CD8α⁺ dendritic cells and immune protection from experimental allergic encephalomyelitis

Abstract

Dendritic cells (DC) represent a phenotypically heterogeneous population endowed with two important biological functions, immunity and tolerance. Here we report that the injection of splenic CD8α⁺ DC, derived from rats with experimental allergic encephalomyelitis (EAE), delayed the onset and suppressed the severity of EAE in Lewis rats. This was accompanied by the lack of magnetic resonance imaging (MRI) lesions in the brain and spinal cord and by reduced numbers of inflammatory cells within the central nervous system. Injection of CD8^α+ DC inhibited T cell proliferation that may relate to increased interferon (IFN)-γ and nitric oxide production. Although CD8⁺CD28^– suppressor T cells, apoptotic cells and co-stimulatory molecules were not altered, CD4⁺ T cells expressing interleukin (IL)-10 were augmented in rats receiving CD8α⁺ DC compared to rats receiving total DC or medium. These results demonstrate that rat splenic CD8α⁺ DC could provide a cellular basis for a novel, individualized immunotherapy using autologous DC as a complement to conventional therapy in diseases with an autoimmune background such as multiple sclerosis.

INTRODUCTION

Multiple sclerosis (MS) is an inflammatory demyelinating disease of the central nervous system (CNS). Pathologically, MS is characterized by multiple, sharply demarcated foci of demyelination and mononuclear leucocyte infiltration. The latter indicates the probability that demyelination is immune-mediated. Traditional therapy of MS with immunosuppressive drugs and even with newer disease-modulating compounds [interferon (IFN)-β, glatiramer acetate] is only partly satisfactory. We have also evaluated a variety of therapeutic regimens in experimental allergic encephalomyelitis (EAE), a model of human MS, including administration of the cytokines interleukin (IL)-10, transforming growth factor (TGF)-β1 and IL-4 as well as mucosal tolerance induced by autoantigens [1,2]. However, these strategies prevented the development of EAE only when given before immunization, but did not treat ongoing EAE when given after immunization.

Dendritic cells (DC) are considered nature's adjuvants, potent stimulators of naive T cells and key inducers of primary immune responses. In recent years it has become clear that DC not only control immunity but also maintain tolerance to self-antigens, i.e. two complementary functions that would ensure the integrity of the organism in an environment full of pathogens. In our previous studies, rat splenic DC derived from healthy rats, exposed to IFN-γ in vitro, were used to treat ongoing EAE successfully in acute and chronic EAE models [3]. However, if DC were derived from EAE animals, IFN-γ-exposed DC did not suppress the development of clinical EAE, thereby limiting the option that autologous DC might be used to treat MS or other diseases with autoimmunity background.

Murine CD8α⁺ DC were described initially as major producers of IL-12 [4,5], and expressed less CD11b [6]. Two subsets of CD8α^– and CD8α⁺ DC have been referred to as myeloid and lymphoid DC, respectively. To date, the function of the high-level expression of CD8α by this DC subtype is still unknown. Based on the fact that CD8α⁺ DC express high levels of adhesion molecules and lack the CD8β chain, it is possible that CD8α⁺ DC may play a role in cell adhesion and phagocytosis. Recent evidence demonstrates that murine CD8α⁺ DC are responsible for inducing peripheral tolerance to tissue-associated antigens [7,8]. CD8α⁺ DC can restrain T cell proliferation by limiting IL-2 production [9] and by killing CD4⁺ T cells [10]. CD8α⁺ DC were very weak stimulators of resting or activated allogeneic T cells despite their mature phenotype and equivalent expression of MHC and co-stimulatory molecules [11].

The activation state of DC is also crucial for the functional outcome of the DC–T cell interaction. Non-activated DC tolerize or delete T cells, whereas activation converts the DC to a stimulatory state that results in T cell activation and memory [12]. We hypothesize that DC under steady state tolerize T cells and that activation of DC from EAE rats promotes T cell activation, suggesting that this may be the reason why IFN-γ-treated DC from EAE rats did not suppress severity of clinical EAE. However, it was proposed that CD8α⁺ DC may mainly activate CD8 T cells, whereas activated CD8α^– DC may activate both CD8 and CD4 T cells [13]. The recent discovery of a CD8α⁺ subset, and suggestions that this subset might mediate peripheral tolerance prompted us to use CD8α⁺ DC from EAE rats for treatment of incipient EAE.

