TGF-β signaling in dendritic cells is a prerequisite for the control of autoimmune encephalomyelitis

Abstract

One unresolved issue in immune tolerance is what prevents self-reactive T cells from activation. In this study, we used a transgenic mouse model of targeted functional inactivation of TGF-βR signaling in CD11c⁺ cells (CD11c^dnR mice) and showed a direct impact on the development of experimental autoimmune encephalomyelitis (EAE). We found that MOG_35–55 immunization of CD11c^dnR mice results in strong inflammation of CNS, high frequency of T cells in CNS, increased levels of T helper 1 (T_H1) and T_H17 cytokines in the periphery, and lack of remission from EAE. Once crossed with mice prone to autoimmunity, double-transgenic CD11c^dnRMog^TCR mice revealed a spontaneous EAE-like disease characterized by early infiltration of activated myelin-specific T cells into CNS, activation of microglial cells, inflammation of CNS, dysfunction of locomotion, and premature death. We constructed chimeric mice and demonstrated that inactivation of TGF-βR signaling in dendritic cells (DCs) results in augmented EAE-associated T cell responses. Our data provide direct evidence that TGF-β can control autoimmunity via actions on DCs.

Experimental autoimmune encephalomyelitis (EAE) serves as an animal model that recapitulates many features of multiple sclerosis (MS) (1). It can be induced by immunization of susceptible animals with a number of myelin antigens, including myelin basic protein (MBP) (2), proteolipid protein (3), and myelin oligodendrocyte glycoprotein (MOG) (4). A strong body of evidence points to a role for TGF-β in preventing EAE: first, the administration of exogenous TGF-β to mice or the pretreatment of MBP-specific T cells with TGF-β was able to prevent or inhibit EAE (5–7); second, the increased expression of TGF-β mRNA or protein was associated with remission of the disease (8, 9); third, the administration of neutralizing antibody to TGF-β enhanced the clinical severity of EAE (10); fourth, T helper 3 (T_H3) cells, known to secrete high amounts of TGF-β, were able to protect mice from EAE, whereas anti-TGF-β antibody treatment abrogated this protection (11); and finally, the administration of MBP via the nasal route induced TGF-β-producing regulatory T cells (Tregs) and prevented relapse of EAE in a rat model (12). Although these observations demonstrate a critical function for TGF-β in protection from EAE, the targets of TGF-β action remain undefined.

In this study, we have revisited the events leading to autoimmune encephalomyelitis in conditions where TGF-β receptor (TGF-βR) signaling has been functionally inactivated in CD11c⁺ cells. We used CD11c^dnR transgenic mice that were previously developed at Yale University to express a dominant negative form of TGF-βR type II under the control of the CD11c promoter (13). Our findings indicate that a lack of TGF-βR signaling in dendritic cells (DCs) caused severe symptoms of EAE disease in CD11c^dnR mice. Most importantly, it induced a spontaneous EAE-like disease in CD11c^dnRMog^TCR mice characterized by early infiltration of T cells into CNS, inflammation of CNS, motor dysfunction, and premature death. Together, these data provide direct evidence that TGF-β controls autoimmune encephalomyelitis via actions on the innate immune system.

Results

Motor Dysfunction and Premature Death in CD11c^dnRMog^TCR Mice.

CD11c^dnR and Mog^TCR mice were crossed, and their progeny were analyzed for behavioral changes. Two major changes were immediately apparent in CD11c^DNRMog^TCR mice: spontaneous motor dysfunction and premature death (Fig. 1).

Fig. 1.

Behavioral phenotype of CD11c^dnRMog^TCR mice. (A) Mice were observed during a 15-s tail suspension test. CD11c^dnRMog^TCR mice (Right) reacted by a clasping phenotype, whereas control mice (Left) exhibited a normal plantar reaction. (B) Motor performance of CD11c^dnRMog^TCR versus control mice was measured on a Rotorod set to accelerating mode. The assay was performed with n = 10 for the combined CD11c^dnR plus Mog^TCR control mice and n = 8 for CD11c^dnRMog^TCR double transgenic mice. Control Mog^TCR and CD11c^dnR mice did not differ when assaying their behavior (P > 0.05). (C) Results show survival curve of CD11c^dnRMog^TCR mice in comparison with control CD11c^dnR or Mog^TCR littermates.

