Wild type (WT) C57BL/6J (B6-CD45.2), B6.SJL-Ptprc^a Pepc^b/BoyJ (B6-CD45.1), and Klrk1^−/− (B6.Cg-Klrk1^tm1Dhr/J) mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). All mice were treated in strict adherence with approved protocols (N13043NAs and N17031NAs) from the CRCHUM Institutional Committee for the Protection of Animals and the Canadian Council on Animal Care.

Active EAE was induced as previously published (20). Briefly, 6–8 week old female mice were immunized subcutaneously with 200 μg of myelin oligodendrocyte glycoprotein 35–55 (MOG₃₅₋₅₅) peptide emulsified in Complete Freund's adjuvant supplemented with 400 μg of Mycobacterium tuberculosis (CFA-MOG₃₅₋₅₅). Two days later, mice were intraperitoneally injected with 400 ng of pertussis toxin (PTX).

For passive EAE, 6–8 week old female donor WT mice were similarly immunized with CFA-MOG₃₅₋₅₅ and injected intraperitoneally with 400 ng of PTX. Eight days later, donor mice were sacrificed; lymph nodes and spleens were harvested and processed as described below. Cells were put in culture at 7 million/ml in complete RPMI [10% (v/v) of fetal bovine serum, 50 μM of β-mercaptoethanol, 1 mM of sodium pyruvate, 0.01 M of HEPES, 1X non-essential amino acids solution, 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin] in the presence of MOG₃₅₋₅₅ (20 ug/ml), recombinant mouse IL-12 (10 ng/ml, R&D Systems distributed by Cedarlane Laboratories Oakville, ON, Canada), recombinant human IL-2 (100 U/ml, Roche, Nutley, NJ) and mouse recombinant IL-15 (1 ng/ml, R&D Systems) pre-complexed (incubation 30 min at 37°C) with recombinant mouse IL-15Rα (4.67 ng/ml, R&D Systems) as published by others (21). After 72 h of culture, cells were washed, resuspended in Hank's Balanced Salt solution, filtered on 70 μm cell strainer, counted and injected intraperitoneally into naïve Klrk1^−/− and wild type (Klrk1^+/+) female mice.

Passive and active EAE animals were scored according to the following scale: 0 = normal, 1 = limp tail, 2 = slow righting-reflex, 3 = paralysis of one hind limb, 3.5 = paralysis of one hind limb and weakness in second hind limb, 4 = paralysis of both hind limbs, 4.5 = paralysis of both hind limbs and weakness in forelimbs, 5 = moribund. Deeply anesthetized mice were perfused with 50 ml of saline 0.9% (w/v) prior to collect organs for RNA extraction, protein extraction, or organ processing for flow cytometry analysis.

Collected organs were put directly into TRIzol® Reagent (Life technologies Thermo Fisher Scientic, Burlington, ON, Canada). Total RNA was extracted from spleen, liver, spinal cord, brain stem-cerebellum and forebrain according to manufacturer's instructions. RNA samples were transcribed into cDNA using Quantitect Reverse transcription kit (Qiagen, Mississauga, ON, Canada) as previously published (22). Gene expression levels were determined by quantitative real-time PCR using primers and TaqMan FAM-labeled probes from Applied Biosystems-ThermoFisher Scientific and the QuantStudio 7 Flex Real-Time PCR System. Amplification of mouse HPRT1 (Mm01545399_m1) was used as an endogenous control to assess the expression of murine NKG2D ligands: MULT1 (Mm01180648_m1), panRAE (Mm00558293_g1), H60b (Mm04243254_m1), and H60c (Mm04243526_m1).

Spinal cord, brain stem-cerebellum and forebrain were, respectively, homogenized in 500 ul of lysis buffer [50 mM Tris pH 7.5, 1 mM EDTA, 150 mM NaCl, 1% (w/v) NP40, 1% (w/v) SDS and cocktail of protease inhibitor containing 5 μg/ml of: Leupeptin (BioShop, Burlington, ON, Canada), Pepstatin A (BioShop) and Chymostatin (Sigma-Aldrich, Oakville, ON, Canada)]. Lysates were kept 10 min at room temperature before storing at −80°C.

Deeply anesthetized mice were placed on the stereotaxic instrument with 135° between head and body. Sagittal incision of the skin was made aligned with the ears, then the subcutaneous tissues and muscles were separated by blunt dissection and were hold by microretractor. The capillary tube was introduced into the cisterna magna though the dura mater, laterally to the arteria dorsalis spinalis. CSF was harvested into capillary tube, collected into Eppendorf tube and immediately frozen on dry ice. Samples were stored at −80°C.

