γ-Aminobutyric Acid Transporter 1 Negatively Regulates T Cell-Mediated Immune Responses and Ameliorates Autoimmune Inflammation in the CNS¹

Multiple sclerosis (MS)³ is a chronic, progressive, degenerative disorder that affects nerve fibers in the CNS. It causes significant disability for most patients, with symptoms including spasticity, dystonia, tremor, ataxia, weakness, sensory loss, and so forth (1, 2). Its pathogenesis involves MHC class II-restricted neuroantigen-specific CD4⁺ T lymphocyte-mediated immune responses, initiating a detrimental autoimmune cascade leading to white matter damage. Experimental autoimmune encephalomyelitis (EAE) is the widely used animal model of MS. It can be induced in susceptible animals by immunization with myelin Ags (3). Studies of transgenic or gene-deficient mice have established critical pathogenic functions of proinflammatory cytokines (e.g., TNF-α, IFN-γ, IL-6, IL-17, IL-23) (4, 5, 6, 7, 8) and transcription factors (e.g., NF-κB, T-bet, STAT1) in EAE (9, 10). These molecules orchestrate a pathogenic cascade resulting in inflammation, demyelination, and axonal damage in the CNS.

Neurotransmitters provide the molecular basis for integrated, bidirectionally coordinated neuroimmune responses to homeostatic disturbance induced by stress, inflammation, or infection (11). Previous studies have demonstrated that manipulating the expression of neurotransmitters and their transporters or receptors can affect the development of MS/EAE (12, 13, 14, 15, 16, 17). For example, high levels of glutamate produced by activated immune cells and the agonist to 5-hydroxytryptamine receptors are associated with enhanced neuron death, damage to myelin-producing cells, and the development of EAE (15, 16). In contrast, acetylcholine degradation inhibition and cannabinoid or vasoactive intestinal peptide administration ameliorate MS/EAE development (13, 14, 17).

γ-Aminobutyric acid (GABA) is the major inhibitory neurotransmitter balanced with glutamate in the CNS. Previous studies have shown that the concentration of GABA and the activity of glutamate decarboxylase (GAD) in blood are reduced in EAE or MS (18, 19). Additionally, the uptake of [³H]GABA by synaptosomes in spinal cord of EAE is decreased (19). Moreover, agonists to GABA could lessen spasticity and improve acquired pendular nystagmus, which are common manifestations of MS (20).

After release from neurons, some GABA molecules combine with GABA receptors in the postsynaptic membrane to mediate inhibitory signals, while most GABA molecules undergo re-uptake by GABA transporters (GATs) in the presynaptic membrane (21). Therefore, GATs are critical in balancing GABA levels in the synaptic cleft as well as in maintaining a GABA reservoir in the body. All four GATs identified in the mammalian CNS belong to the Na⁺- and Cl⁻-dependent transporter family SLC6 (22). GAT-1 (23) is highly expressed in olfactory bulb, neocortex, cerebellum, superior colliculus, and substantia nigra, where it is found predominantly in axons, presynaptic terminals, and glial cells (21). Given its wide expression in the CNS, the roles of GAT-1 in modulation of CNS normal function have been extensively studied and identified, but its involvement in CNS inflammation and immune-mediated disease is poorly understood. In our study with EAE, we found an interesting phenomenon, namely that the expression levels of GAT-1 in spinal cord were much suppressed at the peak of the disease. Thus, we explored further the role of GAT-1 in EAE.

Upon inducing EAE in GAT-1-deficient mice, we found that the induced disease was aggravated. At the same time, mice deficient in GAT-1 developed some nonclassic EAE signs, including tremor, abnormal gait, and excitement. Additionally, mononuclear cells (MNCs) from these mice showed a higher Ag-specific proliferative response. Furthermore, the production of proinflammatory cytokines, including TNF-α, IFN-γ, IL-12, IL-6, IL-17, and IL-23, was significantly increased. In these mice, we also observed enhanced Ag-triggered NF-κB-DNA binding activity, elevated IκB kinase (IKK) triggered degradation of IκBα, and strengthened T-bet/STAT1 circuit signaling pathway. In vitro studies further indicated that GAT-1 was expressed on Ag-stimulated T cells, but not on B cells, and macrophages. Our findings support the notion that, besides being a critical neurotransmitter regulator, GAT-1 is also an important modulator in Ag-specific T cell responses and EAE development, thus bridging the interactions between the nervous system and the immune system.

