De Novo Central Nervous System Processing of Myelin Antigen Is Required for the Initiation of Experimental Autoimmune Encephalomyelitis¹

Experimental autoimmune encephalomyelitis (EAE)³ is a CD4⁺ T cell-mediated disease of the CNS. In C57BL/6 (B6) mice, EAE can be induced by immunization with antigenic myelin components, or by the adoptive transfer of myelin-specific CD4⁺ T cells that have been reactivated in vitro (1, 2, 3). In both cases, T cells specific for CNS myelin infiltrate the CNS, causing the influx of macrophages and activation of CNS resident cells (4). The release of inflammatory mediators causes inflammation and subsequent tissue destruction and demyelination of axonal tracks. This results in acute paralysis with clinical and histopathological similarities to the human demyelinating disease, multiple sclerosis (MS) (5, 6).

Activation of CD4⁺ T cells requires that their TCR must recognize cognate Ags presented by class II MHC. The critical importance of class II MHC Ag presentation in EAE initiation is illustrated by the fact that treatment with anti-class II Abs inhibits or ameliorates the disease (reviewed in Ref. 7). A variety of proteins have important roles in the generation of the functional peptide/class II MHC protein complex. Invariant chain (Ii) and H-2M (DM) are critical components of the class II Ag processing pathway. Newly synthesized class II α- and β-chains associate with Ii in the endoplasmic reticulum (ER), forming a heterotrimeric complex, which then forms a homotrimer, or trimer of trimers (8). The Ii performs at least three critical functions in class II Ag processing and presentation: it aids in the assembly of the class II complex; it provides a signal sequence, targeting the class II/Ii complex through the Golgi apparatus to the endocytic pathway; and a portion of Ii, the class II-associated invariant chain peptide (CLIP) occupies the class II peptide-binding groove, preventing the binding of ER-resident peptides. Within endocytic and/or lysosomal compartments, the Ii is proteolytically degraded in a stepwise fashion by cathepsins (9), until only CLIP remains associated in the peptide-binding groove of the class II αβ heterodimer (10). DM colocalizes with the class II/CLIP complexes in the endocytic compartment, where it catalyzes the removal of the CLIP peptide from the class II MHC peptide-binding groove and enables the binding of antigenic peptides (11, 12, 13). DM also functions as a peptide editor, removing peptides with high off-rates and preferentially allowing the binding of high affinity peptides to class II complexes (14). Upon binding antigenic peptides, the class II/peptide complex traffics to the APC surface where CD4⁺ T cells can recognize the complex and be activated. Loss of expression of Ii or DM results in profound defects in Ag processing and presentation, class II expression, and CD4⁺ T cell development (12, 15, 16, 17). The expression of all of these proteins; class II, Ii, and DM is regulated by the class II transcriptional activator (CIITA) (18, 19, 20, 21). The CIITA is responsible for the constitutive expression of class II Ag processing proteins in “professional” APCs as well as the IFN-γ-inducible expression of class II, Ii, and DM in “nonprofessional” APCs. Mice deficient in CIITA have profound defects in class II, Ii, and DM expression (20, 22, 23) (although the effect on Ii expression is less severe) and substantial decreases in CD4⁺ T cells due to the absence of class II expression and defective thymic selection.

It is not clear which of a variety of candidate professional (e.g., macrophages, dendritic cells, and B cells) and nonprofessional (e.g., endothelial cells, microglia, and astrocytes) APCs present in the CNS during T cell-mediated demyelination are involved in presentation of myelin Ags to autoreactive CD4⁺ T cells active in disease initiation or relapses. In vivo studies using radiation bone marrow chimeras have demonstrated potential roles for both infiltrating macrophages, resident microglia, and CNS parenchymal cells in processing and presenting encephalitogenic Ags to CD4⁺ T cells (24, 25, 26, 27). Each of these cell populations express different levels of CIITA, class II, Ii, and DM, and the expression of these proteins can change with the activation of the different cell populations (28, 29). Thus, there is significant interest in the expression requirements of the class II Ag processing and presentation proteins and their impact on activation of encephalitogenic CD4⁺ T cell populations. Thus, we investigated the role of these proteins in Ag presentation and myelin-specific CD4⁺ T cell activation in the context of CNS disease in vivo, using mice deficient in the expression of Ii, DM, and, via the CIITA knockout (KO) mouse, deficient in Ii, DM, and class II. Mice deficient in CIITA, Ii, or DM are resistant to initiation of EAE by both active priming and adoptive transfer of wild-type (wt) myelin oligodendrocyte protein, MOG_35–55-specific encephalitogenic T cells. Although both Ii- and DM-deficient APCs can present MOG peptide to CD4⁺ T cells, neither is capable of processing and presenting the encephalitogenic peptide of intact MOG protein. Remarkably, DM-deficient mice can prime a potent, peripheral Th1 response to MOG_35–55, comparable to the response seen in wt mice, yet maintain resistance to EAE initiation and can adoptively transfer EAE to wt, but not DM-deficient, mice. Together, these data demonstrate that the inability to process antigenic peptide from intact myelin protein results in resistance to EAE and de novo processing and presentation of myelin Ags in the CNS is absolutely required for the initiation of autoimmune demyelinating disease.

