation of Wnt/β-catenin and subsequently the induction of NKT cell anergy.
IDEN-Associated PGE2 Plays a Role in Induction of Liver NKT Cell Anergy.
Exosome-like nanoparticles have a high capacity for binding PGE218 and maintaining its stability and thus activity. ELISA analysis of circulating exosome-like nanoparticles indicates that the circulating nanoparticles carry PGE2 (Supporting Fig. 5). FACS analysis of these circulating nanoparticles further indicates that they are also A33+ (Fig. 5A). A33+ is an intestinal epithelial marker, suggesting that these PGE2+ nanoparticles are derived from the intestine.
Nanosized particles in the gut migrate into the liver,19,20 where the majority of the NKT cells reside. We tested whether IDENs can induce liver NKT cell anergy. The results from electron microscopy examination showed that they are nanoparticles in size (Fig. 5B). The nanoparticles were enriched for PGE2 (Supporting Fig. 5). We then tested whether IDEN-associated PGE2 plays a role in the induction of NKT cell anergy. NKT cells were purified from the livers of mice that had been administered IDENs or vehicle intravenously. NKT cells were cocultured in vitro with DCs from the livers of untreated mice in the presence of α-GalCer. The results show that the NKT cells purified from the mice that had been administered IDENs had significantly lower production of both IFN-γ and IL-4 of NKT cells to α-GalCer stimulation (Fig. 5C). Liver NKT cells pretreated with circulating exosomes also produce less IFN-γ and IL-4 in response to α-GalCer stimulation (Supporting Fig. 6), suggesting that IDEN-PGE2–mediated induction of NKT cell anergy is physiologically relevant.
To further determine whether the IDEN-associated PGE2 played a role in the induction of NKT cell anergy, mice were treated with indomethacin, a cyclo-oxygenase 2 inhibitor that blocks the generation of PGE2. The effects of IDENs isolated from indomethacin-treated mice on the induction of NKT cell anergy were then evaluated. Indomethacin treatment reduced significantly the amounts of PGE2 associated with IDENs (Supporting Fig. 5), which ultimately led to the attenuation of IDEN-mediated anergy induction in NKT cells to α-GalCer stimulation (Fig. 5D).
To further demonstrate the role of IDEN-mediated anergy induction of liver NKT cells in vivo, we used a concanavalin A (ConA)-induced hepatitis mouse model. ConA-induced hepatitis is dependent on NKT cell activity.21 Using an adoptive transfer approach, we found, as expected, that SCID mice that received liver NKT cells from PBS-treated mice exhibited typical liver necrosis (Fig. 6A) and elevation of serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) (Fig. 6B). In marked contrast, the histologic evidence of necrosis in the liver (Fig. 6A) and the ConA-induced elevation of the levels of serum ALT and AST (Fig. 6B) were reduced significantly in the SCID mice that received liver NKT cells from IDEN-treated mice, indicating that IDENs have a direct effect on induction of anergy in liver NKT cells.
IDENs Also Induce NKT Cell Anergy in the Context of Both Glycolipid Presentation and Toll-Like Receptor–Mediated Pathways.
