|Home | About | Journals | Submit | Contact Us | Français|
Dendritic cells (DCs) appear in higher numbers within the CNS as a consequence of inflammation associated with autoimmune disorders, such as multiple sclerosis (MS), but the contribution of these cells to the outcome of disease is not yet clear. Here we show that stimulatory or tolerogenic functional states of intracerebral DCs regulate the systemic activation of neuroantigen-specific T cells, the recruitment of these cells into the CNS and the onset and progression of experimental autoimmune encephalomyelitis (EAE). Intracerebral microinjection of stimulatory DCs exacerbated the onset and clinical course of EAE, accompanied with an early T-cell infiltration and a decreased proportion of regulatory FoxP3-expressing cells in the brain. In contrast, the intracerebral microinjection of DCs modified by tumor necrosis factor alpha (TNF-α) induced their tolerogenic functional state and delayed or prevented EAE onset. This triggered the generation of interleukin 10 (IL-10)-producing neuroantigen-specific lymphocytes in the periphery and restricted IL-17 production in the CNS. Our findings suggest that DCs are a rate-limiting factor for neuroinflammation.
Since the first discovery of DCs in the CNS (Matyszak and Perry, 1996), these cells have emerged as pivotal players in the development and maintenance of CNS autoimmunity and inflammation (reviewed in (Becher et al., 2006; McMahon et al., 2006). DCs are rarely detected in the healthy CNS, and when they are present, they localize to vascular-rich tissues including the meninges and choroid plexus (Matyszak and Perry, 1996; Hanly and Petito, 1998; McMenamin, 1999; Serot et al., 2000; Greenwood et al., 2003). Several studies have demonstrated a substantial accumulation of DCs in the brain and spinal cord in response to local inflammation induced by autoimmunity, infection, or trauma (Hanly and Petito, 1998; McMenamin, 1999; Suter et al., 2003; McMahon et al., 2005; Newman et al., 2005; Bailey et al., 2007). The mechanism(s) by which DCs accumulate in the CNS under inflammatory conditions are not well understood.
The concordance of i) accumulation of DCs in the cerebrospinal fluid during progression of EAE, the animal model of MS (Fischer and Bielinsky, 1999; Aloisi et al., 2000; Fischer et al., 2000; Fischer and Reichmann, 2001; Juedes and Ruddle, 2001; Kivisakk et al., 2004), ii) the localization of DCs at the proximity of inflamed microvessels in MS lesions (Serafini et al., 2006), and iii) the production of astrocyte-derived chemokines that promote recruitment of DCs into the CNS (Ambrosini et al., 2005) strongly suggest that brain microvessel endothelial cells regulate the recruitment of DCs into the CNS. We have previously demonstrated that the transmigration of DCs across brain microvessel endothelium is regulated by macrophage inflammatory protein-1α (MIP-1α), matrix metalloproteinases (MMPs), and occludin perturbation. In addition, transmigration of DCs across brain microvessel endothelial cell monolayers contributed to the activation of antigen-specific T cells in vitro (Zozulya et al., 2007).
Conflicting data exist concerning the contribution of DCs to the outcome of CNS inflammation. It was proposed that DCs inhibit T-cell responses in the CNS (Suter et al., 2003), thus leading to protection from EAE (Kleindienst et al., 2005). However, other data suggests that DCs contribute to the induction and maintenance of neuroinflammation in EAE (Dittel et al., 1999; Weir et al., 2002). For example, increasing the number of DCs in the brain by systemic injection of FMS-like tyrosine kinase 3 ligand (Flt-3L) leads to a substantial increase in the severity of clinical EAE symptoms (Greter et al., 2005). Conversely, inhibition of Flt-3L signaling ameliorates EAE, providing further evidence that DC numbers in the brain correlate with the outcome of autoimmune responses (Whartenby et al., 2005). In addition, CNS-resident F4/80−CD11c+CD45high cells isolated from brains of animals experiencing relapsing EAE (R-EAE) or Theiler's murine encephalomyelitis virus-induced demyelinating disease (TMEV-IDD) can efficiently present endogenous myelin proteolipid protein (PLP) antigen and activate naïve PLP139–151-specific T cells in vitro (McMahon et al., 2005). Further supporting a stimulatory role for DCs in regulating CNS immune responses, DCs were recently shown to be the only CNS antigen-presenting cells (APC) population capable of inducing memory cytotoxic T-cell responses in lymphocytic choriomeningitis virus (LCMV) infection (Lauterbach et al., 2006).
Taking advantage of methods used to generate stimulatory or inhibitory tolerogenic DCs that can be injected intracerebrally, we addressed the role of functionally different DCs on the generation of neuroantigen-specific T-cell responses and clinical outcome of EAE. Our data demonstrate that the quantity and functional phenotypes of DCs in the brain regulate the onset and progression of EAE.
