IFN-γ and IL-17 Production in Experimental Autoimmune Encephalomyelitis Depends on Local APC–T Cell Complement Production¹

Experimental autoimmune encephalomyelitis (EAE)³ (1) is a rodent model that has been valuable for characterizing immunopathogenic processes that underlie multiple sclerosis (MS). One way that disease is induced is adjuvant and pertussis toxin-boosted immunization of C57BL/6 mice with the encephalitogenic peptide (p35–55) of myelin oligodendrocyte glycoprotein (MOG), an external myelin component. Following this immunization, myelin sheaths of oligodendrocytes are attacked (2).

According to current concepts (3), myelin destruction in EAE (and MS) is caused by T cells autoreactive with myelin, which are peripherally activated, and, upon encountering microglia-bearing myelin peptides in the brain, become further activated (4). A feature of EAE pathogenesis that closely relates to MS pathogenesis is that autoreactivity induced against the myelin immunogen, e.g., MOG, broadens to other myelin components, e.g., proteolipid protein (PLP) and myelin basic protein (MBP). This process, termed epitope spreading, is thought to be due to endogenous priming to new self determinants during initial and subsequent inflammatory interactions (5). For reproducing it, rather than the C57BL/6 H-2^b strain, the autoreactive-prone SJL H-2^s strain generally is used.

In previous studies (6), we found that decay-accelerating factor (DAF or CD55), a regulator originally characterized as a plasma membrane shield that circumvents C3b deposition on self cell surfaces and prevents C5 activation (7), modulates T cell responses (see Discussion). In its absence on APCs or on cognate CD4⁺ and CD8⁺ T cells interacting with them, proliferative and IFN-γ responses are 5- to 22-fold augmented. Findings that double deficiency of DAF and factor D (fD), an alternative pathway component, abolished the enhancement in vitro as well in vivo linked the effect to complement locally produced by APC–T cell partners. Prompted by this result, we analyzed flow-sorted wild-type (WT) APCs and T cells following mixing with specific peptide and found that early during cognate interactions, both partners locally synthesize all three alternative pathway components, i.e., C3, factor B, and fD, together with C5 as well as C5aR and C3aR. We found that concomitant with this local complement synthesis, both APCs and T cells down-regulate surface DAF expression, lifting restraint on junctional activation. We further found that C5a and C3a are locally generated by both partners, and that the interaction of these activation fragments with C5aR and C3aR on both partners promotes both APC and T cell cytokine production (8). Although other investigators have reported that autoimmune diseases are diminished in the absence of C5aR or C3aR and attributed the findings to effects of serum complement (9, 10, 11), we showed that APCs and T cells locally synthesize complement, and as a result of lifted DAF restraint on junctional activation C5a and C3a are locally generated and their interactions with C5aR and C3aR promote cytokine production by both partners. This prompted us to investigate whether interaction of locally produced C5a and C3a with C5aR and C3aR on APCs and T cells impact autoimmunity and, if so, how.

In the present investigation, we exploited Daf-deficient mice in the MOG_35–55-induced EAE model to study the following: 1) the extent to which DAF influences T cell autoreactivity to self Ags; 2) whether its presence is necessary to protect against broadening of the anti-myelin T cell response; 3) the nature of the T cell response elicited in its absence; and 4) the mechanism of any observed differences.

Eight- to 12-wk-old female Daf1^−/− mice (12) and their Daf1^+/+ female littermates backcrossed five generations on the C57BL/6 background were used. Although mice possess a second Daf gene (Daf2 (13)), it encodes transmembrane-anchored Daf expressed almost exclusively in the testes. Studies were conducted using an approved Institutional Animal Care and Use Center protocol.

Rat MOG_35–55 MEVGWYRSPFSRVVHLYRNGK, the immunodominant encephalitogen for C57BL/6 mice (14), was synthesized using standard solid-phase methodology and 9-fluorenylmethoxycarbonyl side chain-protected amino acids, purified >90% by reverse-phase HPLC, and confirmed by mass spectrometry. PLP was purified from a washed total lipid extract of human white matter and converted to aqueous form, as described (15). Mouse MBP was purchased from Sigma-Aldrich. Mouse IL-4 and GM-CSF were purchased from PeproTech. Anti-CD3, anti-CD28, anti-IFN-γ, and anti-IL-4 mAbs were purchased from BD Pharmingen.