The first indication of motor dysfunction was the observation of a clasping phenotype in CD11c^dnRMog^TCR mice (Fig. 1A). Clasping of the limbs is considered to be an early sign of neurological deficits (14). The test is performed by a tail suspension for 15 s under video camera to record abnormal movements. Any abnormal movements of the hindlimbs, forelimbs, or trunk in a dystonic fashion are then scored. Whereas control CD11c^dnR littermates displayed a typical plantar reaction, CD11c^dnRMog^TCR mice showed clasping of the hindlimbs and sometimes even exhibited a full-body clasp. To better assess motor performance of CD11c^dnRMog^TCR mice, an accelerating Rotarod test was used: unlike control mice, CD11c^dnRMog^TCR mice showed a locomotion deficiency (Fig. 1B) that correlated with degree of spontaneous paralysis. Videos are provided to illustrate the different symptoms observed in CD11c^dnRMog^TCR mice, including limp tail, motor dysfunction, and limb paralysis [supporting information (SI) Movies S1–S5] Although these signs do not progress to a moribund stage, premature death occurred in ≈90% of CD11c^dnRMog^TCR mice (Fig. 1C).

Spontaneous and Early Infiltration of T Cells in the CNS from CD11c^dnRMog^TCR Mice.

Consistent with autoimmune encephalomyelitis, two major differences emerged upon examination of CNS from CD11c^dnRMog^TCR mice: spontaneous infiltration of T cells and activation of resident microglial cells (Fig. 2).

Fig. 2.

Spontaneous infiltration of activated T cells in CNS from CD11c^dnRMog^TCR mice. (A) Results are from CNS isolated from 6-week-old mice. Contour plots (blue contours) show total distribution of CD45.2 versus CD11b. Gated CD4⁺TCRαβ⁺ cells (CD4⁺ T cells), DX5⁺TCRαβ⁻ cells (NK cells), or MHC class II-expressing cells are shown in red dots over CD45.2 versus CD11b distribution (blue contours). Gated I-A/I-E^high cells localize in the compartment expressing high levels of CD45.2, indicating a CNS-infiltrating DC type (green dots) and gated I-A/I-E^low cells fall in the compartment expressing low levels of CD45.2 consistent with a CNS-resident microglial cell type (yellow dots). (B) Frequency of each cell population described in A is represented as the mean ± SD (n = 12). Contour plot represents CD44 versus CD62L staining on gated CNS-infiltrating CD4⁺ T cells from CD11c^dnRMog^TCR mice. (C) Brains and spinal cords were processed separately from 6-week-old mice. Dot plots show the distribution of CD4 versus TCR-αβ in each organ. (D–F) Results are from spinal cords and spleens isolated from 2-week-old mice. Contour plots show the frequency of CD4⁺CD45.2⁺ cells (D), CD4⁺CD62L⁻ cells (E), and CD4⁺CD44⁺ cells (F). Data are representative of four (A and B) and two (C–F) independent experiments, with n = 3 pooled mice in each experiment.

Trafficking of immune cells to the CNS is under tight control. The blood–brain barrier serves to isolate the CNS from the circulatory system and to block cellular trafficking between them. Upon injury, however, the blood–brain barrier is disrupted, thereby allowing circulating leukocytes to invade CNS (15). CNS-infiltrating cells are commonly discriminated from CNS-resident cells based on their differential expression of CD45.2: CNS-resident cells are characterized by low levels of CD45.2 (CD45.2^low), whereas CNS-infiltrating leukocytes are identified by higher intensity of CD45.2 expression (CD45.2^high) (16). Virtually all cells isolated from control CNS (CD11c^dnR or Mog^TCR) showed low levels of CD45.2, consistent with a CNS-resident cell type. In contrast, a significant population of leukocytes (CD45.2^high) was found in the CNS from CD11c^dnRMog^TCR mice among which activated T cells (CD45.2^highCD4⁺TCRαβ⁺) were the most represented (Fig. 2 A and B). Consistent with an EAE-type infiltrate, T cells were found more abundant in spinal cord when compared with brain of CD11c^dnRMog^TCR mice (Fig. 2C). Finally, the frequency of infiltrating T cells was determined as a function of age, and results showed that spinal cords from CD11c^dnRMog^TCR mice were invaded by activated CD4⁺ T cells as early as 2 weeks of age (Fig. 2 D and E).