Four to six week old female C57BL/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). 2D2 transgenic mice were a gift from Dr. V. Kuchroo (Harvard Medical School, USA) (Bettelli et al., 2003). DEREG-transgenic mice were a gift from Dr. T. Sparwasser (University of Muenchen) (Lahl et al., 2007). Experimental animals were housed in a pathogen-free facility at the University of Wisconsin, Medical School Animal Care Unit under guidelines of the National Institutes of Health or at the University of Wuerzburg Animal care facility according to German guidelines for animal care. Protocols for animal use were approved by the Animal Care and Use Committees of the University of Wisconsin-Madison and University of Wuerzburg (Regierung von Unterfranken).
For intracerebral injection, the mice were anesthetized by intraperitoneal (i.p.) injection of a ketamine (90 mg/kg) – xylazine (10 mg/kg) mixture. Dendritic cells (2.5×105) loaded or unloaded with myelin oligodendrocyte glycoprotein peptide 35–55 (MOG35–55) (10 µg/ml CyberSyn, Lenni, PA) in 20µl of PBS, or an equal volume of PBS was injected into the right frontal lobe with an insulin syringe attached to a penetrating depth controller as previously described (Ling et al., 2003; Ling et al., 2006). The injection was restricted to the ventral-posterior region of the frontal lobe, and the penetrating depth of the syringe was 1.55 mm from the surface of the brain. For each intracerebral injection, the solution was injected slowly, and then the syringe was held in place for an additional minute to reduce backfilling of injected solution.
In some experiments, 5×106 2D2 transgenic T cells were adoptively transferred into mice (Sewell et al., 2003). Where indicated, DCs were incubated in the presence of 200 ng/ml pertussis toxin (PTX) (List Biological Laboratories, Sacramento, CA) during antigen pulsing (Marriott et al., 1999). PTX-treated DCs were extensively washed before injection. No significant effect of PTX on DC phenotype, maturation and function was observed in vitro (Karman et al., 2004a and data not shown).
For EAE induction, emulsion of equal volumes of CFA (5 mg/ml) and 200 µg MOG35–55 supplemented with M. tuberculosis (Strain H37Ra, Difco, Detroit, MI) were injected subcutaneously in the scapular region of each mouse. PTX (400 ng/mouse) was i.p. injected on the days 0 and 2 relative to immunization. Clinical scores were monitored daily in a blind manner and recorded as follows: 0, no clinical disease; 1, flaccid tail; 2, gait disturbance or hind limb weakness; 3, hind limb paralysis and no weight bearing on hind limbs; 4, hind limb and forelimb paralysis and reduced ability to move around the cage; and 5, moribund or dead. The mean daily clinical score and standard error of the mean were calculated for each group. The significance of differences was calculated by Student’s t and Wilcox tests as described by Fleming et al. (Fleming et al., 2005).
DCs were generated as previously described (Inaba et al., 1992; Karman et al., 2004a). Briefly, bone marrow obtained from femurs and tibias of C57BL/6 mice was washed and plated in 24-well plates in RPMI 1640 with 10% FBS supplemented with 100 U/ml penicillin/streptomycin and 20 ng/ml GM-CSF. GM-CSF was titrated from supernatants of the GM-CSF-secreting X63 cell line (gift from Dr. A. Erdei, Eotvos University, Budapest, Hungary). Seven days following GM-CSF cultures, the nonadherent and loosely adherent cells were removed and re-plated in the absence of GM-CSF and cultured together for 4 hrs with or without MOG35–55 peptide (10 µg/ml). Nonadherent cells were collected for use as previously described (Karman et al., 2004a). To generate semi-matured or fully-matured DCs, cells were treated for 4 hrs with tumor necrosis factor (TNF)-α (500 U/ml; TEBU/PeproTech, Germany) or lipopolisaccharide (LPS) (10 µg/ml), with or without MOG35–55 as described by Menges et al. (Menges et al., 2002).
Spleen and lymph nodes were dissected, weighed, and transferred into cold HBSS (Cellgro, Herndon, VA). The isolated lymphocytes were washed with cold HBSS and resuspended either in staining buffer (PBS containing 1% BSA and 0.1 % NaN3) for a direct cell-surface staining, or in culture medium (RPMI supplemented with 10 %FBS) for an overnight cell culture followed by intracellular cytokine staining. For spleens, red blood cells were lysed with Tris-NH4Cl. Minced brain tissue was processed with Medicon inserts (Becton Dickinson, San Jose, CA). Brain lymphocytes were isolated from the interface of a percoll density gradient as described previously (Sewell et al., 2004). For some samples, the total number of isolated cells per gram of tissue was calculated.
Single-cell suspensions were stained with saturating concentrations of antibodies at 4°C for 30 min. Monoclonal antibodies used for cell surface staining were purchased from BD Biosciences (San Jose, CA) and included anti-CD8, anti-CD4, anti-vβ11, anti-LFA-1, anti-CD45, anti-CD11b, and anti-FoxP3. Non-specific binding to cell surface Fc receptors was blocked with unlabeled FcγRII/FcγRIII-specific antibody (clone 2.4G2) as previously described (Karman et al., 2004a). For intracellular staining of IFN–γ, single-cell suspensions from spleen (106 cells/ml) or brain lymphocyte preparation (105 cells/ml) were cultured for 12 hrs in 96-well plates in RPMI-10% FBS with or without MOG35–55 (10 µg/ml) in the presence of GolgiStop (1 µl/ml) (BD Biosciences) before being stained for extracellular ligands and intracellular cytokines. In some experiments, the supernatants of re-stimulated cells were assessed by cytokine bead array for measurement of IFN-γ, IL-17, and IL-10 according to the manufacturer’s instruction (Bender MedSystems, Wien, Austria). Stained cells were acquired on a four-color FACSCalibur cytometer and were analyzed with FlowJo software (TreeStar, NY) version 7.2.1.