C57BL/6 female mice, 5–6 wk old, were purchased from The Jackson Laboratory (Bar Harbor, ME). CIITA-, DM-, and Ii-deficient mice are described elsewhere (12, 16, 20). CIITA-deficient mice have been backcrossed onto the C57BL/6 background six generations. DM- and Ii-deficient mice have been backcrossed onto the C57BL/6 background 10 and 13 generations, respectively. Mutant mice were bred by homozygous brother-sister matings and all mice were housed in barrier conditions with the Center for Comparative Medicine at Northwestern University (Chicago, IL). Mice were maintained on standard laboratory food and water ad libitum. Paralyzed animals were afforded easier access to food and water.

Hybridomas producing the anti-class I and anti-class II Abs (M1/42 and M5/114, respectively) were purchased from American Type Culture Collection (Manassas, VA) and cultured in DMEM 10% FCS. Supernatants were tested for Ab, filtered, and sterilized before storage. MOG_35–55 (MEVGWYRSPFSRVVHLYRNGK) was purchased from Genemed Synthesis (San Francisco, CA). PLP_178–191 (NTWTTCQSIAFPSK) was purchased from Peptides International (Cleveland, OH). Amino acid composition was verified by mass spectrometry and purity (>98%) was assessed by HPLC. rMOG, consisting of the extracellular portion of MOG (aa 1–125) expressed in, and purified from Escherichia coli, was the generous gift of Dr. M. Gardinier (University of Iowa College of Medicine, Iowa City, IA).

For MOG_35–55-induced EAE, 6- to 7-wk-old female mice were immunized s.c. with 200 μl of an emulsion containing 800 μg of Mycobacterium tuberculosis H37Ra (Difco, Detroit, MI) and 200 μg MOG_35–55 distributed over three spots on the flank. Each mouse additionally received 200 ng pertussis toxin (PTx) (List Biological Laboratories, Campbell, CA) in 200 μl PBS i.p. on days 0 and 2 postimmunization. Animals that were immunized twice received the same immunization 7 days after the first immunization, without PTx. For PLP_178–191-induced EAE, 6- to 7-wk-old female mice were immunized s.c. with 200 μl of an emulsion containing 800 μg of M. tuberculosis H37Ra and 50 μg PLP_178–191 distributed over three spots on the flank. Each mouse additionally received 200 ng PTx in 200 μl PBS i.p. on days 0 and 2 postimmunization. Individual animals were observed daily and clinical scores assessed in a blinded fashion on a 0–5 scale as follows: 0 = no abnormality, 1 = limp tail, 2 = limp tail and hind limb weakness (legs slip through cage top), 3 = hind limb paralysis, 4 = hind limb paralysis and forelimb weakness, and 5 = moribund. The data are reported as the mean daily clinical score ± SEM for all animals in a particular group and/or as the mean peak clinical score ± SEM, i.e. the mean clinical score for all animals at the peak of disease. Unless otherwise mentioned, all mice were age and sex-matched for all experiments.

Female donor mice (6- to 10-wk-old) were immunized s.c. with 200 μl of an emulsion containing 800 μg of M. tuberculosis H37Ra and 200 μg MOG_35–55 distributed over three spots on the flank. Draining lymph nodes (LN) were harvested from donor mice after 7–11 days for in vitro stimulation. LN cells were cultured (10 × 10⁶ cells/ml) in DMEM containing 10% FBS, 1 mM glutamine, 1% penicillin-streptavidin, 1 mM nonessential amino acids and 5 × 10⁻⁵ M 2-ME (D-10; all products from Sigma-Aldrich, St. Louis, MO) with MOG_35–55 peptide (30 μg/ml) and human rIL-12 (20 ng/ml; R&D Systems, Minneapolis, MN). After 72 h incubation, cells were counted, washed, and resuspended (2.5 × 10⁷ T cell blasts/ml; 10–15 × 10⁷ LN cells/ml) in buffered salt solution. T cell blasts were differentiated from other LN cells by size under microscopic observation. On day 0, 6- to 7-wk-old female B6 or KO mice were injected i.p. with 5 × 10⁶ T cell blasts/mouse (in 200 μl). Recipient mice received 200 ng PTx in 200 μl PBS i.p. on days 0 and 2. Individual animals were observed daily and clinical scores were assessed as described above. For adoptive transfer of T cells from DM KO mice, donors and LN cells were prepared as described above, except that donors were primed with MOG_35–55 on both days 0 and 7 and donors were sacrificed at day 14 postinitial priming.

DTH responses were measured using a 24-h ear-swelling assay. Prechallenge ear thickness was determined using a Mitutoyo model 7326 engineer’s micrometer (Schlesinger’s Tool, Brooklyn, NY). Immediately thereafter, mice were ear-challenged by injecting 10 μg of peptide (in 10 μl of saline) into the dorsal surface of the ear using a 100 μl syringe fitted with a 30-gauge needle. The increase in ear thickness was determined 24 h after ear challenge. Results are expressed in units of 10⁻⁴ inches ± SEM. Significance of ear swelling in experimental over naive mice was assessed by the Student’s t test.