Most studies concerning CD8α⁺ DC have concentrated on mouse-derived DC, but there is clear evidence to suggest that rat-derived DC may also be important in controlling T cell responsiveness [14]. In the present study, we explored whether rat splenic CD8α⁺ from EAE rats have the capacity to create immune protection from clinical EAE.

MATERIALS AND METHODS

Animals and reagents

Female Lewis rats (6–8 weeks of age) were purchased from Zentralinstitut fur Versuchstierzucht (Hanover, Germany). All animals were housed in a pathogen-free condition. Guinea-pig myelin basic protein peptides 68–86 (MBP68-86) and 87–99 (MBP87-99) were synthesized in an automatic Tecan/Syro Synthesizer (Multisyntech, Bochum, Germany). Antirat CD68, OX-62 and PE-conjugated antirat CD4, CD45RA, CD161 and MHC class II as well as FITC-conjugated antirat CD3, CD4 and CD8 were from Serotec (Oxford, UK). Mouse antirat CD4, CD8, FITC-conjugated antirat CD40, PE-conjugated antirat CD28, CD80, CD86 and IL-10 were from PharMingen (San Diego, CA, USA). Modified Griess reagent and N^ω-nitrol-l-arginine methylester (NAME) were from Sigma (St Louis, MO, USA). FITC-conjugated Annexin V was from Serotec.

Induction of EAE

Lewis rats were immunized at the base of the tail with 25 µg MBP 68–86, emulsified (1 : 1) in 100 µl Freund's complete adjuvant (FCA) containing 2 mg Mycobacterium tuberculosis (strain H37RA, Difco, Detroit, MI, USA). Animals were evaluated in a blinded fashion by at least two investigators for clinical signs. Clinical scores of EAE were graded according to the following criteria: 0, asymptomatic; 1, complete loss of tail tone; 2, hind limb paraparesis; 3, complete hind limb paralysis; 4, hind limb paralysis with forelimb involvement; 5, moribund/dead.

Magnetic resonance imaging

The magnetic resonance imaging (MRI) recordings were performed on a 4·7 Tesla magnet with a bore diameter of 40 cm (Biospec Advance 47/40 spectrometer Bruker, Karlsruhe, Germany). The system was equipped with a 12-cm self-shielded gradient coil capable of switching 200 mT/m in 250 µs. For the MRI experiments rats were anaesthetized with isoflurane, induction 0·3% in air, maintenance 0·5%. A temperature-regulated warm air stream maintained the rat body temperature at 37·5°C. A RF coil 35-mm (inner diameter) birdcage resonator was used for the brain. A rapid acquisition with relaxation enhancement (RARE) protocol was used with the following parameters: TR = 3000 ms, TE = 21 ms, RARE factor 32, matrix dimensions = 256 × 256, FOV = 4 cm, eight averages, 1·0 mm slice thickness. For spinal cord, a home-made surface coil with a length of 12 cm was used. The parameters were: TR = 3358·12 ms, TE 63·5 ms, RARE factor 32, matrix dimensions = 256 × 256, FOV = 3 cm, 32 averages, 1·0 mm slice thickness.