Animals were injected s.c. with 100 μg of MOG_33–35 in CFA containing 400 μg of Mycobacterium tuberculosis H37RA (Difco). Upon immunization and 2 days later, 200 ng of pertussis toxin (List Biological Laboratories) was injected i.p. Mice were weighed and scored for neurological deficits daily, as follows: 0 = no disease; 1 = decreased tail tone or slightly clumsy gait; 2 = tail atony; 3 = limb weakness; 4 = limb paralysis; and 5 = moribund state.

Eighteen days after priming 8- to 12-wk-old female Daf1^−/− or Daf1^+/+ mice with MOG_35–55 in CFA containing H37RA, splenocytes were cultured for 72 h with 20 μg/ml MOG_35–55 and 10 ng/ml IL-12 or IL-23. In initial studies (3 × 10⁷), washed cells from Daf1^−/− or Daf1^+/+ mice were administered i.p. into Daf1^+/+ recipients that received 400 rad of γ irradiation. Guided by 4- or 3-fold higher numbers of IFN-γ- or IL-17-producing cells, in subsequent studies, washed spleen cells (40 or 30 × 10⁶) from Daf1^+/+ mice vs 10 × 10⁶ from Daf1^−/− mice were transferred to obtain equal numbers of IFN-γ- or IL-17-producing cells. Clinical scores and weights were monitored, as above.

Spinal cords were collected at days 18 and 56 from 4–8 mice in each group, fixed after infusion with 100 ml of 4% paraformaldehyde for 5 min, embedded in OCT, quick frozen in liquid N₂, and stored at −70°C. Cryostat sections (10–20 μm) were evaluated at 10 levels. H&E staining was done by standard methods. Alcohol-dehydrated sections were stained for myelin in Luxol Fast Blue (Sigma-Aldrich) overnight at 37°C and agitated for 10 s in 0.0005% lithium carbonate and 60 s in 70% alcohol.

Spinal cords were fixed in formaldehyde/glutaraldehyde (3%)/0.1 M sodium cacodylate, postfixed in 1% osmium tetroxide, and embedded in LX112 resin (Polysciences). Ultrathin uranyl acetate/lead citrate-stained sections were cut and examined with a JEOL 100cx microscope.

Splenocytes (2 × 10⁵ cells in 200 μl) were cultured for 96 h in triplicate in 96-well microtiter flat-bottom plates with 0.01–100 μg/ml MOG_35–55 and pulsed for 16 h with methyl-[³H]thymidine (1 μCi/well; sp. act., 6.7 Ci/mmol; New England Biolabs), and incorporated radioactivity was expressed as the mean cpm of triplicate Ag-containing wells. In ELISPOT assays, splenocytes (2–6 × 10⁵) were incubated for 48 h in anti-IFN-γ- or IL-17 mAb (R&D Systems)-coated and PBS-1% BSA-blocked 96-well ELISPOT plates (Whatman) with 0.01–20 μg/ml MOG 35–55, 0.01–50 μg/ml PLP, or 100 μg/ml MBP, and spots were developed and quantitated on an ELISPOT analyzer (Cellular Technology).

Bone marrow cells from mice were grown in RPMI 1640 (containing 10% FBS, 10 ng/ml IL-4, and 10 ng/ml GM-CSF). The medium was changed on days 3 and 5. Cells were used on day 6.

Total RNA was purified by Qiagen RNeasy mini kit (Qiagen). cDNA was synthesized from 20 μl of mRNA in Sprint PowerScript Single Shots (BD Clontech). Diluted cDNA (10 μl) was mixed with 2 μl of primer and 10 μl of SYBR Green master mix (Applied Biosystems). Each sample was assayed in duplicate on an ABI prism 7000 cycler. In all assays, fold increases are relative to basal levels and standardized to actin.

All blots were performed as previously described (12), using HRP-conjugated secondary Ab and an ECL enhancer (GE Healthcare).