Spleens were collected from naive, C57BL/6 wt, or KO mice as indicated. RBCs were removed by hypotonic lysis and the remaining cells were used as APCs. The APCs were irradiated (3500 rad), washed, and cultured in 96-well microtiter plates at a density of 5 × 10⁵ cells/well. Varying concentrations of MOG_35–55 peptide or rMOG were added to the different APCs. A MOG_35–55-specific T cell line (10⁵ cells/well) was cocultured with the APCs and Ags in a total volume of 200 μl D-10. Cocultures were incubated for 96 h, being pulsed with 1 μCi/well [³H]TdR for the final 24 h of the 96-h incubation period. [³H]TdR uptake was detected using a Topcount microplate scintillation counter (Packard Instrument, Meriden, CT) and results are expressed as the mean of triplicate cultures ± SEM. The long-term, MOG_35–55-specific T cell line was derived from a C57BL/6 mouse primed with MOG_35–55/CFA as previously described (28). In brief, draining LN cells were isolated and restimulated in D-10 for 4 days with peptide, and rested in D10 plus human rIL-2 (2 U/ml; Roche, Indianapolis, IN) for a minimum of 2 wk. Peptide-specific restimulation and rest were repeated every 14–35 days.

Draining LN were harvested from primed mice, counted, and cultured in 96-well microtiter plates at a density of 5 × 10⁵ cells/well in a total volume of 200 μl HL-1 medium (BioWhittaker, Walkersville, MD; 1% penicillin/streptavidin, 1% glutamine). Cells were cultured with medium alone or different concentrations of peptide Ag for 72 h. Culture wells were pulsed with 1 μCi/well [³H]TdR for the final 20 h of the 72-h incubation period. [³H]TdR uptake was detected using a Topcount microplate scintillation counter and results are expressed as the mean of triplicate cultures ± SEM.

Nitrocellulose-coated, 96-well flat-bottom microculture plates (Whatman, Clifton NJ) were precoated overnight at 4°C with 100 μl of anti-IFN-γ, anti-IL-4, or anti-IL-2 (R46A2 and 11B11 at 4 μg/well and JES6–1A12 at 2 μg/well, respectively) purchased from BD PharMingen (San Diego, CA). Plates were washed four times with sterile PBS and wells were blocked with 200 μl sterile DMEM 1% BSA for 1 h at room temperature. LN cells (5 × 10⁵) or 10⁶ spleen cells were cocultured with Ag at varying concentrations in HL-1 medium (1% penicillin/streptavidin, 1% glutamine). Cultures were incubated at 37°C for 36 h. In mAb coculture experiments, Ab supernatants were diluted 1/5 for a final volume of 200 μl/well. Plates were subsequently washed three times with PBS and three times with PBS/0.05% Tween20 (PBS/Tween). Biotinylated anti-IFN-γ, anti-IL-2, or anti-IL-2 (XMG1.2, BVD6-24G2, and JES6-5H4, respectively) at 2 μg/well diluted in PBS/Tween/1% BSA, were added at 100 μl/well and incubated overnight at 4°C in a humidified chamber. Plates were washed four times with PBS/Tween and incubated for 2 h at room temp with 100 μl/well anti-biotin alkaline phosphatase (Vector Laboratories, Burlingame, CA) diluted 1/1000 in PBS/Tween/1% BSA. Finally, plates were washed with PBS and developed in nitroblue tetrazolium-5-bromo-4-chloro-3-indolyl phosphate substrate solution (Pierce, Rockford, IL). The developing reaction was quenched after 30–45 min using distilled water. ELISPOTs were counted using the ImmunoSpot series 1.0 analyzer (Resolution Technology, Cleveland, OH). Samples were set up in triplicate and anti-CD3 mAb were included as a positive control for each group.

Mice were anesthetized and perfused with 1× PBS on day 20 postimmunization or day 31 postadoptive transfer. Spinal cords were removed by dissection, and 2- to 3-mm spinal cord blocks were immediately frozen in OCT (Miles Laboratories; Elkhart, IN) in liquid nitrogen. The blocks were stored at −80°C in plastic bags to prevent dehydration. Six micrometer thick cross-sections from the lumbar region (approximately L2-L3) were cut on a Reichert-Jung Cyocut CM1850 cryotome (Leica, Deerfield, IL), mounted on Superfrost Plus electrostatically charged slides (Fisher, Pittsburgh, PA), air dried, and stored at −80°C. Slides were stained using a Tyramide Signal Amplification (TSA) Direct kit (NEN, Boston, MA) according to manufacturer’s instructions. Lumbar sections from each group were thawed, air-dried, fixed in 2% paraformaldehyde at room temperature, and rehydrated in 1× PBS. Nonspecific staining was blocked using anti-CD16/CD32, (FcγIII/IIR, 2.4G2; BD PharMingen), and an avidin/biotin blocking kit (Vector Laboratories) in addition to the blocking reagent provided by the TSA kit. Tissues were stained with biotin-conjugated Abs anti-mouse CD4 (H129.19) and anti-mouse I-A^b (AF6-120.1) (BD PharMingen). Sections were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI; Sigma-Aldrich) and then coverslipped with Vectashield mounting medium (Vector Laboratories). Slides were examined and images were acquired via epifluorescence using the SPOT RT camera (Diagnostic Instruments, Sterling Heights, MI) and Metamorph imaging software (Universal Imaging, Downingtown, PA). Eight serial lumbar sections from each sample per group were analyzed at ×100 and ×400 magnification.