Single-cell suspensions from the spleen of a 2D2 mouse were incubated with carboxyfluorescein succinimidyl ester (CFSE) (2.5 µM; Molecular Probes, Eugene, OR) in HBSS for 5 min at 37°C. The reaction was quenched with 20% FBS. Cells were processed for adoptive transfer into C57BL/6 mice (5×106 cells in 100 µL of PBS per mouse).
Differences between groups were determined with unpaired Mann-Whitney and Wilcox tests. P-values less than 0.05 were considered to be significant.
The accumulation of activated neuroantigen-specific T cells in the nervous tissue is a hallmark of autoimmune diseases of the CNS. Although the brain parenchyma is usually devoid of DCs, several recent studies have demonstrated that these cells are prominent components of CNS infiltrates in EAE and MS (Fischer and Reichmann, 2001; Suter et al., 2003; Greter et al., 2005; Bailey et al., 2007; Deshpande et al., 2007). To determine whether increasing the number of stimulatory DCs in the brain induces antigen-specific immune responses in the periphery, we injected MOG35–55 peptide-presenting DCs intracerebrally (i.c.) (i.c. DCMOG) into C57BL/6 mice and subsequently analyzed the proliferation of MOG-specific (2D2) T cells in different organs 5 days following their adoptive transfer. We have previously shown that i.c. microinjection of either ovalbumin (OVA) (Ling et al., 2006) or OVA-loaded DCs (Karman et al., 2004a) initiated antigen-specific T-cell responses in the cervical lymph nodes (CLN), spleen and the homing of OVA-specific T cells into the brain (Karman et al., 2004a). To extend these studies, we addressed whether the spatial localization of DCs in the brain is important for inducing neuroantigen-specific T-cell responses in the brain and CLNs. CFSE-labeled CD4+ T cells from 2D2 mice, expressing a transgenic (Tg) T-cell receptor specific for MOG35–55 peptide and defined by Vα3.2 and Vβ11+ subunit chains (Bettelli et al., 2003), were adoptively transferred into three groups of C57BL/6 animals (Fig. 1 A, upper panel). These groups included i.c. PBS-, i.c. DCMOG-, and intravenously (i.v.) DCMOG-injected mice (Fig. 1 A, left, middle, and right of lower panel, respectively). Five days post-immunization we analyzed the accumulation and activation of CFSE-labeled 2D2 Tg CD4+ T cells in the brain, CLNs and spleen. Our data show that i.c. but not systemic injection of DCs resulted in antigen-specific T-cell accumulation in the brain. Based on high expression of LFA-1, antigen-specific CFSE-labeled T cells in the brain were highly activated (Fig. 1 A, top center of lower panel). As expected, MOG-specific T cells were stimulated in the CLN and spleen, where they underwent several cycles of proliferation prior to their accumulation in the brain (Fig. 1 A, bottom of lower panel, arrows). Only i.c. injection of DCMOG induced the proliferation of MOG-specific T cells in the CLN and spleen and the accumulation of neuroantigen-specific T cells in the CNS. In contrast, i.v. injection of DCMOG resulted in 2D2 T-cell proliferation in the spleen but not in the CLN, and no accumulation of cells was found in the brain (Fig. 1 A, most right column of lower panel). As a control for injection-induced microtrauma, we injected the same volume of sterile PBS intracerebrally. Importantly, PBS injection alone did not induce proliferation of antigen-specific T cells in the periphery or the accumulation of these cells in the brain (Fig. 1 A, far left plots). No MOG-specific T-cell accumulation in the CNS was observed upon i.c. injection of DCs pulsed with ovalbumin (i.c. DCOVA) or other MHCII-restricted peptides (e.g. pigeon cytochrome c, PCC) (data not shown). Since we previously demonstrated that only the relevant antigen-expressing DCs induce antigen-specific T-cell accumulation in the CNS (Karman et al., 2004a) (Karman et al. 2006), irrelevant peptides, such as OVA-expressing DCs were not included in further studies. Analysis of the absolute number of tissue infiltrating CD4+Vβ11+LFA-1highCFSElow cells further confirmed that activated antigen-specific CD4+ T cells were accumulating in the brain when i.c. injected antigen-presenting DCs were present (Fig. 1 B and C).