Activation of Microglial Cells and CNS Inflammation in CD11c^dnRMog^TCR Mice.

Microglial cells represent a population of CNS-resident cells endowed with antigen-presenting capacity upon their activation and up-regulation of MHC class II (16, 17). In the absence of activation, however, microglial cells express undetectable levels of MHC class II antigens consistent with a healthy condition of animals. We found that CNS from CD11c^dnRMog^TCR mice has a significant population of MHC class II-expressing CD11c⁺ cells (Fig. 3A). As expected, low levels of MHC class II and CD11c were detected in CNS from control mice consistent with a healthy condition (Fig. 3A). We next addressed whether these antigen-presenting cells (APCs) are conventional DCs that infiltrated the CNS in CD11c^dnRMog^TCR mice or CNS-resident microglial cells that up-regulated MHC class II upon activation. As shown in Fig. 2A, most MHC class II-expressing cells in CNS from CD11c^dnRMog^TCR mice were CD45.2^low, consistent with CNS-resident microglial cell origin. Finally, several proinflammatory cytokines were measured and levels of mRNA were determined. Data from CD11c^dnRMog^TCR mice provided direct evidence in support of CNS inflammation (Fig. 3B).

Fig. 3.

CNS inflammation in CD11c^dnRMog^TCR mice. (A) Double immunofluorescent staining for MHC class II (green signal) and CD11c (red signal) in tissue sections from brains and spinal cords (DAPI nuclear counterstain is shown in blue). Merged images are shown from spinal cord and brain white matter tracts. (Magnification: ×40.) (B) Total RNA was extracted from CNS harvested from mice between 5 and 8 weeks of age. Levels of cytokine mRNA were determined by quantitative PCR. Levels of iNOS, TNF-α, IL-6, and IL-1β are shown after normalization to levels of HPRT mRNA. Data represent the mean ± SD (n = 6).

Lack of TGF-βR Signaling in Innate Cells Causes Severe EAE in CD11c^dnR Mice.

EAE was elicited and disease was evaluated in CD11c^dnR mice (Fig. 4). Mice were immunized with MOG_35–55 peptide, and EAE was monitored daily as described (18). Although all C57BL/6 mice succumbed to EAE pathology, CD11c^dnR mice showed significantly more severe symptoms of the disease. No gender difference was noted in our study, and both female and male CD11c^dnR mice were equivalently more susceptible to EAE than control mice. Moreover, immunized CD11c^dnR mice did not remit from the disease like control mice did, and some mice even progressed to death (Fig. 4A). Using FACS analysis, we assessed the frequency of T cells in CNS and found five to eight times more abundant T cells in CNS from CD11c^dnR mice (Fig. 4B). Detection of T cells in CNS was further confirmed upon examination of brain tissue on day 20 postimmunization (Fig. 4C). Proinflammatory cytokines were analyzed, and levels of mRNA confirmed high levels of CNS inflammation consistent with severe EAE in CD11c^dnR mice (Fig. 4D).

Fig. 4.

Immunization of CD11c^dnR mice results in severe EAE. (A) EAE was elicited by immunization with MOG/CFA and disease symptoms were assessed by mean clinical score. (B, E, and F) On day 12, mice were killed, and draining lymph nodes and CNS tissues were harvested. Gates and numbers in B represent the frequency of T cells infiltrating the CNS. Cell suspensions from draining lymph nodes were restimulated in vitro with 50 μg of MOG peptide. Supernatants were harvested after 3 days and subjected to Mouse Cytokine Array (E) and ELISA (F) assays. (C) Immunohistochemical detection of CD4⁺ cells in cerebellum and brainstem regions of the brain on day 20. (D) Levels of mRNA of IFN-γ, TNF-α, and iNOS in brain tissues on day 20. Filled bars are CD11c^dnR mice, and open bars are CD11c^wt mice.

Severe EAE in CD11c^dnR Mice Is Associated with Increased T_H1 and T_H17 Cells.