Mice were anesthetized and sacrificed by total body perfusion through the left ventricle using chilled 3% glutaraldehyde in PBS (pH 7.3). Spinal cords were dissected out and cut into 1 mm thick segments and postfixed in OsO₄, dehydrated, and embedded in Epon. Toluidine blue stained sections from 10 segments/mouse were read and scored as follows: ± = mild inflammation without demyelination; 1+ = inflammation with focal demyelination; 2+ = inflammation with multiple foci of demyelination; 3+ = marked inflammation with bilateral, converging areas of demyelination; 4+ = extensive bilateral areas of demyelination and remyelination.

Although the role of class II in the priming of myelin-specific T cells has been addressed using anti-MHC class II blocking Abs, susceptibility to EAE had not been studied with mice genetically deficient in class II expression. To investigate the possibility of nonclass II-restricted mechanisms for initiation of EAE, we tested CIITA KO mice for susceptibility to initiation of EAE by active priming and adoptive transfer of encephalitogenic T cells. Age- and sex-matched C57BL/6 wt and CIITA KO mice were primed with MOG _35–55/CFA and observed for clinical symptoms as described in Materials and Methods. Although 100% of wt mice were susceptible to disease, none of the CIITA KO mice showed any clinical symptoms (Fig. 1,a). In addition to defective class II expression, CIITA KO mice have a radically altered CD4⁺ T cell repertoire (20, 22, 23) and might not have CD4⁺ T cells capable of recognizing the MOG_35–55 epitope. Surprisingly, when MOG_35–55-primed mice were tested for 24-h DTH responses, CIITA KO mice displayed measurable Ag-specific responses (Fig. 1,c), albeit significantly reduced from those seen in wt mice. Analysis of the LN responses from MOG_35–55-primed wt and CIITA KO mice 8 days postpriming by ELISPOT showed that the T cells produced the Th1 cytokines IL-2 and IFN-γ (Fig. 2, a and b, respectively), although the frequency of cytokine-producing LN cells from CIITA KO mice was lower and required higher Ag doses for activation than LN cells from wt mice. The IL-2- and IFN-γ-producing cells from CIITA KO mice were not MOG-specific CD8⁺ T cell responses as coculture with anti-class I mAb (M1/42) did not block cytokine production, while addition of anti-class II mAb (M5/114) completely abrogated Th1 cytokine production (Fig. 2, e and f). The lack of cytokine production with the addition of M5/114 was not due to nonspecific cytotoxicity, as cultures containing anti-CD3 mAb were unaffected by the addition of the anti-class II mAb (data not shown). Splenic Th1 responses were also reduced in CIITA KO mice compared with wt mice (Fig. 2, c and d), suggesting that residual class II MHC was present in both the spleen and LN of CIITA KO mice. The lack of disease in CIITA KO mice could potentially be due to a switch from IFN-γ production to IL-4 production by CD4⁺ T cells, resulting in an inhibition of the inflammatory response. To address this possibility, we compared IFN-γ and IL-4 responses from the LN and spleen of MOG_35–55-primed wt and CIITA KO mice. There was little difference in the numbers of IFN-γ- and IL-4-producing cells between wt and CIITA KO mice 13 days postpriming (Fig. 2, g and h).

To investigate the possibility that the failure to induce active disease in CIITA KO mice may be due to defective presentation of myelin peptides in the CNS, we determined the ability of MOG_35–55-specific T cell blasts from wt B6 donors to adoptively transfer disease into wt vs CIITA KO mice. Similar to active disease, 100% of wt mice developed clinical EAE by adoptive transfer, while none of the CIITA mice were affected (Fig. 1,b). Both wt and CIITA KO mice recipients showed significant DTH responses (Fig. 1 d). This recall response in the CIITA KO mice eliminated the possibility that T cells primed by and restricted to wt IA^b were incapable of recognizing MOG peptide presented by the residual class II present in the CIITA KO mice.

CIITA mutant mice have dramatically reduced levels of class II, Ii, and DM (20, 22, 23). Although the CIITA KO mice could present antigenic peptide to MOG-specific T cells, the possibility remained that due to the reduction of critical accessory molecules, the APCs could not process protein Ags. Thus, mice deficient in Ii or DM expression were tested for their susceptibility to initiation of EAE. Age and sex-matched wt C57BL/6, Ii KO, and DM KO mice were primed with MOG_35–55/CFA and followed for disease as described in Materials and Methods. Neither Ii-deficient nor DM-deficient mice showed any clinical signs of disease (Fig. 3,a), while 100% of wt mice got sick. wt mice had a mean peak clinical score of 2.9 ± 0.5 and histological analysis showed massive infiltration (Fig. 4,a) and severe demyelination (Fig. 3,c, data not shown). When tested for in vivo Th1 responses by DTH, Ii-deficient mice had no response, while, similar to CIITA-deficient mice, DM KO mice had reduced, but significant (p < 0.0002) MOG-specific ear swelling (Fig. 5 a).