Lymph node cells were harvested from five Daf1^+/+, five Daf1^−/−, and five Daf1^−/−C5aR^−/−C3aR^−/− mice, and naive CD4⁺CD62L^high cells were sorted. The sorted (CD4⁺CD62L^high) T cells were stimulated with anti-CD3 (5 μg/ml), anti-CD28 (10 μg/ml), anti-IFN-γ (10 μg/ml), and anti-IL-4 (10 μg/ml) Abs together with TGF-β (20 ng/ml) or TGF-β and IL-6 (20 ng/ml). IL-17 production in culture supernatants was measured by the Beadlyte Mouse 21-plex Cytokine Detection System (Upstate Biotechnology) after 72 h of activation.

A repeated measures cumulative logit model (16), using the equation, log it [P (Y ≤ j x)] = α_j + β₁genotype + β₂time + β₃genotype × time, was used to test for the differences in groups across time points, with α_j corresponding to the P (Y ≤ j) for each threshold or score and β corresponding to the coefficients to be estimated for the main effects of genotype and time, as well as for the interaction between genotype and time. Calculations were performed on SAS for Windows 9.1. Statistics for the in vitro studies were done by Student’s t test.

To determine how DAF deficiency influences CNS neuronal injury in EAE, we immunized 12 Daf1^−/− mice and 12 Daf1^+/+ littermates with MOG_35–55 and monitored them daily. Although all mice in both groups developed EAE, neurologic dysfunction in Daf1^−/− mice was markedly more severe. Fig. 1 A shows clinical scores over a 62-day period from the above mice combined with 12 more Daf1^−/− and Daf1^+/+ mice from an identical repeat experiment (total of 24 mice in each group). In the Daf1^−/− mice, the mean clinical score was 3.5 ± 1.1 vs 1.7 ± 1.3 in Daf1^+/+ littermates (p < 0.001), and the mean weight decrease (over days 10–62) was greater (91.2 ± 2.8% of original weight vs 97.8 ± 3.4%, p < 0.05) (data not shown). Although by day 40 onward most Daf1^+/+ mice exhibited only minimal physical changes, some Daf1^−/− mice had to be euthanized because they barely moved.

To document that the heightened disease severity in Daf1^−/− mice was mediated by increased cellular immunity (because DAF also inhibits cytotoxicity conferred by systemic complement), we performed adoptive transfer experiments in which we injected equal numbers of splenocytes from MOG_35–55-immunized Daf1^−/− or Daf1^+/+ mice into naive Daf1^+/+ recipients. As seen in Fig. 1 B, compared with Daf1^+/+ recipients that received Daf1^+/+ splenocytes, Daf1^+/+ recipients that received Daf1^−/− splenocytes showed ≥2-fold higher clinical scores and more profound weight decreases (data not shown). The difference in disease severity was comparable to that following direct immunization with MOG_35–55.

In contrast to the enhancement of T cell responses in Daf1^−/− mice, assays of day 56 sera showed that both groups elaborated similar amounts of anti-MOG Abs. Moreover, Daf1^−/− mice did not generate more complement-fixing IgG2a, IgG2b, or IgM (data not shown). Staining of (perfused (see Materials and Methods)) spinal cords (day 56) for deposited C3 and C9 showed that whereas some deposition of both components was detectable in Daf1^−/− mice and Daf1^+/+ littermates, pixel counts showed no significant differences (650 ± 123 vs 530 ± 120 for C3 (p > 0.05) and 630 ± 194 vs 570 ± 121 for C9 (p > 0.05) (data not shown)).

To compare pathology, we examined dorsal columns of Daf1^−/− and Daf1^+/+ mice at early (day 18) and late (day 56) disease stages. At the day 18 time point, H&E staining showed markedly increased leukocyte infiltration in Daf1^−/− mice, which on high power examination were >90% lymphocytes and macrophages (Fig. 2, A and B). Luxol Fast Blue staining at both days 18 and 56 showed markedly more myelin loss in Daf1^−/− mice (Fig. 2, C and D). Ultrastructural analyses at both the early and late time points confirmed markedly increased destruction of myelin sheaths (data not shown). Moreover, whereas at late stages of disease (day 84) Daf1^+/+ mice demonstrated remyelination (as indicated by small thinly myelinated axons in lesional areas (Fig. 2,E)), little, if any, remyelination was detectable in the Daf1^−/− mice (Fig. 2 F).