APCs play an essential role in liver NKT cell activation by presenting lipid-related antigen on an APC CD1d molecule–dependent and independent manner.22 First, we tested whether DCs take up IDENs. Both CD11c+ DCs and F4/80+ macrophages from the livers of naïve mice took up IDENs rapidly (as early as 2 hours) and continued to take up IDENs over 24 hours (Fig. 7A, Supporting Fig. 7). We then tested whether IDEN treatment has an effect on NKT cell activation of mice. FACS analysis of liver leukocytes suggested that the expression of MHCII and CD86 by the DCs (Fig. 7B) was reduced in mice that had been administered IDENs. On coculture of DCs purified from the livers of mice that had been administered IDENs or vehicle with carboxyfluorescein succinimidyl ester–labeled liver NKT cells isolated from mice that had been administered IDENs or vehicle in the presence of α-GalCer, DCs from the livers of mice that had been administered IDENs exhibited a reduced ability to stimulate the proliferation of NKT cells, regardless of the source of the NKT cells (Fig. 7C). RT-PCR analysis further indicated that the expression of IL-12 and tumor necrosis factor α, which are critical for activation of DCs and DC-mediated activation of NKT cells, was significantly lower in DCs sorted from the livers of α-GalCer–injected mice that had been administered IDENs (Fig. 7D). Additionally, the levels of the immunosuppressive cytokine IL-10 were higher. DCs also can activate NKT cells in a CD1d-independent manner through Toll-like receptor (TLR)-induced release of soluble mediators, including IL-12 and type I IFNs.23,24 We found that IDENs suppressed the expression of IL-12 and IFN-β in TLR-stimulated DCs (Fig. 7E) and that IFN-γ release was reduced greatly when TLR ligand–treated DCs were cocultured with liver NKT cells in the presence of IDENs (Fig. 7F). Thus, IDENs can also induce NKT cell anergy through modification of the ability of DCs to stimulate NKT cell anergy in the context of both glycolipid presentation and TLR-mediated pathways. To further determine whether IDEN-associated PGE2 plays a role in the inhibition of production of IL-12, the effects of IDENs isolated from indomethacin-treated mice on the production of TLR-stimulated DCs was evaluated. ELISA results (Fig. 7G) indicated that IDENs isolated from indomethacin-treated mice had a reduced potency to inhibit IL-12 production, suggesting that IDEN-associated PGE2 plays a role in inhibiting the production of IL-12 of TLR-stimulated DCs.
We further tested whether IDEN-mediated Wnt/β-catenin activation also plays a role in TLR-stimulated DC activation. Bone marrow–derived dendritic cells (BMDCs) from B6.Cg-Tg(BAT-lacZ)3Picc/J mice have a remarkable increase in β-galactosidase activity, where the β-galactosidase gene expression is driven by a Tcf/LEF1 promoter in the presence of IDENs (Fig. 7H).
Paradoxically, IDEN treatment also led to a reduction in IL-12 from BMDCs stimulated with TLR ligands as listed in Fig. 7I. The addition of the canonical Wnt inhibitors IWR1 and IWP2 led to the partial reversing of IDEN-mediated inhibition of IL-12 production (Fig. 7J). Collectively, these results suggest that IDEN-mediated activation of the Wnt pathway in DCs also plays a role in induction of NKT cell anergy.
The results presented in this study suggest a model (Fig. 8) in which PGE2 or α-GalCer stimulation leads to induction and release of Wnt ligands in the liver, where NKT cells reside. NKT Wnt signaling activation mediated by PGE2, α-GalCer induced Wnt ligands, or PGE2 via inactivation of GSK3β (a β-catenin inactivator) eventually activate β-catenin/LEF1-mediated transcriptional machinery, which causes induction of NKT cell anergy. Alternatively, PGE2-mediated activation of the Wnt/β-catenin pathway in DCs leads to prevention of the TLR stimuli–induced production of IL-12 that is required for CD1d-independent NKT cell activation. In this study, we provide for the first time evidence that activation of the Wnt/β-catenin pathway leads to anergy of NKT cells. We also provide evidence for PGE2 cross-talk with the Wnt signaling pathway, which can occur through regulation of β-catenin/GSK3β activity. The evidence for PGE2 cross-talk with the Wnt signaling pathway is consistent with the literature in which a role for PGE2 in regulation of Wnt signaling at the level of β-catenin stability has been demonstrated in zebrafish hematopoietic stem cells.25
It has also been reported that some factors, through ubiquitin-mediated proteasome degradation, may induce NKT cell anergy.26 The inhibition of α-GalCer–induced phosphorylation of ERK tyrosine kinase in NKT cells plays a role in the induction of NKT cell anergy.27 Lacking costimulatory signals and cytokines provided by DCs may also lead to NKT cell anergy.28–30 Whether these factors also cross-talk with the PGE2/Wnt/β-catenin we identified in this study and lead to NKT cell anergy, will require further investigation to discern.