Both DM KO and Ii KO mice have significantly altered CD4⁺ T cell repertoires (12, 15, 16, 17, 30, 31, 32, 33). Although DM KO mice could prime a MOG_35–55-specific T cell response and elicit Th1 effector function (i.e. DTH), the Ii-deficient mice could not. Although DM- and Ii-deficient mice have diminished Ag processing and presentation in vivo, APCs from both can function in vitro to present peptide to CD4⁺ T cells (16, 34). To confirm our in vivo data, we tested LN cells from MOG_35–55-primed wt, Ii-, and DM-deficient mice for Th1 recall responses by ELISPOT. LN cells from MOG_35–55-primed DM and Ii KO mice produced IL-2 and IFN-γ (Fig. 6, a and c, respectively). In all cases, the frequency of Th1 cytokine-producing cells was less in the KO mice than in wt mice. Strikingly, even though the Ii-deficient mice did not have a MOG_35–55-specific DTH response (Fig. 5,a), their frequency of Th1 recall responses in vitro was greater than that seen in DM-deficient mice (Fig. 6, a and c).

As in the CIITA KO mice, while DM KO and Ii KO mice could prime attenuated MOG_35–55-specific Th1 responses, the possibility remained that the response was below an encephalitogenic threshold. Alternatively, the class II expressed in the mutant APCs could be priming a repertoire of MOG-peptide-specific T cells, which could not recognize MOG peptide processed from the CNS (30). To rule out the possibility of a T cell defect, wt MOG_35–55-specific T cells were adoptively transferred into C57BL/6 wt, Ii KO, and DM KO mice. Once again, 100% of wt mice developed EAE, while DM- and Ii-deficient mice remained healthy (Fig. 3, b and c). Upon measure of Th1 responses, mice from every adoptively transferred group, including Ii KO mice, elicited significant DTH responses (Fig. 5,b), although the responses in the DM KO and Ii KO mice were slightly and dramatically reduced as compared with the wt response, respectively. Surprisingly, when analyzed for their ability to drive Th1 recall responses in vitro, LN cells from Ii-deficient animals showed no responsiveness (Fig. 6, b and d). LN cells from DM KO mice, in contrast, drove a high frequency of IL-2- and IFN-γ- (Fig. 6, b and d) producing T cells, although reduced from the responses seen in wt mice.

DM-deficient mice can both prime and recall significant Th1 responses (Figs. 5 and 6), yet fail to show any overt symptoms of EAE. Although adoptive transfer of wt encephalitogenic T cell blasts failed to initiate disease in DM KO mice, the possibility remained that there was a subclinical disease that exhibited no characteristic motor defects. To eliminate this possibility, MOG_35–55-primed or adoptively transferred wt, Ii KO, and DM KO mice were perfused and their spinal cords were analyzed histologically for CNS infiltration and inflammation as described in Materials and Methods. Although MOG-primed wt mice showed severe signs of inflammation and demyelination (Fig. 3,c), Ii KO and DM KO mice showed no clinical pathology. Immunohistochemical analysis showed that spinal cords from wt mice either MOG_35–55-primed or adoptively transferred with MOG_35–55-specific T cell blasts showed extensive infiltration of CD4⁺ T cells (Fig. 4, a and e), and widespread class II (IA^b) expression (Fig. 4, c and g). F4/80 expression, indicative of macrophage infiltration and/or microglial activation was also widespread in the wt tissues (data not shown). In contrast, spinal cord sections from DM KO mice showed no CD4⁺ T cell infiltration (Fig. 4, b and f), and no class II or F4/80 expression (Fig. 4, d and h, and data not shown, respectively). Similar to the DM KO mice, Ii-deficient mice, either MOG_35–55-primed or adoptively transferred, showed no infiltration or CNS cell activation (data not shown), confirming a lack of CNS inflammation in the absence of Ii or DM expression.

The preponderance of data supported the hypothesis that DM KO mice, and to a lesser extent, Ii KO mice do not have defective MOG_35–55-specific T cell responses. Both mutant mice could prime Th1 responses and present peptide for recall responses, yet failed to show any clinical signs of EAE. Indeed, the DM KO mice could prime and elicit a significant peripheral Th1 recall response to peptide in vivo. Thus, with the DM KO mice in particular, the defect appeared to be at the level of Ag presentation in the target organ. The Ii-deficient mice had a more profound defect in vivo; failing to elicit peripheral T cell responses, suggesting a broader defect than that seen in DM-deficient mice. Both Ii and DM deficient mice have defects in the processing of intact protein Ags, but the level of deficiency can be Ag-specific (35, 36, 37). To assess the ability of the mutant APCs to present intact MOG, splenocytes from wt, Ii KO, and DM KO mice were isolated and assessed for their ability to present peptide vs recombinant protein to a MOG_35–55-specific T cell line. Although all of the APCs could present the MOG_35–55 peptide and stimulate the line to proliferate (Fig. 7,a), only the wt APCs could process and present the rMOG protein (Fig. 7 b). Thus, resistance to EAE initiation in DM KO mice appeared to be at the level of protein Ag processing and presentation.