GAT-1-null mutant mice (B6 × 129 F₂) were generated as previously described (24). Briefly, GAT-1 knockout heterozygotes (GAT-1^+/−) were crossed with C57BL/6 mice for two generations. Then, the heterozygotes were selected for sib mating to create wild-type (WT) mice (GAT-1^+/+) and homozygous mice (GAT-1^−/−) for experiments. C57BL/6 mice were purchased from The Jackson Laboratory. All experimental procedures have been reviewed and approved by the Institutional Laboratory Animal Care and Use Committees.

For active EAE induction, mice were immunized s.c. on the flank with 200 μl inoculum containing 300 μg myelin oligodendrocyte glycoprotein (MOG_35–55) (MEVGWYRSPFSRVVHLYRNGK) (Invitrogen) in 100 of μl PBS and 0.5 mg Mycobacterium tuberculosis H37Ra (Difco) in 100 of μl IFA (Sigma-Aldrich). Mice received 200 ng pertussis toxin (List Biological Laboratories) by i.v injection at the time of immunization and again 48 h later. For passive EAE induction, donor mice were immunized s.c. with 300 μg MOG_35–55 in IFA supplemented with 0.5 mg M. tuberculosis. Ten days after immunization, mice were sacrificed and spleens were removed and homogenized and RBC were lysed. The cells were cultured for 3 days in complete DMEM supplemented with 5% FCS (Invitrogen) and 30 μg/ml of MOG_35–55. Then cells were harvested and dead cells were removed. Cells were then washed and injected i.p. into recipient mice irradiated at 400 rad (20 × 10⁶ cells/mouse). Animals received 200 ng/mouse pertussis toxin on days 0 and 2 after transfer (25). Clinical signs of disease were assigned scores on a scale of 0–5 daily as follows: grade 0, normal; grade 0.5, partially limp tail; grade 1, completely limp tail; grade 2, unilateral partial hindlimb paralysis; grade 2.5, bilateral partial limb paralysis; grade 3, complete bilateral hindlimb paralysis; grade 4, total paralysis and unilateral forelimb paralysis; and grade 5, moribund or death. After the onset of EAE, food and water were provided on the cage floor. To eliminate any diagnostic bias, scores were assigned by researchers blinded to mouse identity.

Mice were sacrificed and extensively perfused with PBS and then with 4% (w/v) paraformaldehyde in PBS. Spinal cords were fixed in 4% paraformaldehyde and then dissected and embedded in paraffin. Spinal cord sections (5 μm) were stained with Luxol fast blue (Sigma-Aldrich) to assess the degree of demyelination. Semiquantitative analysis of inflammation and demyelination was performed in a blinded manner as previously described (26). Immunohistochemistry was performed using anti-CD4, anti-Gr-1, anti-Mac1, and anti-B220 (all from BD Biosciences) to assess infiltration of immune cells. Cells positive for each marker were counted and tissue measurements were performed using a spot advance image analysis system. The number of stained cells per 10⁴ square pixel tissue area was calculated (27).

Blood samples were collected into tubes containing EDTA, stored on ice, and centrifuged within 60 min. Plasma was separated and then stored at −80°C for assay at a later time. GABA levels were determined using the HPLC procedure described as follows: samples were acidified by the addition of 0.4 N perchloric acid on ice and were centrifuged to remove protein. The supernatant was retained, basified, and shaken with the liquid ion-pairing agent di-(2-ethylhexyl)phosphate. The aqueous layer was retained and briefly centrifuged. The sample was then reacted with fluoraldehyde (o-phthaldialdehyde, OPA) reagent and injected onto an HPLC system attached to a fluorescence detector and an integrator. All plasma samples were assayed in the same batch, and all assays were performed by a laboratory technician blinded to the study conditions.

MNCs (2.5 × 10⁶/ml) were plated in 96-well plates and stimulated in triplicate with or without 20 μg/ml Ag. Culture supernatants were collected after 48 h and analyzed for IFN-γ production using ELISA (R&D Systems). To assess cell proliferation, 1 μCi [³H]thymidine (ICN Radiochemicals) was added at 56 h and cells cultured for another 16 h.

Total RNA was extracted and mRNA expression of GAT-1, the subunits of GABA_A receptor, and cytokines was quantitated using real-time PCR as previously described (28). mRNA expression was normalized to endogenous β-actin expression in the same sample. Relative expression was calculated as the difference (ΔCt, cycle threshold) between the Ct values of the target gene and of β-actin, and given as 2^−ΔCt. Primer sequences are detailed in Table I.