To determine the immunological basis for the heightened disease in Daf1^−/− mice, we measured T cell responses to MOG_35–55 peptide. In [³H]thymidine incorporation assays (data not shown) at day 28 postimmunization, Daf1^−/− splenic T cells expanded 5-fold more than Daf1^+/+ spleen cells (2000 vs 400 cpm/10⁵ spleen cells at 1 μg/ml MOG_35–55). IFN-γ and IL-17 ELISPOT assays at 30 days showed 5- to 6-fold more IFN-γ-producing cells and 2-fold more IL-17-producing cells than in Daf1^+/+ spleens (Fig. 3). Later in disease (60 days), these assays showed 4- to 5-fold more IFN-γ-producing cells and 5- to 7-fold more IL-17-producing cells than in Daf1^+/+ spleens. At an even later time point (75 days), the ELISPOT assays showed that the ratio of IL-17-producing cells to IFN-γ-producing cells in Daf1^+/+ mice remained the same, whereas the ratio of IL-17-producing cells to IFN-γ-producing cells in Daf1^−/− mice increased from 1:2 to 3:1, indicating that the IL-17 response in Daf1^−/− mice is more sustained. ELISAs of supernatants of spleen cells from Daf1^−/− and Daf1^+/+ mice at day 56 showed increases in IL-2, IFN-γ, and IL-17, and decreases in IL-5 and IL-10 in the Daf1^−/− mice, consistent with augmented Th1 and Th17 responses (data not shown). IFN-γ and IL-17 ELISAs of cultured spleen supernatants at day 56 verified 3- to 4-fold elevated levels of IFN-γ and 5- to 6-fold elevated levels of IL-17 (data not shown). Computer analyses of mean ELISPOT sizes showed similar IL-17 spot sizes, but larger IFN-γ spot sizes (Fig. 3,C). Repeat adoptive transfer experiments this time adjusting spleen cells for equal numbers of IFN-γ- or IL-17-producing cells showed that, even under these conditions, Daf1^−/− spleen cells in each case conferred greater pathology (Fig. 3 D).

To determine whether the enhanced immunoreactivity in Daf1^−/− mice was sufficient to overcome the normal resistance in H-2^b mice to epitope spreading (5), we performed IFN-γ ELISPOT recall assays at days 13–64 with PLP and MBP. Consistent with past literature (17), no PLP reactivity was observed with spleen cells from Daf1^+/+ controls (Fig. 4,A). In contrast, dose-dependent PLP reactivity (Fig. 4,A) developed between day 56 and 64 (Fig. 4,B) in Daf1^−/− mice. As with PLP, Daf1^−/− spleen cells also showed reactivity to MBP (Fig. 4 C). Reactivity to neither PLP nor MBP was detectable in spleens of unprimed Daf1^−/− mice, verifying that the elicited autoreactives were dependent upon MOG immunization.

Our previous studies (6) showed that when APCs and cognate T cells interact, both partners turn on the local synthesis of complement and concomitantly down-regulate DAF. Because DAF regulates C3/C5 convertases (7), prior reports have suggested that the anaphylatoxin receptors C3aR and C5aR can be involved in T cell-mediated disease models (9, 10), and we showed that exogenous C5a augments alloreactive T cell immunity (6), we tested whether increased local C5a/C3a–C5aR/C3aR interactions by Daf1^−/− APCs and T cells can account for the greater anti-MOG autoreactive response in Daf1^−/− mice as well as explain previous findings by others with C5aR^−/− and C3aR^−/− mice attributed to cross-talk from serum complement.