DM KO and Ii KO mice can differentially present peptide Ags, in an Ag-dependent manner (35, 36, 37). Although unlikely, it was possible that the defect in EAE initiation in the mutant mice could be specific to CNS processing and presentation of the MOG_35–55 peptide. We thus tested the ability of a peptide of proteolipid protein, PLP_178–191, to initiate disease in C57BL/6 mice. Similar to MOG_35–55-induced disease, wt mice developed clinical disease with 100% incidence, while DM- and Ii-deficient mice failed to show any signs of disease (Fig. 8,a). Immunohistochemical analysis of spinal cords from PLP_178–191-primed mice showed significant CD4⁺ T cell infiltration and IA^b and F4/80 expression in wt tissues, while DM KO and Ii KO spinal cords showed little evidence of pathology (data not shown). Interestingly, when measuring Th1 recall responses in vivo, both DM- and Ii-deficient mice had significant PLP_178–191 reactivities (Fig. 8,b), although the response in Ii KO mice was not as significant as the response elicited in DM KO mice (p < 0.02 and p < 0.005, respectively). Ex vivo, LN cells from PLP_178–191-primed Ii KO and DM KO mice proliferated to specific peptides in a dose-dependent fashion, albeit at reduced levels compared with wt LN cells (Fig. 8 c). Thus, the phenotype of peripheral Th1 responsiveness upon myelin peptide immunization without initiation of clinical disease in DM- and Ii-deficient mice does not appear to be an Ag-specific phenomenon.

DM KO mice displayed the ability to prime significant peripheral MOG_35–55-specific T cell responses without initiating EAE. This data, in conjunction with their inability to process rMOG protein and the failure to initiate disease following adoptive transfer of wt encephalitogenic T cells, supported the hypothesis that the failure to initiate EAE in the DM mutant is the result of a failure to process CNS protein Ags. However, there remained a slight possibility that the wt T cells transferred into the KO mice were incapable of recognizing CNS Ags processed and presented by DM KO APCs and that the peptide-primed T cell population in DM KO mice was not potent enough to initiate clinical EAE. To address this concern, we primed C57BL/6 wt and DM-deficient mice with MOG_35–55/CFA two times, 7 days apart, and compared their Th1 responses in vivo and in vitro to mice primed only once. Once again, DM KO mice were resistant to disease initiation, whether primed one or two times, while all of the wt mice developed EAE (Fig. 9,a). Significantly, the DTH response in DM KO mice primed twice was equivalent to the response in wt mice primed once (Fig. 9,b), suggesting that the magnitude of Th1 responses in the two groups was identical. ELISPOT analysis of LN cells from mice primed one or two times showed similar results, with DM KO mice primed twice displaying similar frequencies of IL-2- and IFN-γ-producing T cells as seen in wt mice primed once (data not shown). This suggested that unless there was a previously undescribed defect in T cells from DM-deficient mice, these T cells should have encephalitogenic potential. To test this hypothesis, we used DM KO mice primed twice as adoptive transfer donors. All of the wt recipients of MOG_35–55-specific DM KO T cell blasts showed significant disease, while none of the DM KO recipients had any clinical signs (Fig. 9 c), demonstrating that T cells from DM KO mice were encephalitogenic.

The critical role of CD4⁺ T cells in the pathogenesis of MS and EAE has been clearly described. Indeed, the work of Ben Nun and colleagues (38, 39) demonstrating a lack of cross-reactivity of MOG-specific T cells from IA^b and IA^bm12 congenic mice demonstrates that the class II locus is absolutely critical for MOG-specific autoimmune T cell responses. Additionally, treatment with anti-IA Abs had been shown to inhibit EAE and a variety of other autoimmune diseases (7). Yet, the importance of different APC populations and the requirement for processing and presentation of CNS protein Ags remains controversial. Numerous studies have demonstrated differential capacities of CNS-resident APC populations to up-regulate the expression of the Ag-processing machinery and/or costimulatory molecules (24, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52) and activate myelin-specific CD4⁺ T cells. We have previously shown that following IFN-γ treatment, astrocytes from SJL/J mice up-regulate the expression of the CIITA, class II, Ii, and DM, as well as a variety of costimulatory molecules (28). Additionally, after IFN-γ exposure, the SJL/J astrocytes could process the immunodominant PLP epitope, PLP_139–151, from intact protein and present it to encephalitogenic CD4⁺ T cells (28).