It was previously shown that MOG-specific FoxP3+ regulatory T (Treg) cells accumulate in the CNS during EAE (Korn et al., 2007; O'Connor et al., 2007). To determine whether i.c. DCs could also induce the accumulation of antigen-specific Treg cells in the brain, we intracerebrally injected 2D2 Tg mice with MOG-pulsed DCs and tracked the appearance of both MOG-specific and non-specific Treg cells in the CNS and peripheral tissues (Fig. 2, upper panel). We also confirmed our previous results showing that i.c. DCMOG injection induces Vβ11+CD4+ MOG-specific T-cell accumulation in the brain (Fig. 2, top right panel of lower panel). Intracerebral DCMOG injection resulted in approximately five times higher infiltration of MOG-specific Vβ11+CD4+ T cells in the brain compared to the i.c. PBS injections (Fig. 2, top left panel). Unpulsed DCs were used as a control to test whether the observed effects depended on the accessibility of MOG antigen to DCs. Interestingly, we detected two times higher infiltration of MOG-specific Vβ11+CD4+ T cells in the brain following the injection of unpulsed DCs compared to i.c. PBS injections but this was clearly less compared to i.c. DCMOG injection (Fig. 2, top center panel). This may indicate that DCs can uptake neuroantigens in the brain and induce the accumulation of antigen-specific T cells in the CNS. Irrespective of the antigen, i.c. DCs induced the presence of FoxP3-expressing CD4+ T cells with the prevalence of MOG-specific Vβ11+FoxP3+ Treg cells. Although, approximately 2% of MOG-specific Treg cells were found in the CLN in all groups of animals (Fig. 2, middle panel), no MOG-specific Treg cells were detected in the spleen following i.c. injections (Fig. 2, lower panel). These experiments demonstrate that i.c. DCs elicit the accumulation of both encephalitogenic and regulatory T cells in the CNS. Altogether, these data suggest that antigen-carrying DCs in the brain contribute to the neuroantigen-specific T-cell accumulation in the CNS.
When we used transgenic mice termed "depletion of regulatory T-cell" (DEREG) mice (Lahl et al., 2007) in which FoxP3+ cells can easily be detected and followed for tissue distribution based on GFP expression regardless of their antigen specificity, we did not observe any differences in the fraction of GFP+CD4+FoxP3-expressing Treg cells in response to different types of DCs or PBS injections in the periphery (Supplementary Fig. 1). However, higher percentages of CNS-derived GFP+ cells were detectable in i.c.-injected animals with no differences between i.c. PBS and i.c. DCMOG injected groups compared to i.v. DCMOG. This suggests that microtrauma induced by i.c. injection of PBS or DCs was not sufficient to induce the accumulation of antigen-specific T cells in the brain, but attracts Treg cells that can be found in the brain five days post injection (Supplementary Fig. 1).
In figure 1 we demonstrated that i.c. injected DCs that carry MOG antigen contribute to the activation of MOG-specific T cells in the periphery and the accumulation of these cells in the CNS. From our previous studies, we also learned that in addition to presenting antigen in the CNS, DCs can deliver antigens from inflamed CNS and induce homing of peripheral antigen-specific T cells into nervous tissue (Karman et al., 2004b; Ling et al., 2006). To further understand the in vivo significance of these results, we extended our work to analyze whether i.c. injected DCs loaded with MOG peptide would influence the clinical course or onset of EAE. In these experiments, we also used DCs that were pulsed with MHC class II-specific OVA peptide (DCOVA) in parallel with DCMOG. Thus, DCMOG, DCOVA, or equal volume of sterile PBS was i.c. injected into the CNS, and five days later EAE was actively induced by immunization with MOG35–55 antigen (Fig. 3 A, upper panel). Mice i.c. injected with DCMOG experienced a significantly accelerated EAE onset and a more progressive course of the disease as compared to i.c. PBS and i.c. DCOVA injected animals (Fig. 3 A, lower panel and Table 1). Accordingly, the frequency of CD4+LFA-1+ double positive IFN–γ-producing cells in the brain of i.c. DCOVA-injected mice (3.6%) was significantly lower at 7 days post EAE induction compared to i.c. DCMOG (16.7%)-injected mice (Fig. 3 B). Also, a higher percentage of activated LFA-1+CD4+ T cells were observed in i.c. DCMOG (approximately 80%) compared to i.c. DCOVA (37%) and i.c. PBS (23.5%) injected mice. Still, modest acceleration of EAE clinical scores (Fig. 3 A) and cellular infiltration (Fig. 3B) was observed in response to i.c. DCOVA injections. This might suggest that OVA-pulsed and i.c. injected DCs could amplify neuroinflammation by contributing to CNS antigen-specific T-cell activation. In support of this, myeloid DCs were recently described as a superior CNS cell population in EAE mice, capable of inducing naïve CD4+ T-cell proliferation and cytokine production with endogenous peptides (Bailey et al., 2007) (and reviewed in (Miller et al., 2007b)).
To test whether the induction of neuroantigen-specific T cells by i.c. DCMOG in conjunction with CFA is sufficient to induce clinical EAE, one group of i.c. injected animals (i.c. DCMOG) received only subcutaneous CFA and no MOG antigen during EAE induction (Table 1). These mice did not develop clinical signs of EAE, indicating that subcutaneously injected MOG antigen is still required for full susceptibility to clinical disease (Table 1).