For Western blot analysis, cytoplasmic proteins and membrane proteins were prepared as described previously (29, 30). Protein aliquots (30 μg) were mixed with an equal amount of 2× SDS sample buffer, boiled at 98°C for 5 min, centrifuged, and resolved on 10% SDS-PAGE. The gels were then transferred onto polyvinylidene difluoride membranes (Millipore). Membranes were blocked with 5% nonfat milk powder in TBS/Tween 20 and incubated overnight at 4°C with specific primary Abs: anti-IκBα, anti-pIκBα, anti-IKKα, anti-IKKβ, and anti-pIKKα/β (all obtained from Cell Signaling Technology); anti-STAT1, anti-pSTAT1, anti-STAT4, and anti-STAT6 (all purchased from BD Biosciences); and anti-GATA3 and anti-T-bet (Santa Cruz Biotechnology) and anti-GAT-1, which was generously provided by Dr. J. Fei. Anti-actin (Sigma-Aldrich) was used to detect β-actin as loading control. After washing, subsequent incubation with appropriate HRP-conjugated secondary Abs for 1 h at room temperature, and extensive washing, signals were visualized by ECL (Cell Signaling Technology).

Gel mobility shift assays were performed on nuclear extracts as previously described (30). DNA-protein complexes were resolved on 4% native polyacrylamide gels. The probes used were double-stranded ³²P-labeled oligonucleotides containing the consensus binding sequences for NF-κB (5′-AGTTGAGGGGACTTTCCCAGGC-3′). For competition assays, an excess of unlabeled double-stranded oligonucleotides was added. Gels were vacuum-dried and visualized by autoradiography.

For surface staining of CD3, CD4, CD8, CD11b, CD19 (all from BD Biosciences) and GAT-1 (Millipore), cells were permeabilized with fixation/permeabilization (Cytofix/Cytoperm) solution (BD Biosciences) for 20 min and incubated for 30 min at 4°C with fluorochrome-conjugated Abs, and then analyzed with a FACSAria (BD Biosciences).

Mice were injected i.p. with 2 ml of 4% fluid thioglycolate medium (Sigma-Aldrich). Two days later, the peritoneal exudate cells were harvested by washing the peritoneal cavity with 10 ml of PBS. Cells were centrifuged at 300 × g for 10 min and the cell pellet was resuspended in complete DMEM supplemented with 5% FCS and plated in 12-well tissue-culture plates. After 2 h of incubation, nonadherent cells were removed by washing three times with PBS. Adherent cells, consisting of 95% macrophages (detected with CD11b, which was purchased from BD Biosciences), were supplemented with fresh complete DMEM supplemented with 5% FCS and incubated at 37°C for 2 h before experimentation. In all purifications, cell viability was >95% as determined by trypan blue staining. Macrophage cell morphology was examined under light microscopy.

To assess macrophage activation, macrophages were stimulated with 1 μg/ml of LPS (Sigma-Aldrich) and supernatants were collected 24 h later. Production levels of IL-6, IL-12, and TNF-α were determined using ELISA kits (R&D Systems).

Mice received s.c. injections of 25 μg OVA in 50 μl CFA at days 1 and 8. At day 15, all mice received a s.c. challenge with 10 μl OVA (1 mg/ml) injected into the ear. Ear thickness was measured by micrometer (Mitutoyo) at 24 and 48 h following the challenge.

Significance between two groups was examined using Student’s t test after analyzing the variance. A p-value of <0.05 was considered significant.

In MS and EAE, a broad range of neurotransmitters, including glutamate, 5-hydroxytryptamine, and cannabinoids and their receptors or transporters have been found to be up-regulated. GAT-1 maintains the level of the major inhibitory neurotransmitter, GABA, and is widely expressed in the CNS, but its role in MS and EAE is unknown. We first measured the expression of GAT-1 after EAE induction. We immunized C57BL/6 mice with MOG_35–55 to induce EAE, or with OVA as an irrelevant Ag. At the peak of EAE (day 15 postimmunization (p.i.)), we obtained spinal cords from naive mice, OVA-immunized mice (OVA mice), and MOG_35–55-immunized mice (EAE mice) and compared their GAT-1 expression. We found that there was no difference in GAT-1 expression at mRNA levels between naive mice and OVA mice. However, GAT-1 mRNA expression in spinal cord from EAE mice was reduced >70% compared with that of control mice (Fig. 1,A). We further confirmed our findings using Western blot, obtaining similar results. The protein levels of GAT-1 were significantly decreased after EAE induction (Fig. 1 B). These results indicate that GAT-1 expression is greatly suppressed in the CNS after EAE induction.