The CIITA has been described as the master switch for both constitutive and IFN-γ-induced class II expression. Findings from three distinct CIITA-deficient mice have demonstrated a profound, but not total, loss of class II (and accessory molecule) expression (20, 22, 23). Functional studies have shown that CD4⁺ T cells from CIITA-deficient mice cannot elicit a class II-restricted allogeneic response or prime an anti-keyhole limpet hemocyanin T cell response (22). Mora et al. (53) demonstrated that while NOD mice deficient in CIITA expression did have pancreatic infiltration of CD8⁺ T cells, B cells, and macrophages, they did not develop diabetes, although an autoaggressive CD8⁺ T cell clone could initiate a delayed diabetes independent of CIITA function. To test for the requirement of class II in EAE, we attempted to initiate disease in CIITA-deficient mice. Not surprisingly, the CIITA KO mice were resistant to initiation of EAE. This defect was not solely due to a failure in T cell activation, as adoptively transferred syngeneic encephalitogenic T cells could not initiate disease. Additionally, resistance to EAE was not the result of a failure in MOG peptide presentation, as CIITA-deficient mice could present peptide in vivo and in vitro. Because the CIITA-deficient mice express little or no class II, there was a possibility that the DTH and IFN-γ responses measured were class I-restricted. Other disease models have provided evidence supporting the role of CD8⁺ T cells in EAE, e.g. Huseby et al. (54) recently described a role for myelin basic protein (MBP)-specific CD8⁺ cytotoxic T cells in the initiation of EAE in C3H (H-2^k) mice. In a more similar model, Sun et al. (55) described a potential role for CD8⁺ T cells in the initiation of MOG_35–55-specific EAE in B6 mice. Thus, there was a distinct possibility that in the CIITA mutant mice, CD8⁺ T cells were mediating the MOG-specific immune responses. To address this possibility, we preformed immunohistochemistry on DTH lesions from MOG_35–55-primed mice. Immunohistochemical analysis of ears following DTH showed significant CD4⁺, but not CD8⁺, T cell infiltrates in MOG_35–55-challenged ears of both wt and CIITA KO mice (data not shown), suggesting that the inflammatory response was not mediated by CD8⁺ T cell responses. Additionally, in ELISPOT analysis of MOG_35–55-primed wt and CIITA KO mice, the inhibition of IL-2 and IFN-γ production by addition of anti-class II mAb (Fig. 2, e and f) argues that trace levels of class II were presenting MOG peptide to CD4⁺ T cells, and that CD8⁺ T cells were not mediating the inflammatory responses. Moreover, in vitro MOG_35–55-specific IFN-γ and IL-2 responses were maintained using column purified CD4⁺ T cells (data not shown). Lastly, both MOG_35–55- and PLP_178–191-specific proliferative responses from C57BL/6 mice were inhibited by coculture with anti-class II mAb, while the addition of anti-class I mAb had no effect (data not shown). These data show that while there are almost undetectable levels of class II proteins expressed in CIITA KO mice, that level is sufficient for the priming of some peptide-specific, peripheral CD4⁺ T cell responses. This conclusion has implications regarding the levels of class II required for the initiation of peptide-induced immune responses and suggests that a broader population of tissues may function as APCs as long as prerequisite costimulatory molecules are present in cis or in trans. This in turn may have important implications in peptide vaccine design. Our data indicate that MHC class II levels are sufficient for priming a Th1 peptide response in CIITA-deficient mice, but the mice are not susceptible to EAE initiation perhaps due to the lack of Ag-processing accessory molecules (e.g. Ii and DM) in the CNS or the inability of nonprofessional APC to express functional class II.

Ii and DM KO mice have significant defects in class II expression and/or function. In the absence of Ii, class II α- and β-chains fail to fold properly in the ER and the majority of the class II is degraded. By associating with ER-resident peptides, a small but significant population of class II αβ dimers does assemble, and is shunted directly to the cell surface by the default pathway (16, 17). Thus, Ii-deficient mice express low levels of class II that are occupied with peptides easily exchanged for high-affinity peptides at the cell surface. Loss of DM function, in contrast, does not dramatically decrease class II expression. Class II is assembled and transported to the endocytic compartment for peptide loading, but without DM, CLIP is not removed form the class II peptide-binding groove. The class II αβ/CLIP complex is transported to the APC surface where exogenous peptides could selectively exchange with CLIP and be presented (12, 15, 34, 56). Thus in both cases, there was the possibility of priming a myelin-specific T cell response and it was of considerable interest whether the specific components of the class II Ag-processing machinery were required for the activation of encephalitogenic T cells in vivo.

Professional APCs are incapable of processing and presenting the MOG_35–55 determinant from a rMOG protein in the absence of Ii or DM. Additionally, while the efficiency of peptide presentation by mutant APCs is decreased at lower Ag doses, both Ii- and DM-deficient mice can present both MOG_35–55 and PLP_178–191. Interestingly, the Ii-deficient mice had little response to MOG_35–55 challenge in vivo, yet responded more vigorously to in vitro Ag restimulation than DM KO LN cells. This is likely due to the enhanced ability of mutant APCs to present peptide Ags in vitro, which is often more efficient in Ii KO APCs than peptide presentation by wt APCs (16). In vivo, we established that both Ii- and DM-deficient mice are resistant to initiation of EAE by active priming with either MOG_35–55 or PLP_178–191. In addition to defective in vivo Ag presentation, DM- and Ii-deficient mice have additional defects that could potentially contribute to disease resistance. Although both mutant mouse strains have altered CD4⁺ T cell development (12, 16), the inability to transfer disease to KO mice using wt MOG_35–55-specific encephalitogenic T cell blasts, in conjunction with the demonstration of encephalitogenicity of MOG_35–55-specific T cell blasts from DM-deficient mice when transferred into wt mice, argues against a T cell defect preventing disease. Indeed, the ability of DM KO mice to prime a T cell response that is encephalitogenic in wt mice indicates that, in this case, there is not a defect in the periphery of these mice. Ii KO mice, in contrast, have additional deficiencies; in particular, defective B cell maturation (57). However, it is unlikely that the B cell defect in Ii-deficient mice mediated resistance to active EAE, as Lyons et al. (58) demonstrated B cell deficient mice to be susceptible to MOG_35–55-induced EAE. Moreover, Lyons and colleagues demonstrated that B cell-deficient mice had similar clinical symptoms and disease severity as compared with wt mice, suggesting that the loss of Ag presentation function in Ii-deficient mice, and not the absence of B cells, is responsible for the resistance to MOG_35–55-induced EAE.