DC preparation and injection

Spleens were removed on day 14 post-immunization (p.i.) under aseptic conditions from rats immunized with MBP 68–86 + FCA, and cell suspensions were prepared by grinding spleen through a 40 µm nylon mesh (Falcon, Franklin Lakes, NJ, USA). After erythrocytes were osmotically lysed, cell suspension was filtered to remove debris. DC were obtained as the non-adherent component after overnight culture of plastic adherent cells as described previously, with minor modifications [15], while macrophages represent the cells persistently adherent to flasks [16,17]. Briefly, DC were cultured in serum-free Dulbecco's modification of Eagle's medium (DMEM) (Gibco, Paisley, UK) containing 50 IU penicillin, 50 µg/ml streptomycin, 1% MEM and 10 mm Hepes. After 2 h, non-adherent cells were removed gently by swirling the flasks and aspirating the medium. Flasks were washed five times with serum-free medium to remove remaining non-adherent cells. New medium containing 10% fetal calf serum (FCS) (Gibco) was added to the flasks. After 18 h, re-floating cells were collected as a DC-enriched fraction (total DC), while adherent cells consisted mainly of macrophages. The DC-enriched population contained ∼75–85% DC by staining with OX62 MoAb which recognizes the αE2 subunit of an intergrin expressed specifically on rat DC [18]. Antirat CD3 (T cells), CD45RA (B cells) and CD161 (NK cells) were used to detect the purity of total DC and CD8α⁺ DC. Other cells were routinely 1·6–2·4% T cells, 1·7–2·2% of B cells and ≤1% NK cells in total DC, and 1·9–2·9% T cells, 0·4–1·0% B cells and <1% NK cells in CD8α⁺ DC.

For preparation of the CD8α⁺ subset, the bulk DC were separated using positive selection columns and CD8α MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). After cell fractionation, the CD8α⁺ fraction (CD8α⁺ DC) was made up of approximately 95% CD8⁺ cells.

Total DC and CD8α⁺ DC were injected s.c. at a dose of 1 × 10⁶ DC/rat into four sites along the back on day 5 p.i. with MBP68-86 + FCA or used for in vitro experiments.

Preparation of mononuclear cells

Peripheral blood was obtained on day 12 p.i., and mononuclear cells (MNC) were isolated by centrifugation over Lymphoprep density gradient (Nycomed, Oslo, Norway). After washing three times with medium, cells were adjusted to 2 × 10⁶/ml for experiments.

Preparation of T cells

T cells were separated using positive selection of antirat CD3 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany). After cell fractionation, the purity of T cells was made up of >98% CD3⁺ cells.

Flow cytometry

MNC or DC were stained with MoAbs to CD4, CD8, CD28, MHC class II, CD40, CD80 or CD86 and then subjected to flow cytometry acquisition and analysis with CellQuest software (BD Biosciences, Mountain View, CA, USA). For analysis of apoptosis, cells were suspended in ice-cold binding buffer (Serotec, 1 : 4 dilution with distilled water) at 2–5 × 10⁵ cells/ml. PE-antirat CD4 (5 µl) and FITC-Annexin V (5 µl) were added to the cell suspension (195 µl) for 15 min at room temperature.

Proliferation assay

MNC (2 × 10⁶/ml) were cultured in the presence of irrelevant MBP 87–99 or specific MBP 68–86 (10 µg/ml) or NAME (5 mm). For DC-stimulated T cell proliferation assay, DC were radiated (3000 rad) and co-cultured with T cells (ratio 1 : 20) in the presence of MBP 68–86 (10 µg/ml). After 60 h of incubation at 37°C, cells were labelled for an additional 12 h with 10 µl aliquots containing 1 µCi of [³H]-methylthymidine (Amersham, Little Chalfont, UK). Cells were harvested and thymidine incorporation was measured with a liquid scintillation counter. Cultures were run in triplicate and the results expressed as counts per minute (cpm).

Cytokine enzyme-linked immunosorbent assay (ELISA)

MNC (2 × 10⁶/ml) were cultured in the presence of irrelevant MBP 87–99 and specific MBP 68–86 (10 µg/ml). After 48 h of incubation at 37°C, the supernatants were collected, and IFN-γ and IL-10 were measured by a sandwich ELISA kit (PharMingen) following the manufacturer's instructions. Determinations were performed in duplicate and results expressed as pg/ml.

Intracellular cytokine staining

Cells were first incubated with FITC-antirat CD4 MoAb for 30 min at 4°C, then fixed with 4% paraformaldehyde for 20 min at 4°C and permeabilized with 0·2% saponin in 1% bovine serum albumin–phosphate-buffered saline (BSA–PBS). Cells were incubated subsequently with PE-antirat IL-10 MoAb for 30 min at 4°C, washed and enumerated by flow cytometry.