To study further the role of GAT-1 in EAE, age-matched GAT-1^−/− and WT mice were immunized with MOG_35–55 and disease progression was monitored by clinical assessment. All GAT-1^−/− mice developed neurological signs of EAE starting at approximately days 12–15 after immunization (Fig. 2,A and Table II), followed by severe ascending paralysis, reaching a mean maximum clinical score of 2.8 ± 0.66. In contrast, 60% of WT mice developed EAE, with less severity, showing only mild paralysis and reaching a mean maximum clinical score of 1.38 ± 1.37 (p < 0.01) (Fig. 2,A and Table II). Taken together, the data indicate that GAT-1^−/− mice are more susceptible to EAE than are WT mice, suggesting that GAT-1 might play an important role in suppressing the inflammatory response that drives EAE pathogenesis.

At the same time, we observed some interesting nonclassic EAE signs in GAT-1^−/− mice. Soon after EAE induction, GAT-1^−/− mice displayed readily observable continuous tremors in the limbs and tail. They walked with an abnormally large paw angle relative to the direction of walking, and with abnormal gait. Additionally, when suspended by the tail, GAT-1^−/− mice displayed trembling and flexor contraction, which resemble a typical mouse model for anxiety. In WT control mice with EAE or GAT-1^−/− mice immunized with only CFA, or CFA + OVA, we observed none of the above phenomena. The details of clinical sign development in the two groups are described in Table III.

Recruitment of autoreactive T cells into the CNS is crucial for the onset of EAE, which attracts further autoreactive T cells and leads to an amplification of the immune responses. Subsequently, additional T cells, macrophages, and granulocytes are recruited from the peripheral lymphoid organs to the inflammatory lesion, aggravating the tissue damage (31). To differentiate further pathological changes between GAT-1^−/− and WT mice with EAE, we performed histologic staining and examined inflammatory infiltration and demyelination. We did find that, in contrast to WT mice, spinal cord sections from GAT-1^−/− mice had abundant cellular infiltration and demyelination lesions (Fig. 2,B), particularly in the lumbar and thoracic segments. Detailed analysis of invading cells of the lesions showed increased numbers of CD4⁺ T cells, granulocytes, macrophages, and B cells in GAT-1^−/− mice (Fig. 2 C). Therefore, the absence of GAT-1 appeared to promote trafficking of immune cells to the target organ, leading to enhanced inflammation and tissue damage.

Since GAT-1 is the major transporter involved in maintaining a GABA reservoir in the CNS, we asked whether GABA concentration is affected in GAT-1^−/− mice. We found that GAT-1 deficiency has no effect on plasma GABA concentration either before or after EAE induction (Fig. 3,A). Several major subunits of GABA_A receptor on MNCs have been reported to be involved in GABA-mediated immune suppression, including α₁, α₂, β₁, β₂, δ, γ₃, and so forth (32). Thus, we next investigated the expression of these subunits on MNCs between WT and GAT-1^−/− EAE mice. We found no significant difference (Fig. 3 B). Therefore, GAT-1 appears to play a negative role in EAE pathogenesis with little interference in systemic GABA concentration or GABA_A receptor expression.

To determine whether the deficiency of GAT-1 might enhance T cell activation and proliferation, we examined the recall responses to MOG_35–55 by T cells at the peak of EAE (days 18–22 p.i.). Compared with those of WT mice, the proliferative responses of MOG_35–55-reactive T cells were higher in GAT-1^−/− mice (Fig. 4 A). We also cultured naive T cells from GAT-1^−/− and WT mice with anti-CD3 and anti-CD28 and obtained similar results (data not shown).

Inflammatory cytokines have been identified as playing pivotal roles in the establishment and maintenance of EAE. Proinflammatory cytokines, including IFN-γ, TNF-α, IL-12, and IL-6, are well known for their EAE-promoting functions. Recently, several laboratories have discovered that IL-23 induces the expansion of IL-17-producing CD4⁺ T cells, which are now widely thought to be an important pathogenic population in autoimmune inflammation (7, 8). Therefore, we quantified the expression of these cytokines by real-time PCR in both spleen (the immune-priming organ) and spinal cord (the target organ) at the peak of EAE.