Graphical representation of EAE day onset for i.c. DCMOG immunized animals compared to i.c. DCOVA and i.c. PBS injected mice confirms an earlier onset of disease (Fig. 3 C). Likewise, graphical representation of the disease severity in number of days with EAE clinical scores equal or higher than 2.5 from all animals involved in EAE experiments (Table 1) shows a higher number of mice with EAE score of 2.5 or above in i.c. DCMOG compared to i.c. PBS and i.c. DCOVA injected groups (Fig. 3 D). A significantly increased disease severity was also observed between i.c. PBS and i.c. DCOVA-injected animals, proving DC contribution to CNS-triggered neuroinflammation (Fig. 3 D). Immunohistopathology was performed in addition to clinical EAE scoring, and cellular infiltration was analyzed in the animal cohorts described above. On day 17 following EAE induction, immunohistopathology directly correlated with the EAE development in all experimental groups and was followed by a higher cellular infiltration and demyelination degree at lesion sites in the optic nerves (Supplementary Fig. 2), brain and spinal cord tissues (not shown) in i.c. DCMOG and i.c. DCOVA compared to i.c. PBS-injected animals. These data show that injected DCs contribute to cellular infiltration and demyelination in the CNS.
The i.c. DCOVA group in the last set of experiments has helped us conclude that antigen-specific T cells do not accumulate in the CNS in the absence of their specific antigen and neuroinflammation, and that i.c. injections of PBS or DCs (pulsed or not pulsed with antigen) induces the same level of microtrauma in the brain that does not result in the accumulation of antigen-specific cells in this tissue.
We next wanted to determine whether the accumulation of T cells in the brain following i.c. DCMOG injection and EAE induction correlates with clinical disease (Fig. 4, experimental scheme). Our data show that i.c. DCMOG injection resulted in earlier accumulation of CD4+ and also CD8+ T cells in the brain (preclinical phase; Fig. 4 A). At day 14 following EAE induction (peak of disease), differences in T-cell accumulation between i.c. PBS and i.c. DCMOG were not statistically significant (Fig. 4 B). These data indicate that increasing the number of MOG-presenting DCs in the CNS induces early immune responses in the periphery, and the accelerated accumulation of T cells in the brain contributes to the amplification of neuroinflammation during early stages of the EAE clinical course.
To address the question whether the migration of i.c. injected DCs out of the brain would be required to influence EAE disease onset, we inhibited the migratory capacity of DCs with pertussis toxin (PTX), blocking chemokine receptor signalling in vitro. Following in vitro treatment, PTX pre-treated DCs were injected into the brain. In figure 5 we show that pre-treatment of DCMOG prior to their i.c. injection does not exacerbate the onset of EAE (no significant differences between i.c. PBS versus i.c. DCMOG(PTX) (Fig. 5 A and B). The i.c. DCMOG injection led to a significantly accelerated EAE clinical onset compared to both i.c. PBS and i.c. DCMOG(PTX)-groups (Fig. 5 A and B). Furthermore, the disease severity was significantly enhanced if i.c. PBS and i.c. DCMOG groups were compared, however, we did not find statistical differences between i.c. DCMOG and i.c. DCMOG(PTX)-groups (Fig. 5 C).
To further define the mechanisms by which intracerebral DCs regulate the early development and severity of clinical EAE symptoms, we studied the kinetics and frequency of IFN-γ-producing, MOG-specific peripheral T cells. Intracerebral delivery of DCMOG induced significantly higher numbers of IFN-γ-secreting, MOG-reactive CD4+ T cells in the peripheral lymphoid organs (spleen, Fig. 6 and lymph nodes (not shown)). These cells were detected earlier in the course of EAE compared to i.c. PBS injected animals (as measured at day 7 post EAE induction, Fig. 6 A) and also remained significantly higher at the peak of EAE (day 14, Fig. 6 B). We also detected the frequency of IL-17-producing CD4+ T cells in the peripheral immune organs, which appeared to be slightly elevated in the i.c. DCMOG compared to the i.c. PBS group at earlier time points of EAE (day 7, Fig. 6 A) and significantly higher at later time points (day 14, Fig. 6 B). Interestingly, IFN-γ-producing CD4+ T cells from cultured splenocytes isolated at different time points post EAE induction (day 7 and day 14) could also be detected in media without MOG re-stimulation from i.c. DCMOG, but not from i.c. PBS injected animals. In the clinical phase of EAE (day 14), double positive IFN-γ+IL–17+ CD4+ T cells could also be detected in i.c. DCMOG but not in i.c. PBS group (Fig.6 B).
Growing evidence indicates a bi-directional interaction between Treg cells and DCs (Tang et al., 2006; Tang and Bluestone, 2006). We therefore speculated that the impact of intracerebral DC modulation on CNS autoinflammation might be reflected in a change in the absolute numbers or in the ratio between encephalitogenic and regulatory CD4+ T cells in the CNS. Lymphocytes isolated from non-inflamed (naïve) brains contained only a few CD4+ T cells that predominantly exhibited the FoxP3high regulatory phenotype (Fig. 7 A). The expression of FoxP3 on CD4+CD25high cells was clearly decreased in the brain of i.c. DCMOG injected animals (Fig. 7 B, lower panel) compared to i.c. PBS injected control group (Fig. 6 B, upper panel). This data shows that in parallel with early EAE onset and higher accumulation of MOG-specific T cells after i.c. DCMOG injection, the frequency of FoxP3high cells in the brain is decreased and i.c. DCMOG alters the ratio between encephalitogenic and regulatory T cells.