Given the critical role of T_H17 cells in the development of autoimmune diseases (19, 20), we asked whether a lack of TGF-βR signaling in the innate system can restrict T cell polarization into T_H17. To investigate this possibility, mice were killed at peak of disease (day 12), and the profile of T cell cytokines was determined in draining lymph nodes after restimulation in vitro. Data from T cell cytokine array assay showed a dramatic increase in IFN-γ and IL-17 in draining lymph nodes isolated from immunized CD11c^dnR mice (Fig. 4E). Exact amounts of IFN-γ and IL-17 were determined by ELISA, and the results obtained were consistent with increased production of IFN-γ and IL-17 in CD11c^dnR mice (Fig. 4F). We investigated changes in other T cell functions; however, no significant differences were observed in T cell proliferation (Fig. S1) or T cell activation (Fig. S2) from CD11c^dnR mice. Although we cannot rule out differences in their functions, the frequency of regulatory T cells were similar among CD4⁺ T cell populations in both animal groups (Fig. S3).

Lack of TGF-βR Signaling in APC But Not NK Cells Is Responsible for Severe EAE in CD11c^dnR Mice.

The CD11c^dnR transgene is expressed in both DCs and NK cells (13). To distinguish the role of TGF-β on DCs versus NK cells in EAE, we used adoptive cell transfer strategy to restrict the transgene expression either to DCs or NK cells. Four groups of chimeric mice were generated (Fig. 5): Chimera 1 was constructed to have TGF-β-resistant DCs but normal NK cells; chimera 2 had normal DCs but TGF-β-resistant NK cells; in chimera 3 both DCs and NK cells were resistant to TGF-β; and chimera 4 was used as control (both DCs and NK cells from WT mice). We took advantage of the lack of NK cells in Rag^−/−γc^−/−-deficient mice (21, 22) and used them as recipient mice for NK cell transplantation to generate chimeras 2 and 4. In parallel, we crossed γc^−/− mice with CD11c^dnR mice, and CD11c^dnRγc^−/− mice lacking NK cells but having TGF-β-resistant DCs were used as recipients for NK cell transplantation to generate chimeras 1 and 3. In addition to NK cell transplantation, recipient mice received donor T cells, and a combination of 90% WT T cells + 10% Mog^TCR T cells was injected together with donor NK cells. Subsequently, EAE was induced and the frequency of T cells was determined in the CNS at peak of disease (Fig. 5). Results revealed approximately four times more abundant T cells in chimeras having TGF-β-resistant DCs (groups 1 and 3) when compared with control chimeras with TGF-β-sensitive DCs (group 4). Surprisingly, no difference was observed in chimeras with TGF-β-resistant NK cells (group 2). These data highlight the importance of the interplay of TGF-β signaling in DCs as a means to control the extent of T cell activation and prevent autoimmunity.

Fig. 5.

Lack of TGF-βR signaling in DCs is sufficient to promote massive T cell infiltration in CNS. Chimeras were constructed by using different combinations of NK cell adoptive transfers. Splenic NK cells were obtained from CD11c^dnR or control donor mice. NK cells together with a mixture of 90% WT T cells + 10% Mog^TCR T cells were i.v.-injected into Rag^−/−γc^−/− or CD11c^dnRRag^−/−γc^−/− recipient mice. After 4 weeks, mice were immunized with MOG/CFA and killed on day 12. Gates and numbers represent the frequency of T cells infiltrating CNS in each chimera group.

Lack of TGF-βR Signaling in Conventional APC But Not CNS-Resident APC Is Responsible for Severe EAE in CD11c^dnR Mice.

We asked whether severe EAE occurs as a result of a lack of TGF-βR signaling in conventional DCs or CNS-resident microglial cells. To address this issue we developed bone marrow (BM) chimeras in which transgene expression was restricted either to DCs or microglial cells. The model of radiation-BM chimeras is commonly used to selectively manipulate the genotype of the peripheral hematopoietic immune system from that of the CNS resident cells, given that microglial cells are radioresistant and are not repopulated after BM reconstitution (23, 24). To distinguish between donor-derived cells versus host cells, we used congenic donor and recipient mice. Two groups of BM chimeric mice were constructed in which a blockade of TGF-βR signaling was limited to the radioresistant (CNS) or radiosensitive (peripheral) compartment of the host (Fig. 6). We ensured >95% engraftment of donor BM cells, and cellular chimerism in CNS was assessed after immunization by using CD45.1 and CD45.2 congenic markers (Fig. S4). We found no effect of a blockade of TGF-βR signaling in CNS-resident microglial cells during the development of EAE. In contrast, chimeras with restricted blockade of TGF-βR signaling in conventional innate cells revealed severe symptoms of EAE that resulted in ≈70% of death within 36 days after MOG/CFA immunization (Fig. 6).