We found that when EAE was induced in GAT-1^−/− mice, as compared with controls, the mRNA expression of IFN-γ, TNF-α, and IL-6, as well as that of IL-23 and IL-17, was significantly elevated in both spleen and spinal cord. The expression of IL-12 in GAT-1^−/− mice was up-regulated in spinal cord but not in spleen. The expression of IL-5 was decreased in spinal cord of GAT-1^−/− mice, while its expression in spleen was undetectable (Fig. 4, B and C). Therefore, GAT-1 deficiency appears to significantly enhance the production of major inflammatory cytokines, which in turn exacerbates EAE.

To investigate further the molecular mechanisms underlying GAT-1 deficiency-induced EAE aggravation, we examined transcriptional factors orchestrating inflammatory responses. First, we studied NF-κB, which has a central role in coordinating the expression of a wide variety of genes that control immune responses, including cell activation and the production of Th1 cytokines (33). EAE induced the nuclear transcription of NF-κB in WT mice, as measured by gel shift. However, GAT-1^−/− mice with EAE showed more vigorous NF-κB-DNA binding activity (Fig. 5,A). Since nuclear translocation of NF-κB took place after NF-κB/IκBα dissociation, we examined the effect of GAT-1 deficiency on phosphorylation and degradation of IκBα. As shown in Fig. 5 B, phosphorylation and degradation of IκBα were both enhanced in GAT-1^−/− mice with EAE.

We next investigated whether GAT-1 modulates IKK, the kinase upstream of IκBα. Activated IKK promotes the phosphorylation of IκBα and induces its degradation. We found that the phosphorylation of IKKα/β was increased in GAT-1^−/− mice (Fig. 5 C). Taken together, our data clearly indicate that the deficiency of GAT-1 affects IKKα/β phosphorylation, inducing the active NF-κB formation triggered by the degradation of IκBα.

We have shown that GAT-1 deficiency induces high levels of IFN-γ production in EAE. Several transcription factors, especially T-bet, STAT1, and STAT4, are essential for the differentiation of IFN-γ-producing Th1 cells (34, 35). To determine which transcription factor is involved in promoting IFN-γ production in GAT-1^−/− mice, we examined the expression of T-bet, STAT1, and STAT4 in spleen. As shown in Fig. 5,D, the expression levels of T-bet were much increased in GAT-1^−/− mice. Recent studies indicate that STAT1 is also critical for the induction of T-bet (35, 36). Consistent with the expression of T-bet, the expression of STAT1 and pSTAT1 were up-regulated in GAT-1^−/− mice. However, there was little difference in STAT4 expression between the two groups. Additionally, GATA3 expression was inhibited in GAT-1^−/− mice, but there was little difference in STAT6 expression between the groups (Fig. 5 E). These data indicate that enhanced T-bet-IFN-γ-STAT1 circuit activity and suppressed GATA3 expression are involved in the mechanisms underlying the exacerbation of EAE in GAT-1^−/− mice.

To confirm the function of GAT-1 deficiency in the effector phase of EAE, we adoptively transferred encephalitogenic MNCs derived from WT or GAT-1^−/− donor mice into WT recipient mice. Fully primed and activated encephalitogenic MNCs derived from WT mice induced a mild EAE. However, more vigorous EAE was induced by GAT-1^−/− MNCs as characterized by higher clinical scores and more extensive inflammatory infiltration in the spinal cord (Fig. 6). Therefore, the absence of GAT-1 in MNCs has more potent capability in triggering and aggravating EAE in mice.

Previous studies demonstrated only that GAT-1 is expressed in the CNS. However, our data indicate that GAT-1 in the immune system is critical in EAE induction. Therefore, we wondered whether GAT-1 might be expressed on immune cells and, more importantly, on which type(s) of immune cells. To answer this question, we first sought GAT-1 expression in MNCs from naive mice, but found none (Fig. 7,A). When these MNCs were stimulated in vitro with Con A or LPS, GAT-1 remained undetectable (Fig. 7,A). However, when MNCs from naive mice were stimulated with anti-CD3 and anti-CD28 to induce T cell activation, a high expression level of GAT-1 mRNA was observed (Fig. 7 B).