Nitrite assay

NO was assayed by measuring the end-product nitrite, which was determined by a colorimeter assay based on the Griess reaction. MNC (2 × 10⁶/ml) were cultured in the presence of irrelevant MBP 87–99 and specific MBP 68–86. After 3 days of incubation at 37°C, supernatants (100 µl) were mixed with 100 µl of Griess reagent at room temperature for 10 min. Absorbance was measured at 540 nm in an automated plate reader. Concentration of nitrite was determined by reference to a standard curve of sodium nitrite (Sigma).

Immunohistochemistry

On day 16 p.i., animals were killed and spinal cords were dissected. Cryostat sections were cut at 10 µm and fixed in acetone for 10 min. Endogenous peroxidase activity was blocked by incubation of slides in 0·1% H₂O₂ solution in PBS at 10°C for 10 min. Non-specific binding sites were blocked further with 5% normal rat serum (DACO, Copenhagen, Denmark). The sections were incubated with primary mouse antirat CD4, CD8 and CD68 for 1 h at room temperature, followed by secondary antibodies and corresponding substrate kits. Specificity of the staining was tested by incubating sections without the primary antibodies. Sections were evaluated under a microscope and positive cells were counted by automatic video scanning using a Leica Q500MC.

Statistics

Statistical analysis was performed using one-way anova and Tukey–Kramer multiple comparisons when anova showed significant differences (P < 0·05).

RESULTS

Generation and characteristic of rat splenic CD8α⁺ DC

At present, the most important rat DC marker is the OX62 molecule [19]. By using this antibody, both CD8α⁺ and CD8α^– DC were observed and isolated from rat spleen [19]. We assessed the purity of rat splenic DC by OX62 and CD3 MoAbs. As shown in Fig. 1a, OX62⁺CD3^– DC were about 77%, while CD3⁺ cells were only 1·7%. The CD8α⁺ subpopulation (single stain) represented about 15% of the total rat splenic DC-enriched fraction. OX62⁺CD8α⁺ DC were about 14% and OX62^–CD8α⁺ DC were only 2% before CD8α⁺ DC selection. After CD8α⁺ DC were isolated using positive selection columns and CD8α MicroBeads, CD8α⁺ DC (single stain) were routinely about 95%, while OX62⁺CD8α⁺ DC were made up of about 81% (Fig. 1a).

Fig. 1.

Generation and characteristic of rat splenic CD8α⁺ DC. Splenic MNC were obtained from EAE rats on day 14 p.i. (n = 4 rats). The DC-enriched population was prepared by the adherent-floating process and stained with antirat OX62 and CD3, while CD8α⁺ DC were isolated using CD8α MicroBeads and stained with antirat OX62 and CD8α before and after isolation (a). Phenotype of DC was stained with antirat MHC class II, CD80 and CD86 after isolation (b). Representative pictures are shown.

Expression of various surface molecules was assessed on both total DC and CD8α⁺ DC after purification (Fig. 1b). Both DC expressed similar level of MHC class II and co-stimulatory molecules, such as CD80 and CD86 (P > 0·05, respectively).

T cell stimulatory capacity of DC was tested on both total and CD8α⁺ DC after purification (Fig. 2). DC were co-cultured with T cells in the presence or absence of MBP 68–86. Total DC stimulated antigen-induced T cell proliferation (P < 0·05). A major difference between total DC and CD8α⁺ DC was that CD8α⁺ DC did not stimulate T cell proliferation under antigen stimulation compared with T cells alone (P > 0·05).

Fig. 2.

T cell stimulatory capacity of DC after purification. Both total and CD8α⁺ DC were radiated (3000 rad), and co-cultured with T cells (ratio 1 : 20) in the absence or presence of MBP 68–86 (10 µg/ml). T cell proliferation was measured by [³H]-thymidine. Data are presented as the mean ± s.d. of triplicate samples from four rats.