Fig. 6.

Lack of TGF-βR signaling in microglial cells has no effect on the susceptibility to EAE. Chimeras were constructed by BM cell transfers into lethally irradiated mice (1,000 rads). BM cells were obtained from CD11c^dnR or control donor mice, and 10 × 10⁶ cells were injected (i.v.) into irradiated CD11c^dnR or WT recipient mice. After 6 weeks, all mice were immunized with MOG/CFA. Data show the mean of clinical scores (Lower Left) and the survival curves (Lower Right).

Discussion

In the present study, we provide direct evidence that the interplay between TGF-β and DCs is a key event in the control of autoimmune encephalomyelitis. Although the role of TGF-β in preventing autoimmunity is indisputable, the issue of whether TGF-β can control autoimmunity via suppression of the innate immune system has not been directly investigated. One way to address this issue was by crossing CD11c^DNR mice having a blockade of TGF-βR signaling in innate cells (13) and MOG^TCR mice known to be predisposed to autoimmune encephalomyelitis (25). MOG^TCR mice were previously described to be predisposed to autoimmunity as 47% developed spontaneous optic neuritis but only 1–4% showed signs of spontaneous EAE (25). The goal from this crossing was to investigate how a lack of TGF-β signaling in innate cells can influence the development of autoimmunity in predisposed mice. Mice were crossed (CD11c^DNR × MOG^TCR) and double transgenic offspring (CD11c^DNRMOG^TCR) were observed and tested. Our results showed that the lack of TGF-β signaling in the innate system was sufficient to induce a severe and spontaneous autoimmune encephalomyelitis that led to death. This spontaneous disease resembled MS as indicated by (i) early infiltration of T cells into CNS, (ii) activation of microglial cells and inflammation in the CNS, and (iii) dysfunction of locomotion and premature death.

Given the wide expression of the CD11c^dnR transgene in the innate immune system of CD11c^dnRMog^TCR mice (13), two scenarios were considered: either TGF-β controls autoimmune encephalomyelitis via suppression of one particular innate cell type or several TGF-β-resistant innate cell subsets are needed to act in concert to provoke autoimmune encephalomyelitis. To restrict the expression of CD11c^dnR transgene to an individual innate cell subset, different combinations of adoptive cell transfers were used, and several groups of chimeric mice were generated for this study. We developed chimeras in which blockade of TGF-βR signaling was restricted to (i) microglial cells only, (ii) DCs only, or (iii) NK cells only. Induction of EAE in these chimeras showed no effects of a lack of TGF-βR signaling in NK cells or microglial cells. In contrast, a lack of TGF-βR signaling in DCs was sufficient for increased susceptibility to EAE, indicating the importance of the interplay of TGF-β and DCs in the control of autoimmune encephalomyelitis. In line with our data, myeloid DC emerged recently as a unique APC population capable of driving T_H17 differentiation in CNS in the relapsing EAE model (26).

Our finding that lack of suppression in the innate immune system results in spontaneous autoimmune encephalomyelitis is in support of an infectious trigger for MS. The idea that infection plays a role in the development of MS has prevailed for more than a century (27). The data supporting an infectious trigger for MS include an association of relapses with viral infection (28), a linkage of MS with latent viral infections (27), and lack of disease development in transgenic mice expressing a T cell receptor (TCR) for an encephalitogenic peptide when housed in a pathogen-free facility (29). Several hypotheses exist to explain how infection could activate myelin-specific T cells. In this regard, the major postulated hypothesis was attributed to a molecular mimicry mechanism predicting that structural similarities between T cell epitopes of microbial antigens and myelin proteins can lead to functional cross-reactivity at the level of TCR (30–32). In this scenario, the execution of this cross-reactivity will need the collaboration of three parameters: (i) the type of APC that initiates the response to the mimic peptide, (ii) the ability of a single TCR to functionally bind to different peptides; and (iii) the innate immune environment in which the presentation of mimic peptide to T cells occurs. If this is truly the case, then it is conceivable to think that when the innate immune environment is enriched with suppressive factors, this will lead to a prevention of autoimmune encephalomyelitis. In this regard, our data demonstrate a role of innate TGF-βR signaling in the prevention of autoimmunity.