It has been proposed that CNS DCs have a dual role during EAE, as they could provide stimulatory or suppressive signals at different stages of disease (Deshpande et al., 2007). Modifying DC phenotypes with TNF-α elicits the generation of IL-10-producing T cells and leads to antigen-specific prevention of EAE (Menges et al., 2002), and reviewed in (Steinman et al., 2003)). We therefore asked the question whether the functional state of intracerebral DCs is decisive for the qualitative and quantitative outcome of experimental CNS inflammation. We injected MOG-pulsed and TNF-α− or LPS-treated (DCTNF-α or DCLPS) semi-mature DCs i.c. and subsequently induced EAE. Among other microbial products, LPS treatment results in an upregulation of maturation markers and an increase in the ability of DCs to stimulate T cells [reviewed by R. Sousa (Reis e Sousa, 2006)]. Intracerebrally injected DCLPS significantly exacerbated EAE onset and increased EAE severity in comparison to i.c. PBS-injected animals (Fig. 8 and Table 2). Intracerebral injection of DCTNF-α beneficially modulated EAE in comparison to the i.c. DCLPS injected group. Protection ranged from delaying disease onset to full prevention of clinical symptoms (only 35% disease incidence) (Fig. 8 and Table 2). Onset of EAE clinical symptoms was on average 3–4 days later in i.c. DCTNF-α and i.c. PBS injected mice in comparison to i.c. DCLPS injected animals. Also, the severity of disease post-onset was lower in i.c. DCTNF-α compared to i.c. DCLPS injected groups. To assess the mechanism of how i.c. DCTNF-α injections promoted EAE protection, we measured the frequency and function of peripheral and CNS infiltrating immune cells in i.c. DCTNF-α−, i.c. DCLPS− and i.c. PBS-injected mice. Although clinical disease was delayed or EAE was ameliorated (Fig. 8 and Table 2), i.c. DCTNF-α injected EAE mice did not have decreased numbers of CNS infiltrating CD4+ and CD8+ T cells at any time points (Fig. 9 A and B). The ratio of encephalitogenic to regulatory T cells was higher in i.c. DCLPS and i.c DCTNF-α groups compared to i.c. PBS-injected mice, similar to that observed between i.c. DCMOG and i.c. PBS-injected groups (Fig. 7 and data not shown). We observed a higher accumulation of CD45+CD11bhigh macrophages in i.c. DCLPS injected mice and similar subsets of CD45+CD11bhigh macrophages and CD45+CD11blow microglia in i.c. PBS and i.c. DCTNF-α groups (Fig. 9 A). Similar to i.c. DCMOG injected animals (Fig. 4), CD4+ T cells persisted at later stages of EAE (day 17) in i.c. DCTNF-α and i.c. DCLPS injected groups in comparison to i.c. PBS controls (Fig. 9 B). To find factors that may be responsible for earlier EAE onset (i.c. DCLPS) or a delay in EAE onset (i.c. DCTNF-α) at earlier time points of disease, we measured the cytokines produced both in the CNS and periphery at different stages of EAE. Notably, lymphocytes isolated from the CNS of i.c. DCTNF-α and i.c. DCLPS-injected mice produced comparable levels of IL–10 at earlier (day 7) and later (day 17) time points, which were higher in comparison to i.c. PBS injected animals (Fig. 9 C and D, left panels). Although there was no IFN-γ signal detected in the CNS for i.c. PBS injected animals, 962 pg/ml of IFN-γ was detected in i.c. DCLPS group, which was ten fold higher than the concentration of IFN-γ in i.c. DCTNF-α mice (95 pg/ml) before EAE onset (day 7 post EAE induction). At this point, i.c. DCTNF-α completely protected both the CNS compartment and peripheral immune organs from IL-17-producing cells (26 pg/ml and 263 pg/ml, respectively) compared to the i.c. DCLPS injected group (1235 pg/ml and 10910 pg/ml, respectively) (Fig. 9 C). At later time points of EAE (day 17), the i.c. DCLPS injected group contained cells with a phenotype polarized towards Th-17 profile with 1350 pg/ml of IL-17 in the CNS and 7841 pg/ml of IL-17-producing cells in the periphery (spleen), which was two fold higher than IL-17 production detected in i.c. DCTNF-α and i.c. PBS injected mice in the CNS and periphery (Fig. 9 D). However, a great majority of CNS lymphocytes in i.c. DCTNF-α injected mice included IFN-γ-producing cells (2468 pg/ml of IFN-γ in i.c. DCTNF−α compared to 962 pg/ml in i.c. DCLPS mice), suggesting a Th1 phenotype in this group at later time points of EAE (d17). At the same time, i.c. DCTNF-α injections induced a higher amount of peripheral IL-10 production after both 7 and 17 days post EAE induction compared to the i.c. DCLPS group (Fig. 9 C and D).