To further verify that GAT-1 expression was triggered by Ag signaling, we immunized C57BL/6 mice with OVA and, after 7 days, obtained MNCs. Compared with MNCs from naive mice, MNCs from OVA-immunized mice (OVA-I) showed elevated levels of GAT-1 at both mRNA and protein levels. Furthermore, when these cells were restimulated ex vivo with OVA for 48 h, GAT-1 expression reached even higher levels (Fig. 7, C and D). Additionally, we analyzed GAT-1 expression on OVA-reactive cells by flow cytometry. We found that GAT-1 was indeed expressed on activated CD4⁺ T cells and CD8⁺ T cells (Fig. 7 E), but not on macrophages and B cells (data not shown).

This implies that GAT-1 signaling might be more important in T cells than in APCs. To further test this hypothesis, we purified peritoneal macrophages and performed in vitro studies. First, we found that WT macrophages did not express GAT-1, either before or after LPS activation (Fig. 7,F). Second, after activation, macrophages from WT and GAT-1^−/− mice produced similar levels of TNF-α, IL-6, and IL-12 (Fig. 7 G). Taken together, the data show that GAT-1, triggered by Ag stimulation, is expressed not only in the CNS, but also on Ag-activated T cells. These data suggest that GAT-1 might be an important molecule involved in modulating T cell-mediated Ag-specific immune responses.

The ability of GAT-1 to interfere with T cell-mediated cellular immunity was further assessed in a DTH model. DTH response is mediated by infiltrating IFN-γ-secreting Th1 cells in response to formerly encountered Ags, resulting in a specific inflammation at the site of local challenge. Mice immunized with OVA at days 1 and 8 and then challenged locally on the ear at day 15 develop a significant DTH response as measured by ear swelling at days 16 and 17. Mice deficient in GAT-1 show a significantly enhanced DTH response as compared with that of WT mice (Fig. 8 A).

We then examined whether GAT-1 deficiency enhanced DTH by promoting T cell activation and cytokine production. As compared with cells from WT mice, MNCs from GAT-1^−/− mice proliferate more vigorously (Fig. 8,B). Additionally, IFN-γ production in GAT-1^−/− mice increases nearly 25-fold as compared with that in WT mice (Fig. 8 C). Therefore, besides EAE, T cell-mediated DTH responses are also much enhanced in GAT-1^−/− mice.

In this study, we addressed a novel and original finding concerning GAT-1 function. Although GAT-1 is known to be involved in CNS homeostasis, we have shown herein that GAT-1 exerts a critical inhibitory effect upon EAE pathogenesis. In the absence of GAT-1, mice develop exacerbated EAE, accompanied by increased production of proinflammatory cytokines, including IFN-γ, IL-23, TNF-α, IL-17, and IL-6. The effect of GAT-1 deficiency on T cell immunity in EAE appears to be largely attributable to augmented NF-κB, T-bet, and STAT1 signaling pathways. Notably, in addition to the wide expression of GAT-1 in the CNS, GAT-1 is now seen also to be expressed on activated CD4⁺ and CD8⁺ T cells, a response triggered by Ag stimulation.

MOG-induced EAE in mice shares many common characteristics with MS in humans. It usually manifests itself as an ascending progressive paralysis eventually leading to forelimb paralysis and/or to a moribund state in severe cases (classic EAE) (3). Interestingly, there have been some reports of nonclassic manifestations of EAE with ataxia, spasticity, and tremor (37, 38). In the present study, we found that besides exacerbated classic signs of ascending paralysis, GAT-1^−/− mice with EAE also developed tremor, anxiety, ataxia, and spasticity. These nonclassic signs are rare in classic EAE, but they are quite common in MS patients. Thus, EAE as induced in GAT-1^−/− mice may provide us a more specific and accurate model for MS.

The development of EAE is dependent on the infiltration of activated mononuclear cells into the CNS from the periphery, an event that is accompanied by abundant production of proinflammatory cytokines, including IL-6, TNF-α, and IFN-γ (5, 6). In GAT-1^−/− mice with EAE, the expression of IL-6, TNF-α, and IFN-γ are all up-regulated. As a critical transcription factor, NF-κB controls the expression of those proinflammatory cytokines and the corresponding signal transduction, modulating cell activation and differentiation (39). As expected, IκBα phosphorylation and degradation were enhanced and the following NF-κB-DNA binding activity was increased in GAT-1^−/− MNCs. The finding that GAT-1 deficiency activated IKKβ provides a mechanistic basis for explaining the strengthened NF-κB activity. Previous studies have shown that activation of NF-κB in cells is essential for the induction of EAE (9, 39), while inactivation of NF-κB or disruption of the integrity of the IKK complex can protect mice from EAE (40). Given that IKKβ and IκB are the regulators of NF-κB activity, our results provide persuasive evidence that GAT-1 deficiency affects the upstream regulatory protein of NF-κB activity. Since the dynamic regulation of IKKβ-NF-κB activity controls the balance of survival and death of activated T cells (41), future studies should be aimed at identifying whether GAT-1 influences multiple levels of NF-κB activity associated with the modulation of cell activation and survival.