Methods

Mice.

CD11c^dnR transgenic mice (13) were maintained on a C57BL/6 background. Mog^TCR transgenic mice (25) on a C57BL/6 background were obtained from Vijay K. Kuchroo (Harvard Medical School, Boston). Rag^−/−γc^−/−-deficient mice were obtained from Taconic. C57BL/6 and congenic CD45.1⁺ C57BL/6 mice were obtained from Jackson Laboratories. All animals were maintained in specific pathogen-free conditions. All animal experiments were performed under approved Institutional Animal Care and Use Committee protocols (Yale University).

EAE Induction.

EAE was elicited by s.c. tail-base injection of 50 μg of MOG_35–55 peptide in complete Freund's adjuvant (CFA) containing 500 μg of heat-inactivated Mycobacterium tuberculosis (Difco Labs). A dose of 200 ng of Bordetella pertussis toxin (LIST Biological Labs) was injected i.p. on days 0 and 2. Mice were monitored daily and scored as follows: 1, flaccid tail; 2, inability to right; 3, one hindlimb paralysis; 4, paralysis of both hind limbs; and 5, moribund.

Adoptive Transfers.

Splenic NK cells were obtained from CD11c^dnR or control donor mice. A total of 5 × 10⁶ of a mixture of donor NK and T cells was prepared in sterile PBS and transplanted (i.v.) into Rag^−/−γc^−/− or Rag^−/−γc^−/−CD11c^dnR recipient mice. After 4 weeks, recipient mice were immunized with MOG/CFA. For BM transplantation, CD11c^dnR and control mice were lethally irradiated (1,000 rads) and subsequently transplanted (i.v.) with 10 × 10⁶ donor BM cells obtained from CD11c^dnR or control mice. After 6 weeks, recipient mice were immunized with MOG/CFA.

In Vitro Recall Response Assay.

Spleens or draining lymph nodes were isolated on days 6 or 12 after immunization with MOG peptide. Cells were cultured for 3 days in the presence of the indicated amounts of MOG peptide. Supernatants were collected and subjected to analysis by ELISA (anti-IFN-γ from Pharmingen and anti-IL-17 from Biosciences) or a mouse cytokine array detection kit (RayBiotech).

Real-Time PCR.

Total RNA was isolated with TRIzol reagent (Invitrogen) and then treated with a DNA-free kit (Ambion). Five micrograms of mRNA was used in a reverse transcriptase reaction to produce cDNA, and 2 μl of cDNA was used for quantitative RT-PCR using an icycler (Applied Biosystems). Customized primers and probes for inducible NOS (iNOS), TNF-α, IL-6, IL-1β, and IL-17 were obtained from Applied Biosystems. Primers and Probes for IFN-γ, hypoxanthine guanine phosphoribosyltransferase (HPRT), and CD11c^dnR transgene have been described (13).

Histology.

CNS tissues were fixed in 4% paraformaldehyde overnight at 4°C, cryoprotected in a graded series of sucrose diluted in PBS (10% to 20% to 30%), and embedded in optimal cutting temperature compound (OCT Tissue-Tek™) for cryosectioning. Para-median sagittal sections were cut at 12 microns on a freezing microtome, applied to Superfrost Plus Gold slides (Fisher), and used for immunohistochemistry or immunofluorescence.

Behavioral Analyses.

Mice were suspended by their tails for 15 s and videotaped to record any abnormal movement. An abnormal movement was defined as dystonic movements of the hindlimbs or a combination of hindlimbs and forelimbs and trunk, during which the limbs were pulled toward the body. The Rotarod test was performed in accelerated mode. The acceleration was increased from 4 to 40 rpm over a period of 1 min, and time to fall was recorded. A total of three trials were conducted with a 30-min intertrial interval.

Statistical Analyses.

Means and SDs were calculated according to standard practice, and one-way ANOVA was performed to assess significance for Rotarod experiments. Survival analysis was performed using Kaplan-Meier survival curves. P < 0.05 were considered to be statistically significant, and analyses were conducted using the Statistical Package for the Social Sciences, release 13.0 (SPSS).