Injection of CD8α⁺DC mediates immune protection from EAE

Compared with total DC, CD8α⁺ DC from the spleen have a decreased capacity to stimulate T cell function. To evaluate the ability of CD8α⁺ DC to induce peripheral tolerance in EAE, splenic total DC and CD8α⁺ DC were obtained from EAE rats on day 14 p.i. and injected s.c. into rats with incipient EAE, i.e. on day 5 p.i. with MBP 68–86 + FCA. Compared to rats receiving total DC and medium, single injection of CD8α⁺ DC (1 × 10⁶/rat) on day 5 p.i. delayed the onset of clinical signs and reduced the severity of clinical symptoms in all three independent experiments (Fig. 3). In the first experiment (Exp. 1), injection of total DC delayed clinical onset, but there was no difference in subsequent experiments (Exp. 2 and Exp. 3, Fig. 3). After the mixture of three experiments, there was a significant difference between rats injected with CD8α⁺ DC and medium, but no difference between rats injected with total DC and medium (Total, Fig. 3).

Fig. 3.

Clinical suppression induced by CD8α⁺ DC in EAE. Splenic MNC were obtained from EAE rats on day 14 p.i. CD8α⁺ DC were prepared with CD8α MicroBeads by positive selection. Five days p.i. with MBP 68–86 + FCA, such rats were injected s.c. with total DC (n = 12 for three experiments) or with CD8α⁺ DC (n = 8 for three experiments) or with medium (n = 14 for three experiments). Mean clinical score is presented from three independent experiments with identical results and as the mean ±SD of total of three experiments (*P < 0.05).

Injection of CD8α⁺ DC reduces MRI lesions in brain and spinal cord

In preliminary experiments, we did not observe any difference on damage detectable by MRI in brain and spinal cord between control EAE rats and rats injected with total DC [3]. We also observed that the clinical score and inflammatory cell infiltrates in the spinal cord reached a peak on day 14 p.i. in MBP 68–86-induced Lewis rat EAE, while the damage detectable by MRI became more severe on day 18 p.i. and could last for 1 week, regardless of the recovery phase of EAE [3]. Based on this information, we chose here to perform MRI examination on day 18 p.i., and to explore whether injection of CD8α⁺ DC can inhibit damage in the brain and spinal cord detectable by MRI, compared to control EAE. Figure 4 shows that injection of CD8α⁺ DC clearly reduced the extent of the lesions of both lateral ventricles and the third ventricle, while MRI of the lumbar spinal cord revealed less T2 lesions compared to control EAE rats.

Fig. 4.

MRI axial slices at brain and lumbar spinal cord level were performed ín Lewis rat EAE on day 18 p.i. (3–4 rats/Exp. 2). Note enlarged lateral ventricles and third ventricle in medium-injected control EAE rats, and normal size of ventricles in the CD8α⁺ DC-injected rats. Note also scattered T2 lesions in axial slices at the lumbar spinal cord level in medium-injected control EAE rats, and normal T2 appearance in CD8α⁺ DC-injected rats. Representative pictures are shown.

Phenotypic analysis of MNC

To examine the phenotypic diversity of MNC, blood was obtained on day 12 p.i. Immediately following separation, MNC were stained for expression of CD4, CD8, CD28, CD40, CD80, CD86 and MHC class II, and analysed by flow cytometry. CD8⁺CD28^– cells were considered as T suppressor cells. The injection of CD8α⁺ DC did not induce the generation of CD8⁺CD28^– T suppressor cells (29%, mean = 28% ± 9, P ≥ 0·05) compared to rats injected with total DC (26%, mean = 25% ± 10) and rats injected with medium (30%, mean = 28 ± 6) (Fig. 5). There was no difference in the expression of MHC class II and co-stimulatory molecules (CD40, CD80 and CD86) expressed by blood MNC from the three groups of rats (Fig. 5). Figure 5 also showed that injection of CD8α⁺ DC was unable to induce apoptosis of CD4⁺ T cells (CD4⁺annexin⁺ cells).

Fig. 5.

Phenotypic analysis of MNC. To examine phenotypic diversity of MNC, blood was obtained on day 12 p.i. (3–4 rats/Exp. 2), and MNC were prepared by centrifugation over Lymphoprep density gradient. MNC were stained for CD4, CD8, CD28, CD40, CD80, CD86 and MHC class II, and analysed by flow cytometry. Representative patterns are shown.