To determine whether increased local C5a/C3a–C5aR/C3aR interactions by Daf1^−/− APCs and T cells is involved in the increased IFN-γ and IL-17 production in Daf1^−/− mice and, if so, how the two processes are mechanistically linked, we performed two sets of in vitro studies. In one, we stimulated transgenic OT-II T cells with Daf1^+/+, Daf1^−/−, and Daf1^−/−C3aR^−/−C5aR^−/− DCs in the presence of specific OVA_323–339 peptide and examined the production of IL-12 and IL-23 by the DCs. In the other, we stimulated Daf1^+/+, Daf1^−/−, Daf1^−/−C3aR^−/−, and Daf1^−/−C5aR^−/− T cells with anti-CD3 and anti-CD28 mAbs and assessed the effects of IL-12 or IL-23 addition on T cell IFN-γ and IL-17 production. The two sets of studies showed the following: 1) Daf1^−/− DCs produced 4-fold more IL-12 and 5-fold more IL-23 than Daf1^+/+ DCs (Fig. 5,A); 2) the addition of IL-12 to anti-CD3/CD28-stimulated Daf1^−/− T cells produced 4-fold more IFN-γ-producing cells, and the addition of IL-23 produced 5-fold more IL-17-producing cells compared with Daf1^+/+ controls (Fig. 5,B); 3) the heightened responses of Daf1^−/− T cells were associated with up-regulation of their IL-12Rβ2 and IL-23R levels (Fig. 5,C); and 4) the augmented IFN-γ and IL-17 production by Daf1^−/− T cells depended on T cell C3aR and C5aR (Figs. 5,B and 6 C).

In support of the dependence of the augmented IFN-γ and IL-17 responses on C5aR and C3aR on both APCs and T cells: 1) the increased production of IL-12 and IL-23 by Daf1^−/− APCs was abrogated in Daf1^−/−C3aR^−/−C5aR^−/− triple-knockout APCs (Fig. 6,A); 2) Western blots of supernatants of anti-CD3/CD28-stimulated Daf1^+/+, Daf1^−/−, Daf1^−/−C3aR^−/−, and Daf1^−/−C5aR^−/− T cells showed that Daf1^−/− cells, but not Daf1^−/−C3aR^−/− or Daf1^−/−C5aR^−/− cells, produced more C3a and C5a (the ligands for C3aR and C5aR) than Daf1^+/+ cells (Fig. 6,B); 3) the increased IFN-γ and IL-17 responsiveness of Daf1^−/− T cells to IL-12 and IL-23 was abrogated in Daf1^−/−C5aR^−/− and Daf1^−/−C3aR^−/− mice (Fig. 5,B); 4) ELISPOT assays showed that the increased numbers of IFN-γ- and IL-17-producing cells generated by anti-CD3/CD28-stimulated Daf1^−/− T cells were abolished with Daf1^−/−C5aR^−/− and Daf1^−/−C3aR^−/− T cells (Fig. 6,C); and 5) ELISAs showed that the augmented IFN-γ and IL-17 also depended on C3 and fD (Fig. 6 D).

Recent studies (18) of sorted naive CD4⁺CD62L^high cells have shown that IL-17-producing cells are not generated directly from IL-23 stimulation, but rather are initially induced by IL-6 and TGF-β. Once differentiated by these cytokines, IL-23 augments and sustains IL-17 production. To ascertain how the results in this study relate to this finding, we flow sorted Daf1^+/+ and Daf1^−/− CD4⁺CD62L^high cells and incubated the sorted cells with anti-CD3 and anti-CD28 mAbs for 72 h in the presence of IL-6 and TGF-β together with anti-IFN-γ and anti-IL-4 mAbs, exactly as described in the above published study (18). An ELISA of supernatants (Fig. 7 A) of the treated cells showed 3 times more IL-17 production by naive Daf1^−/− T cells than by naive Daf1^+/+ cells. Compared with WT T cells, IL-17 production was reduced 4-fold when Daf1^−/−C3aR^−/−C5aR^−/− cells were used.

To document that the above in vitro linkage of C5a/C3a–C5aR/C3aR interactions with T cell IFN-γ and IL-17 production applies in vivo and correlates with EAE disease severity, we compared clinical scores in Daf1^−/−, Daf1^−/−C5aR^−/−, Daf1^−/−C3aR^−/−, and Daf1^+/+ mice following immunization with MOG_35–55. Consistent with the in vitro data in Fig. 6, the increased disease severity in Daf1^−/− mice over that in Daf1^+/+ mice was not only abrogated in Daf1^−/−C5aR^−/− and Daf1^−/−C3aR^−/− mice, but clinical scores went below those of Daf1^+/+ mice (Fig. 7 B).