These data indicate that TNF-α-treated, MOG-pulsed DCs injection into the brain induces a predominantly IL-10 dominated peripheral immune response that restrains Th17-mediated pathology during EAE.
DCs have emerged as pivotal regulators of CNS autoimmune and inflammatory responses. Taking advantage of the targeted i.c. microinjection technology, combined with methods to generate stimulatory or inhibitory tolerogenic DCs, we have analyzed the role of functionally different DCs on the generation of neuroantigen-specific immune responses and their relevance for modulating experimental neuroinflammation using the mouse model of MOG35–55-induced EAE.
Our results demonstrate that immunogenic DCs pulsed with MOG antigen and injected into the brain (i.c. DCMOG) activated naïve antigen-specific T cells in peripheral immune organs and promoted CNS invasion of neuroantigen-specific CD4+ T cells. However, the i.c. DCMOG did not induce EAE if injected without active immunization. This implies that peripheral induction for the disease is necessary. In this model, a mixture of MOG and CFA needs to be delivered subcutaneously in order to induce EAE, and in this case the i.c. DCMOG accelerated EAE initiation. Taken together, this may suggest that peripheral antigen presentation and immunization are necessary and sufficient to induce EAE, and that i.c. DCMOG can only modify the kinetics of disease. Thus, the main role of DCs in the brain appears to be in the effector phase of the immune response, rather than in the priming phase. The increasing number of MOG-loaded stimulatory DCs in the brain also accelerated the onset of EAE disease and resulted in exacerbated clinical severity and the extent of the disease. This was accompanied by an early infiltration of IFN-γ-producing MOG-reactive CD4+ T cells from the periphery into the CNS and a decrease in the proportion of FoxP3-expressing CD4+ regulatory T cells, suggesting a shift toward the enrichment of encephalitogenic T cells. In contrast, i.c. injection of TNF-α-treated and MOG-loaded tolerogenic dendritic cells (i.c. DCTNF-α) prior to EAE induction prevented clinical signs of EAE disease (35% disease incidence) or delayed EAE onset, followed by decreased IL-17 production in the CNS and increased level of IL-10-producing peripheral and CNS CD4+ T cells.
Recently, several studies have contributed to the notion that antigen-presenting cells in the brain play a key role in determining the outcome of CNS inflammation (Greter et al., 2005; McMahon et al., 2005; Bailey et al., 2007). Under inflammatory conditions (e.g. autoimmunity and infectious disease) DCs accumulate in the CNS parenchyma, suggesting that CNS DCs are critical for the initiation, regulation, and/or maintenance of immune responses in the CNS. Although numerous studies unambiguously emphasize the potential relevance of DCs for CNS immune surveillance or autoimmune reactions, the true contribution of DCs in the initiation and perpetuation of neuroantigen-specific T-cell responses remains elusive. Our work tested the relevance and impact of CNS-derived DCs, by taking advantage of a DC delivery method into the CNS. We found that the activation and CNS accumulation of neuroantigen-specific T cells critically depended on the route of MOG antigen delivery by DCs and the site of DC origin. Local brain DCs are likely to provide critical signals for attracting antigen-specific CD4+ and CD8+T cells into the inflamed CNS (Carson et al., 1999; Tang and Cyster, 1999). A combination of different cytokines, chemokines and inflammatory mediators might be responsible for the regulation of local inflammatory responses via intracerebral DCs. At the same time, several chemokines such as MIP-1α (CCL3), MIP-3β (CCL20), MCP-1 (CCL2), and RANTES (CCL5) are produced in the CNS during acute and chronic EAE (Serafini et al., 2000; Bailey et al., 2007). Defined by chemokine responsiveness, DC populations can differentially migrate to CNS compartments (Bailey et al., 2007). Myeloid DCs were recently shown to preferentially accumulate in the perivascular inflammatory foci of the spinal cord and cerebellum, clustering there with T cells at the peak of EAE (Bailey et al., 2007). The establishment of such DC-T and DC-B cell clusters in the CNS may be necessary to sustain intrathecal clonal expansion of T cells and production of anti-myelin specific antibodies as detected in EAE and MS lesions.
Spatial CNS localization of MOG antigen-pulsed DCs is critical in the accumulation of antigen-specific T cells in the brain. To further strengthen the importance of specific antigen presentation by intracerebral DCs, we performed experiments directly in 2D2 mice. We show that i.c. DCMOG injection significantly increased MOG-specific CD4+Vβ11+ T-cell accumulation in the CNS (4.2% vs 21.15%). Unpulsed DC injection into the brain induced a detectable accumulation of neuroantigen-specific cells in the brain under these experimental conditions, which might be due to the intrinsic ability of intracerebral DCs to pick up local CNS antigen and present it to naïve T cells, resulting in their expansion.
Severity of EAE as well as the number of MS plaques seem to correlate with the presence and function of DCs (Pashenkov et al., 2001; Greter et al., 2005; Serafini et al., 2006). One of the most important observations from our studies is that the amount of DCs in the brain could be a limiting factor in CNS autoimmune diseases. Our data show that in the early phase of EAE autoimmune disease, the number of DCs in the brain is critical, and this is a rate-limiting factor for the development of the disease. This and other studies (Greter et al., 2005; McMahon et al., 2005) collectively suggest that CNS-associated DCs are capable of inducing pathogenetically relevant T-cell responses locally in the brain.