Injection of CD8α⁺ DC leads to increase of IFN-γ secretion

To assess whether injection of CD8α⁺ DC can alter the cytokine pattern, we examined levels of IFN-γ, IL-10 and TNF-α secretion from blood MNC by ELISA. Rats injected with CD8α⁺ DC showed an increase of spontaneous and MBP68–86-induced IFN-γ secretion in supernatants of blood MNC (mean 60 pg/ml for spontaneous secretion and 344 pg/ml for MBP68–86-induced secretion) compared to rats injected with total DC (mean 4 pg/ml for spontaneous secretion and 156 pg/ml for MBP68–86-induced secretion) or rats injected with medium (mean 6 pg/ml for spontaneous secretion and 178 pg/ml for MBP68–86-induced secretion) (Fig. 6). There was no difference between spontaneous and MBP68–86-induced IL-10 or TNF-α secretion among the three groups of rats (Fig. 6).

Fig. 6.

MBP87–99- and MBP68–86-induced cytokine levels. Blood was obtained on day 12 p.i., and MNC were prepared by centrifugation over Lymphoprep density gradient. IFN-γ, IL-10 and TNF-α were measured by ELISA kits. Data are presented as the mean ± s.d. of duplicate samples from three to four rats/Exp. 2 (*P < 0·05).

Injection of CD8α⁺ DC inhibits T cell proliferation and enhances NO production

The injection of CD8α⁺ DC inhibited spontaneous and MBP68–86-induced T cell proliferation (mean 555 cpm for spontaneous proliferation and 546 cpm for MBP68–86-induced proliferation) compared to injection of total DC (mean 785 cpm for spontaneous proliferation and 1910 cpm for MBP68–86-induced proliferation) or injection of medium (813 cpm for spontaneous proliferation and 1710 cpm for MBP68–86-induced proliferation). There was no difference between rats injected with total DC and medium (Fig. 7a).

Fig. 7.

MBP87–99- and MBP68–86-induced proliferative response and NO production. Blood was obtained on day 12 p.i., and MNC were prepared by centrifugation over Lymphoprep density gradient. T cell proliferation was measured by [³H]-thymidine incorporation (a). NO production was measured by the Griess reagent (b). Blood MNC were obtained from control EAE rats on day 12 p.i. and cultured in the presence of MBP 68–86 (10 µg/ml). The relationship between NO production and proliferative response was assessed by addition of NAME (5 mm) (c). Data are presented as the mean ± s.d. of duplicate samples from three to four rats/Exp. 2 (*P < 0·05; **P < 0·01).

In view of the two facts that IFN-γ is an inducer of NO production and that NO can inhibit T cell proliferation, we then measured the levels of NO production. Injection of CD8α⁺ DC increased NO production (mean 17·3 µm for spontaneous production and 86 µm for MBP68–86-induced production) compared to rats injected with total DC (mean 7·4 µm for spontaneous production and 57 µm for MBP68–86-induced production) and rats injected with medium (mean 6·6 µm for spontaneous production and 47 µm for MBP68–86-induced production) (Fig. 7b).

When cells were cultured with NAME, NO production was partly inhibited, accompanied by the increase in T cell proliferation (mean cpm = 2656) compared with those in the absence of NAME (mean cpm = 1821) (Fig. 7c). These data implicate that reduced T cell proliferation in rats injected with CD8α⁺ DC may be related to increased NO production.

Injection of CD8α⁺ DC leads to increase of CD4⁺ T cells expressing IL-10

A population of T regulatory cells (CD4⁺ T cells expressing IL-10) can be visualized directly by immunofluorescence flow cytometry according to intracellular expression of IL-10. Consistent with previous results [43], the injection of CD8α⁺ DC resulted in increase of CD4⁺ T cells expressing IL-10 (10·7%) compared to rats receiving total DC (5·1%) and medium (3·9%) (Fig. 8). Taken collectively, these results raised the possibility that CD8α⁺ DC have the ability to induce the differentiation of CD4⁺IL-10 expressing cells with potential suppressive activity [20].

Fig. 8.

Frequency of CD4⁺ T cell expressing IL-10. Blood was obtained on day 12 p.i. (3–4 rats/Exp. 2), and MNC were prepared by centrifugation over Lymphoprep density gradient. Immediately following separation, CD4⁺ T cells expressing IL-10 were stained with FITC-CD4 and PE-IL-10 antibodies, and analysed by flow cytometry. Representative patterns are shown.