Our data have also provided evidence that in GAT-1^−/− mice, T-bet/STAT1-induced IFN-γ production is enhanced. Although the complex and seemingly paradoxical functions of IFN-γ in the regulation of EAE are emphasized by the different outcomes of systemic blockade of this cytokine, the activation and differentiation of myelin-specific precursor CD4⁺ T cells into encephalitogenic Th1-type cells are always considered essential for initiation of the disease (34, 42, 43). In EAE, it has been proposed that the enhancement of the T-bet/STAT1 and IL-12/STAT4 pathways results in the production of high levels of IFN-γ, which is the central molecule regulating Th1 cell differentiation (44). GAT-1^−/− mice with EAE show elevated T-bet and STAT1 expression levels. T-bet has been found to play a key role in regulating pathogenic T cells in EAE (45), and its action in regulating IFN-γ production appears to be mediated by the IFN-γ/STAT1 pathway (46). Although loss of STAT1 induces severe EAE by creating disequilibrium in the balance of IL-10/IFN-γ (10), residual T-bet expression in STAT1^−/− mice probably accounts for the generation of pathogenic Th1 cells capable of initiating EAE, as (STAT1 × T-bet)^−/− mice do not produce IFN-γ and are also resistant to the development of EAE (10). However, in GAT-1^−/− mice, the expression of IL-12 and STAT4 was seen to be the same as that in WT control mice, suggesting that the IL-12/STAT4 pathway might not be modulated by GAT-1. T-bet also influences IL-23 responsiveness and IL-17 production through IL-23R, and both IL-23 and IL-17 play important pathogenic roles in EAE (47, 48). We demonstrate herein that increased expression of IL-17 and IL-23 is detectable in GAT-1^−/− mice with EAE. It remains to be determined whether the upstream regulators of T-bet and IL-17 are affected by GAT-1^−/− deficiency.

Although wide expression of GAT-1 in the CNS has been established (21), our data for the first time reveal that GAT-1 is also expressed on Ag-activated T cells. Therefore, GAT-1 deficiency increases the Ag-pulsed signaling cascade through TCR and costimulatory signals, triggering IκBα degradation and NF-κB translocation associated with cell activation and inflammatory cytokine production. At the same time, GAT-1 deficiency affects T cell differentiation by up-regulating T-bet/STAT1 expression and down-regulating GATA3 expression. However, the complex mechanisms bridging GAT-1, Ag signaling TCR activation, IKK degradation, and T cell differentiation still need further study (Fig. 9).

Our findings indicate that GAT-1 plays a negative role in EAE pathogenesis. Previous reports have shown that GAT-1 deficiency alters the activity of GABA receptors in hippocampus (49), and that GABA administration suppresses diabetes and DTH through inhibition of the development of proinflammatory T cell responses and arresting TCR-mediated T cell cycle (32, 50). Therefore, we had to exclude the possibility that GABA concentration and GABA receptor expression levels may be altered in GAT-1-deficient mice, thus leading to the more severe EAE. However, we find that WT mice and GAT-1^−/− mice have similar plasma GABA concentrations as well as expression of major subunits of GABA_A receptor in MNCs. Thus, GAT-1 plays negatively roles in EAE pathogenesis with little interference with GABA concentration and GABA_A receptor expression (Fig. 9).

In summary, our findings suggest that GAT-1, whose role in autoimmune disease is still uncertain, has a crucial inhibitory effect upon EAE. When GAT-1 is absent, immune responses are directed toward generating pathogenic T cell subsets. Thus, GAT-1 bridges the relationship between the nervous system and the immune system and acts to orchestrate the cytokine microenvironment associated with immune responses. Therefore, strategies targeting GAT-1 might prove an attractive and useful approach for the treatment of autoimmune inflammatory diseases of the CNS.

The authors have no financial conflicts of interest.