Activated T cells can cross the blood-brain barrier independent of Ag specificity, and traffic through the CNS (59). If the T cells encounter their cognate Ag (myelin peptide plus class II MHC), they remain in the CNS and execute their effector function. Activated T cells that enter the CNS express the necessary inflammatory mediators to up-regulate the expression of class II and costimulatory molecules (e.g. IFN-γ) in a variety of CNS cells. It is unclear whether the initial source of peptide Ag is derived from free myelin peptides present in the CNS, or if myelin proteins are endogenously processed and presented by CNS APCs. In support of the former possibility, Krogsgaard et al. (60) used a mAb specific for DR2 complexed to the encephalitogenic MBP_85–99 peptide, demonstrating that the naive CNS tissues display low levels of class II/myelin Ags. We have shown that myelin peptides can be presented in vitro, to myelin specific CD4⁺ T cells, independent of Ii or DM expression. Moreover, we have demonstrated that DM- and Ii-deficient mice can prime and elicit a myelin peptide-specific DTH response, which is indicative of a Th1-mediated T cell response (61, 62). However, adoptively transferred encephalitogenic T cells blasts, which can recognize peptides presented by DM- and Ii-deficient APCs, and can infiltrate the CNS (data not shown; Ref. (59)) fail to initiate EAE and do not remain in the CNS (Fig. 4 f). These data support the hypothesis that free myelin peptides are not present in naive CNS tissues of the various KO mice displaying defects in the ability to process intact myelin proteins and that de novo endocytic processing of myelin Ags and subsequent presentation by CNS APCs is required for the reactivation of encephalitogenic T cells in the target organ, enabling effector function and the initiation of EAE.

The observed absence of class I-restricted, CD8⁺ T cell responses in MOG_35–55-primed wt and mutant mice contrasts with the work of Sun et al. (55). Our data suggest that potentially MOG-specific, CD8⁺ T cells cannot initiate disease in C57BL/6 mice. Sun et al. (55) proposed that CD4⁺ T cell help is required to elicit active MOG_35–55-specific CD8⁺ T cell-mediated disease. We demonstrated that DM-deficient mice prime a potent peripheral CD4⁺ T cell response, yet we failed to observe any disease symptoms or CNS infiltration of CD8⁺ T cells, and could not elicit any MOG-specific CD8⁺ T cell responses. Moreover, Sun et al. (55) suggested that as few as 5 × 10⁵ previously activated MOG_35–55-specific CD8⁺ T cells could adoptively transfer disease independent of CD4⁺ T cell (and presumably class II) function. Our findings in the B6 CIITA-, Ii-, and DM-KO mice contrast with these results. In each adoptive transfer experiment, an average of 10 × 10⁶ CD8⁺ T cells elicited in wt mice and sensitized to MOG_35–55 were transferred along with the encephalitogenic CD4⁺ MOG-specific T cells into the various KO mice, all of which express normal levels of MHC class I. In all experiments, there was never any incidence of disease (Figs. 1 b, 3b, 9c, and data not shown) or evidence of CNS infiltration of CD8⁺ T cells. In support of our conclusions, a recent publication using green fluorescent protein-expressing cells to track adoptively transferred MOG_35–55-specific T cells in B6 mice demonstrated that while a significant proportion of transferred cells were CD8⁺, only CD4⁺ green fluorescent protein-positive cells were detected in the CNS of recipient mice (63). Although our results and those of others downplay the role of CD8⁺ T cells in our disease model, they do not conflict the work of Huseby et al. (54), as the authors specifically primed the class I Ag-processing pathway with a vaccinia virus expressing MBP to generate the encephalitogenic T cells from C3H.shi (MBP-mutant), H-2^k mice. In a disease model distinct from MOG-specific EAE in C57BL/6 mice, Huseby and colleagues demonstrate that CD8⁺ T cells can mediate encephalomyelitis. Their findings have important implications regarding factors contributing to autoimmune disease and MS in particular, but do not directly impact this study.

After submission of this manuscript for publication, Slavin et al. (64) published a paper entitled “Requirement for endocytic Ag processing and influence of invariant chain and H-2 M deficiencies in CNS autoimmunity.” Similar to the results shown in this study, the authors use Ii- and DM-deficient mice to demonstrate the requirement of Ag processing in the CNS for the initiation of EAE in C57BL/6 mice by either active priming or adoptive transfer. We are very pleased by the similarities in results, as we strongly support each other’s conclusions.

We thank Ann M. Girvin and Todd N. Eagar for many fruitful discussions and technical assistance.