Injection of CD8α⁺ DC inhibits infiltrates of inflammatory cells within the CNS

CNS inflammation is a characteristic of disease progression in EAE. The degree of inflammatory cell infiltration correlates with disease severity. To examine the influence of CD8α⁺ DC versus total and medium on degree of infiltration within the CNS in Lewis rat EAE, we evaluated infiltrating CD4⁺ T cells, CD8⁺ T cells and CD68⁺ macrophages in spinal cord by immunohistochemistry staining. Low levels of infiltrating CD68⁺ macrophages were detected in rats injected with CD8α⁺ DC compared to rats receiving medium (P < 0·05) (Fig. 9). There was no difference on infiltrating CD4⁺ T cells or CD8⁺ T cells within the CNS among the three groups of rats (Fig. 9).

Fig. 9.

Infiltrates of CD4⁺ T cells, CD8⁺ T cells and CD68⁺ macrophages in spinal cords. Rats were sacrificed on day 16 p.i., and lumbar spinal cords were dissected (3–4 rats/Exp 2). Infiltrating CD4⁺ T cells, CD8⁺ T cells and CD68⁺ macrophages were detected by immunohistochemical staining (*P < 0·05).

DISCUSSION

DC constitute a heterogeneous population of professional antigen-presenting cells. Several subsets of DC have been defined in humans and mice using DC-specific or non-specific markers. For instance, in humans, DC1 and DC2 have been described using the cell surface marker CD11c. Based on CD8α expression, there are at least two subsets among mouse splenic DC that interact with T cells, resulting in different outcomes. However, the expression of CD8α on rat splenic DC is contradictory. A study by Trinite et al. [21] observed that DC freshly prepared from spleen did not express CD8α, while DC from lymph nodes exhibited a low level of CD8α expression. In contrast, DC isolated by the adhere-floating procedure showed co-expressing CD8α and OX62 molecules on rat splenic DC [19]. One explanation for the contradiction is that freshly prepared DC may loss a population of OX62^–CD8α⁺ DC. Another possibility is that certain stimulation may trigger CD expression. CpG or viral stimulus resulted in the acquisition of CD8 expression on the CD8^– populations [22]. It was also reported that expression of CD8 was higher on cells derived from rat lymphoid DC than myeloid DC [23]. The finding that rat CD8α⁺ splenic DC could represent a rat lymphoid tissue-related DC is supported by the loss of CD8α expression on rat myeloid bone marrow-derived DC [19,23]. In the present study, we observed a DC subset co-expressing CD8α and OX62 on rat splenic DC, and we also show that the CD8α⁺ DC subset in the rat spleen closely resembles mouse splenic CD8α⁺ DC, including lower stimulatory capacity.

Functionally, splenic CD8α⁺ DC are less potent inducers of T cell proliferation compared to CD8α^– DC, despite the expression of similar levels of co-stimulatory molecules [6,24]. Recent findings have challenged the concept that maturation of DC is related to immunity or tolerance. Mature DC also induced regulatory CD4⁺ T cells in vivo [25] and suppressed alloreactive Th1 response, resulting in a delayed skin graft rejection [26], and polarized Th2 response, suppressing EAE [27]. This inefficiency to stimulate T cells has been proposed to result from several possibilities: (1) CD8α⁺ DC induce the Fas-mediated death of activated T cells [10]; (2) CD8α⁺ DC expressing IDO mediate apoptosis of T cells through tryptophan catabolism [28]; and (3) mucosal CD8α⁺ DC are able to differentiate naive T cells into T regulatory 1-like cells with regulatory properties [20]. These findings therefore suggest that a particular CD8α⁺ DC subset could play a role in peripheral tolerance in vivo. Our results confirm that rat splenic CD8α⁺ DC suppress the development of clinical EAE in Lewis rats compared to total DC. The hypothesis that CD8α⁺ DC have tolerizing capacity was challenged by subsequent reports that CD8α⁺ DC induce a Th1 response, whereas CD8α^– DC lead to Th2 differentiation [5]. Consistent with these results generated by mouse splenic CD8α⁺ DC, rat splenic CD8α⁺ DC suppressed the severity of clinical EAE, accompanied by an increase of IFN-γ production in vivo.