Taken together, the data indicate that heightened disease activity in Daf1^−/− mice is due to the following: 1) heightened production of IL-12/IL-23 by Daf1^−/− APCs and 2) heightened IFN-γ and IL-17 responses by Daf1^−/− T cells to the two APC signal 3 cytokines, resulting (at least in part) from higher induced levels of IL-12β2R and IL-23R. The studies with Daf1^−/− cells doubly deficient in C5aR or C3aR show the following: 1) the augmented cytokine production by both partners depends on APC–T cell C5a/C3a–C5aR/C3aR interactions, 2) the dependence on these interactions applies in vivo in that they control EAE disease severity, and 3) initial commitment into the IL-17 lineage depends on these interactions.

The results of this study document an important role for DAF function in shielding against T cell autoreactivity. T cell autoimmunity in Daf1^−/− mice immunized with MOG_35–55 is stronger; EAE is more severe; more T cells are present in the CNS; myelin loss is more extensive; and remyelination is impaired. That the heightened CNS injury is, in fact, mediated by augmented T cell reactivity was formally established by adoptive transfer. The autoreactive response is both quantitatively and qualitatively different in that it is not only associated with more IFN-γ-producing cells, but even more IL-17-producing cells. As assessed by both proliferation and by IFN-γ and IL-17 ELISPOT assays, the increased lymphocyte infiltration in Daf1^−/− mice is due to up to a 5-fold more robust T cell IFN-γ response and 7-fold more robust T cell IL-17 response to the priming by MOG immunogen. When equalized for IFN-γ-producing cells or IL-17-producing cell numbers, adoptively transferred spleen cells induce greater disease severity. As a result, despite the inherent resistance of the H-2^b background to epitope spreading, the enhanced autoreactivity allows intermolecular spreading to PLP and MBP. Taken together, the findings argue that DAF serves as a barrier to the induction of T cell autoreactivity in EAE, and that its protection is conferred via inhibition of the induced T cell autoreactivity to myelin and spreading of the autoreactivity to other myelin components.

Until this report, significant epitope spreading has been observed primarily in SJL/J or SWXJ mice and has not been extensively described in C57BL/6 H-2^b mice. Our findings of markedly stronger T cell responses in C57BL/6 H-2^b Daf1^−/− mice and spreading of autoreactivity to PLP and MBP are consistent with the observed augmented myelin destruction. To our knowledge, the finding that T cell reactivity to PLP and MBP is induced by MOG_35–55 in the absence of DAF, i.e., that the barrier to extension of T cell reactivity to myelin components other than the MOG_35–55 immunogen is overcome, constitutes the first definitive demonstration of H-2^b-restricted intermolecular epitope spreading.

The adoptive transfer experiments (Fig. 1 B) showed that the enhanced disease was directly related to Daf1^−/− donor cells. Thus, priming to self in a DAF-deficient milieu produces an uncontrolled hyperautoreactivity that leads to severe disease sequelae. Although it is clear that DAF serves as a global negative regulator of T cell activation, the overall permissive impact of DAF deficiency in EAE severity is the overrepresentation of the Th17 as well as the Th1 lineage. Although the association is not causally proven in this study, the linkage of severe EAE outcome to increased frequencies of IL-17-producing T cells is consistent with many recent studies showing the following: 1) greater autoimmune severity following transfer of such Th17 T cells; 2) amelioration of disease following IL-17 neutralization in WT mice; and 3) reduced EAE susceptibility in IL-17^−/− mice (19, 20, 21, 22). It is possible that DAF expression may tip this fine balance away from IFN-γ- and Th17-mediated enhanced disease, and that unregulated local complement production by APCs, T cells, or both may shape the autoimmune repertoire by allowing excessive selection of the IFN-γ and Th17 lineages. In support of this, in addition to the number of anti-MOG_35–55 IL-17-producing cells being greater in the Daf1^−/− mice, the rate of decline in the number of these cells was slower than in Daf1^+/+ mice, possibly accounting for the absence in these mice of myelin repair. Thus, DAF may serve overall to prevent or limit severe autoimmune outcomes due to IFN-γ- and particularly to Th17-mediated events.