Recent studies suggest that exosome-like nanoparticles play a critical role in cell-to-cell communication.31,32 Intestinal epithelial cells are known to release nanosized microvesicles,33,34 and the nanoparticles have been shown to migrate into the liver.19,35 Our study further demonstrates that the gastrointestinal tract communicates directly with the liver via IDENs that carry PGE2. PGE2 carried by IDENs has at least two unique characteristics in comparison with the free form of PGE2. First, the stability of PGE2 carried by IDENs is increased significantly, as shown by the fact that the half-life of PGE2 is approximately 30 seconds in the circulator system,36,37 and intravenous injection of chemically synthesized PGE2 did not have any effect on the induction of IFN-γ and IL-4 of mice treated with α-GalCer (data not shown) and therefore could have no effect on the activation of Wnt signaling in NKT cells. Besides the stability of PGE2 regulated by the local balance between the COX2-driven synthesis and 15-hydroxyprostaglandin dehydrogenase–mediated degradation of PGE2,38,39 in this study, we demonstrated that the amount of PGE2 carried by IDENs is also associated with the potency of induction of liver NKT cell anergy (Supporting Figs. 5 and 6). It is conceivable that the factors regulating the amount available and the affinity of IDEN binding to PGE2 may also contribute to PGE2-mediated Wnt signaling. The role of ceramide40 and others factors that affect COX2/15-hydroxyprostaglandin dehydrogenase–mediated PGE2 synthesis and degradation warrants further study. In addition, factors regulating gut permeability which are critical factors in regulating the amount of nanoparticle trafficking from the gut to the liver41–43 needs further study to. Caution should be exercised when drawing conclusions regarding the biological effect of PGE2 on IDENs. Effects on the Wnt signaling pathway may be different when comparing PGE2 on IDENs to that of free form of PGE2, since microRNAs and other lipids are packed in the IDENs and may also contribute to the PGE2-mediated Wnt signaling pathway. Identifying whether IDEN microRNAs and/or lipids have a role in PGE2-mediated Wnt signaling pathway needs further study. Second, PGE2 carried by IDENs induces anergy of NKT cells not only through direct targeting of NKT cells but also through DC activation via a TLR-mediated pathway. The finding that IDENs can carry a number of therapeutic agents44 and target APCs may provide an avenue to pursue IDEN modulation of APC function and their role in gut immune tolerance.
These findings also open up a new avenue for investigating further the possible role of IDENs carrying other molecules released in gut that could induce both gut and liver immune tolerance. Furthermore, from therapeutic standard point, IDENs from intestines of other species may also be a useful vehicle for delivering therapeutic reagents44,45 to treat gastrointestinal diseases as well as diseases such as liver diseases treated by oral administration.
In this study, the finding that IDEN-PGE2 activated the Wnt pathway and suppressed cytokine expression via inactivation of the GSK3/β-catenin pathway raises a number of important questions that need to be addressed in future studies. PGE2 binds and activates four G-protein–coupled E receptors (EP1-EP4). Unlike EP1 and EP3, EP2 and EP4 have been shown to activate the GSK3/β-catenin pathway, as well as the adenylate cyclase-triggered cyclic adenosine monophosphate (cAMP)/protein kinase A (PKA)/exchange protein directly activated by cAMP pathway.46–48 Whether IDEN-PGE2 also suppresses IFN-γ and IL-4 expression via cAMP/PKA/cAMP responsive element binding protein (CREB)-dependent pathway is unknown. If IDEN-PGE2 does suppress cytokine production, also it needs to be determined if there is cross-talk with the cAMP/PKA/CREB pathway at unidentified points to ultimately regulate the production these cytokines.
Finally, the Wnt signaling pathway is known to play a crucial role in the prevention of autoimmune responses and in promotion of tumor growth. PGE2 is a potent signaling molecule that regulates immune tolerance and promotes tumor growth in addition to having a role in hematopoiesis, regulation of blood flow, renal filtration and blood pressure, regulation of mucosal integrity, and vascular permeability.49 Our findings provide a basis for further studies regarding the biological effects of PGE2 cross-talk with the Wnt/β-catenin pathway on these systems as well.
We thank the National Institutes of Health Tetramer Facility for providing PBS-57 ligand complexed to CD1d monomers or tetramers and Mitchell Kronenberg, who provided the NKT1.2 hybridomas. We also thank Jerald Ainsworth and Fiona Hunter for editorial assistance.