In view of the established Th1-mediated autoimmune pathogenesis of EAE, the therapeutic effect of IFN-γ is unexpected. However, the traditional view has been challenged recently by a number of studies describing unexpected disease-ameliorating effects by IFN-γ in EAE [29–32], as well as in lethal autoimmune myocarditis [33] and collagen-induced arthritis [34]. IFN-γ can result in apoptosis in multiple cells by different pathways, including by the promotion of caspase-8-dependent apoptosis [35], up-regulation of Fas [36] and down-regulation of Bcl-2 and Bcl-x(L) [37]. IFN-γ is also a powerful stimulus of NO production that inhibits T cell proliferation [38]. DC induce cell apoptosis which is mediated partially by NO [39]. In our studies, NO production was enhanced in CD8α⁺ DC-injected rats, indicating that the IFN-γ–NO pathway may be related to inhibition of T cell proliferation. The defective ability to sustain T cell proliferation was observed selectively by CD8α⁺ DC cultured with Th1 cells. During co-culture, the Th1 cells were induced to express considerable amounts of CTLA-4 [28]. We do not exclude another possibility, namely that IFN-γ can up-regulate the expression of IDO by DC, which in turn inhibits T cell function through tryptophan catabolism [40]. Peripheral CD8α⁺ DC have been shown to actively suppress immune responses in vivo via IDO, suggesting that this subset of CD8α⁺ DC may utilize multiple mechanisms to limit T cell responsiveness [41].

T regulatory cells represent another mechanism by which tolerogenic DC mediate peripheral tolerance. It is now clear that CD4⁺ regulatory T cells encompass more than one cell type, including CD4⁺CD25⁺ T cells, CD4⁺CD25^– T cells and CD4⁺ T cells producing IL-10. Both steady state and mature antigen-processing DC induce proliferation of adoptively transferred CD4⁺ CD25⁺ T cells in vivo and expand CD4⁺ CD25⁺ T cells in the presence or absence of specific antigen in vitro [42]. CD11c_lowCD45RB_high DC specifically induce tolerance through the differentiation of CD4⁺ T cells producing IL-10 in vitro and in vivo [43]. T regulatory cells suppress immune responses via cell-to-cell interactions and/or the production of IL-10. Type I T regulatory cells (CD4⁺ T cells expressing IL-10) are defined by their ability to produce high levels of IL-10. Our data show that injection of CD8α⁺ induced the expansion of CD4⁺ T cells expressing IL-10, indicating that immune protection mediated by CD8α⁺ DC may not depend on CD4⁺ T cells expressing IL-10. A possible explanation for the discrepancy (augmented CD4⁺ T cells expressing IL-10 and unchanged levels of IL-10 production) is that injection of CD8α⁺ DC may just trigger expansion of CD4⁺ T cells expressing IL-10, but do not promote these T cells to produce IL-10. In spinal cords, the frequency of CD4⁺IL-10⁺ cells was low; no significant difference was observed among three groups (data not shown).

CD8α⁺ DC appear to be responsible for induction of both cross-tolerance and cross-priming [44]. What are the mechanisms that select between these opposing outcomes? One possibility is that CD8α⁺ DC may be subdivided further into immunogenic and tolerogenic subsets. It is more probable that tolerance induction is the default outcome, and that additional stimuli associated with foreign antigens are responsible for their immunogenicity [45]. Other studies support the notion that the CD8α⁺ DC can be converted to an immunogenic form by CD40 signalling [46], indicating that the immunogenicity of these DC can be influenced by environmental signals. We observed that CD8α⁺ DC, pulsed with MBP 68–86 in vitro, did not affect the development of clinical EAE (data not shown). This dual role for a DC subset should be noted in designing immunotherapeutic strategy for autoimmune diseases.

In conclusion, these results provide a novel possibility that autologous DC, after CD8α^– DC are depleted, may be used to mediate immune protection for treatment of multiple sclerosis or other diseases with an autoimmune background.