Our studies with Daf1^−/− DCs and T cells doubly deficient in Daf1^−/−, C5aR^−/−, C3aR^−/−, or both relate the mechanism underlying DAF’s T cell immunomodulatory activity to our recently uncovered findings (6) that an early event (<1 h) during APC–T cell cognate interactions is that both partners turn on local synthesis of C3, factor B, and fD (the three alternative pathway components), and down-regulate DAF. More recent work (8) has shown that both cognate partners additionally synthesize C5, up-regulate C5aR and C3aR, and, in the context of diminished junctional DAF, generate C5a and C3a from the locally synthesized components. This recent work has shown that interactions of these locally generated anaphylatoxins with up-regulated C5aR and C3aR in both partners confer costimulation. The in vitro experiments in this study with Daf1^−/−, Daf1^+/+, and doubly deficient Daf1^−/− C5aR^−/− or Daf1^−/−C3aR^−/− DCs and T cells show that the augmented IFN-γ and IL-17 production in Daf1^−/− mice is due to the following: 1) more IL-12 and IL-23 production by Daf1^−/− DCs than WT DCs; 2) up-regulated expression of IL-12Rβ2 and IL-23R on Daf1^−/− T cells; and 3) increased production of IFN-γ and IL-17 of Daf1^−/− T cells in response to IL-12 and IL-23, respectively. The Western blot analyses of culture supernatants showing increased C5a and C3a production by Daf1^−/− T cells and the ELISPOT assays and ELISAs showing reversal of the heightened IFN-γ and IL-17 production when Daf1^−/−C5aR^−/− or Daf1^−/−C3aR^−/− T cells were used document that the T cell autoreactivity is dependent on local C5a/C3a generation and consequent C5aR/C3aR signaling. Although the work in this study was done with Daf-deficient cells, our previous findings in WT cells (6) that in concert with local complement/C5aR/C3aR synthesis, DAF down-regulates on both APCs and T cells, establish that Daf1^−/− T cells simulate activation events in normal cells. Because the heightened reactivity of T cells occurs in the absence of DAF on APCs (6) and T cells (8), this newly uncovered process is not related to older studies (23, 24), putatively linking DAF to direct signaling per se.

Our studies with naive CD4⁺CD62L^high cells relate our results to recent findings (25) that differentiation of naive T cells into the Th17 lineage is dependent on TGF-β and IL-6. It has been shown that TGF-β mediates up-regulation of IL-23Rs on T cells (18, 26, 27, 28, 29, 30). Our incubations of sorted CD4⁺CD62L^high naive Daf1^−/− and Daf1^+/+ T cells with TGF-β and IL-6 showed that Daf1^−/− T cells produce 3-fold more IL-17. Our studies with Daf1^−/− C5aR^−/−C3aR^−/− cells showed that the increases depended on C5aR/C3aR function. More work is needed to determine the source of these cytokines and whether local complement synthesis is involved in their production.

Humoral immunity against myelin has been implicated in EAE, but its involvement generally relates to induction of disease with whole protein rather than peptide (31). When immunized with whole MOG protein, B cell-deficient mice develop less severe EAE (31). In contrast, if MOG_35–55 is used, demyelination in the B cell knockouts is similar to that in WTs (32). The experiments in this study showing no significant difference in complement-fixing Igs or C3/C9 deposition document that DAF’s activity on T cells is distinct from DAF’s traditional self cell-shielding role against systemic complement-mediated injury. Lower clinical scores in C3^−/− mice (33) and in C3aR^−/− mice (11) are consistent with the results in this study concerning locally generated complement activation fragments enhancing activation during T cell priming (6). Whether the findings in this study that DAF participates in protecting myelin from autoreactive T cells in EAE is a general phenomenon for autoimmunity and T cell reactivity in other contexts must await further work. Results available to date with corneal transplantation (34) would support this possibility.

In MS, like EAE, MOG is a target autoantigen (35). As in EAE, MHC class II-restricted T cells reactive with MOG are present in cerebrospinal fluid of MS patients (36). Currently, acute attacks are treated with i.v. steroids, IFN-β, Copaxone, and immunosuppressives, but, overall, available treatments are unsatisfactory (37). Our findings that markedly enhanced EAE occurs in Daf1^−/− mice due to lowering the threshold to T cell autoreactivity raise the possibility that appropriately targeted rDAF could have therapeutic relevance for MS.

We thank Dr. B. P. Morgan (Cardiff, U.K.) for rabbit anti-rat C9, Dr. Mireya Diaz-Insua (Department of Statistics, Case Western Reserve University) for statistical analyses, and Denny Hatala for preparing tissue sections.

The authors have no financial conflict of interest.