Some data indicate that regulatory T cells are critical in the maintenance of immune privilege status of the CNS (Korn et al., 2007; O'Connor et al., 2007). We investigated the influence of i.c. DCMOG injections on FoxP3 Treg accumulation in the CNS. Although the naïve brain per se contained a very low number of CD4+ T cells, the majority of these cells expressed high levels of FoxP3. This might indicate that the high level of FoxP3 cells in the brain could be a part of the mechanisms maintaining an “immunological privileged” milieu of the CNS. Our experiments show that i.c. DC injections induce the CNS-accumulation of FoxP3 negative neuroantigen-specific T cells, which occurs at the expense of FoxP3 Treg cells. Thus, immunogenic DCs clearly alter the ratio of encephalitogenic to regulatory T cells. Whether regulatory T cells play a critical role in modifying CNS autoimmunity needs to be further studied (Zozulya, 2008).
DCs with regulatory or tolerogenic properties are capable of attenuating EAE (reviewed by (Miller et al., 2007a)). It was previously demonstrated that DCMOG matured with TNF-α and systemically injected into mice prior to EAE induction induced antigen-specific protection from EAE in mice (Menges et al., 2002). Here we show that semi-mature DCs have a protective effect on subsequent CNS-directed autoimmune responses, emphasizing the notion that the functional state of DCs has a clear impact on the quantity and quality of CNS-directed immune responses. The mechanism of EAE-attenuation by tolerogenic TNF-α-matured DCs is noteworthy. Strikingly, absolute numbers of CNS-infiltrating immune cells are equally high in animals receiving i.c. injections of DCTNF-α or DCLPS. Although we did not test the production of cytokines by CNS DCs in this study, it is well known that multiple immune mediators can be produced by DCs. It was proposed that DCs can polarize CD4+ T cells to become either Th17 or Th1 cells, producing IL-17 or IFN-γ, respectively. Tolerogenic or stimulatory DCs influence this polarization differently (Shortman and Liu, 2002; Shortman and Naik, 2007). Thus, IL-17 polarized CNS T cells with pathological function in the relapsing EAE model and their clustering with myeloid DCs in the CNS indicate the latter as the only CNS APC population that biased antigen-specific T cells toward a Th17 profile (Bailey et al., 2007). Endogenously collecting CNS antigen and highly producing IL-6 and TGF-β, DCs polarized Th17 responses which could be an intrinsic property of myeloid DCs or the CNS environment (Bailey et al., 2007; Miller et al., 2007b). Our data show that i.c. injection of DCTNF-α completely protects the CNS from infiltration of IL-17-producing cells but promotes IL-10 cells both in the CNS and in the periphery at early time points of EAE. This supports the idea that CNS DCs, depending on their functional state, can accelerate IL-17 appearance, leading to earlier EAE onset or restrict IL-17 generation in the target organ, thus delaying the disease.
In summary, our study shows that intracerebral DCs are capable of mounting CNS-specific T-cell responses and, in contrast to peripheral DCs, of inducing specific accumulation of neuroantigen-specific T cells in the brain. The quantity of stimulatory DCs in the CNS is a rate-limiting factor for the onset of subsequent EAE. Furthermore, the functional state of intracerebral DCs is decisive for the outcome of a subsequent autoimmune CNS inflammation: the presence of tolerogenic DCs in the brain protects from early development of EAE clinical signs by inducing IL-10 and restricting CNS-infiltrating IL-17 cells. Intracerebral DCs can therefore be considered as a crucial immune cell population during CNS-specific immune responses and can change the outcome of autoimmune inflammatory CNS disorders. This has clear implications for understanding the pathogenesis of MS and considering DCs as potential targets for future therapies.
Regulatory T cells accumulate in the CNS only in response to i.c. injections. Intracerebral injections of PBS and DCMOG and systemic injection of DCMOG (i.v.) were used to track Treg cell population in DEREG mice based on GFP expression on CD4+ T cells. Five days post injection mononuclear cells were isolated from the brain, CLN and spleen and directly analyzed by flow cytometry. Results are representative of two independent experiments.
Intracerebral DCMOG injection induces a stronger demyelination and enhanced cellular infiltration in comparison to i.c. PBS and i.c. DCOVA-injected animals. Mononuclear cellular infiltrates associated with demyelination (LFB) (upper panel) and H&E stains lower panel) in optic nerves taken from EAE mice with EAE scores of 2 or higher at the peak of EAE (day 17). Red squares indicate higher magnifications of indicated area. Visible sites of infiltration (arrows) and demyelination (asterisks) in the optic nerve tissue are shown.
We thank Dr. Dana C. Baiu, Melissa G. Harris and Heidi A. Schreiber for helpful discussions and critical review of the manuscript; K. Macvilay and B. Reuter for technical assistance. This work was supported by National Institutes of Health grant RO1-NS 37570-01A2 to Z.F. and BMBF 01gz0707, TPA8, and SFB581 to H.W.