Proteins isolated from organs were electrophoresed under reducing conditions and transferred onto PVDF membrane as previously published (23). CSF samples were run on a stain free gel (Bio-Rad Laboratories, Mississauga, ON, Canada) containing trihalo compounds reacting with tryptophan residues and used to revel total proteins on gel before transfer. MULT1 (rat monoclonal antibody clone 1D6 provided by Dr. S. Jonjic, University of Rijeka, Rijeka, Croatia) and Rae-1 (polyclonal goat anti-Rae-1 pan specific R & D Systems), were quantified by chemiluminescence relative to total protein as published (24), actin (monoclonal antibody clone C4, MP Biomedicals, Fisher Scientific) or albumin (rabbit anti-albumin, Novus Biologicals, distributed by Cedarlane Laboratories) using the Chemidoc MP imaging system (Bio-Rad Laboratories).

Brain and spinal cord were processed from individual mice as previously published (20). Briefly, minced CNS was digested 15 min with collagenase D and DNase I at 37°C prior to being mashed. Myelin was removed using Percoll. Harvested spleens and lymph nodes were put in complete RPMI and then mashed through cell strainer (70 μm) to obtain single cell suspension. Red blood cells from spleens were lysed using 0.83% (w/v) of ammonium chloride for 4 min at room temperature.

Splenocytes from Klrk1^+/+ or Klrk1^−/− mice were put in culture in complete RPMI in 48 well plates pre-coated with anti-CD3 antibody (2 ug/ml in 100 mM NaHCO3 overnight at 4°C, BD Biosciences, Mississauga, ON, Canada), anti-CD28 antibody (1 ug/ml BD Biosciences), recombinant human IL-2 (100 U/ml), and the complex recombinant mouse IL-15/IL-15Rα (described above). On day 5, recombinant murine MULT1 protein (10 μg/ml, R&D Systems) was added or not to splenocytes. On day 7, cells were activated for the detection of intracellular mediators and analyzed by flow cytometry as described below.

Flow cytometry staining was performed as previously published (22, 25, 26). Briefly, cells were incubated 15 min in blocking solution containing normal mouse immunoglobulins and rat anti-mouse CD16/CD32 prior to be incubated in the presence of fluorochrome-conjugated antibodies. Antibodies targeted CD45.1 (Biolegend, San Diego, CA, USA, clone A20), CD45.2 (BioLegend, distributed by Cedarlane Laboratories, clone 104), CD11b (BD Biosciences, clone M1/70), CD3 (BD Biosciences, clone 145-2C11), CD4 (BD Biosciences, clone RM4-5), CD8 (BD Biosciences, clone 53-6.7), and NKG2D (eBioscience or R&D systems, respectively, clone CX5 or 191004). LIVE/DEAD Fixable Aqua Dead Cell Stain was used to exclude dead cells. To trace proliferation of MOG-specific T cells, leukocytes from MOG-immunized donor mice were labeled with CFSE as previously done (22, 25) prior to being activated in vitro for 72 h. For cytokine detection, cells were stimulated 5 h with phorbol 12-myristate 13-acetate (20 ng/ml, Sigma-Aldrich) and ionomycin (1 ug/ml, Sigma-Aldrich) in the presence of brefeldin A (5 ug/ml, Sigma-Aldrich) and monensin sodium (1 μM Sigma-Adrich). Intracellular staining was accomplished as previously published (25). Antibodies targeted interferon-γ (IFNγ, BD Biosciences clone MP6-XT22), granulocyte-macrophage colony-stimulating factor (GM-CSF, BD Biosciences, clone MP1-22E9), interleukin-17 (IL-17, BD Biosciences, cloneTC11-18H10) and granzyme B (eBioscience ThermoFisher Scientific, clone 16G6). Appropriate isotype controls were used in all steps. Staining specificity was confirmed using fluorescence minus one (FMO: all antibodies minus one). The median fluorescence intensity (MFI) was calculated by subtracting the fluorescence of the isotype from that of the stain. Cell numbers were quantified using either cell counting prior to cytometry staining or beads added to samples prior to sample acquisition as previously described (20).

Deeply anesthetized mice were perfused with 30 ml of saline 0.9% (w/v) and then with 50 ml of paraformaldhehyde 4% (w/v in PBS). Spinal cord was collected and soaked into 4% paraformaldehyde for 1 day prior to being transferred into sucrose 30% (w/v) for 2 days and then put into OCT for freezing at −80°C. Nine micron sections were stained for FluoroMyelin™ Green fluorescent myelin stain (Thermofisher Scientific) and 2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride, 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI) (Sigma-Aldrich) for nucleus detection according to manufacturers' instructions. Slides were observed using a SP5 Leica confocal microscope and confocal images acquired sequentially in different channels using LASAF software and overlay using Adobe Photoshop software.

Data analysis was performed using Prism 7.0 software (GraphPad, La Jolla, CA, USA). Statistical tests used are indicated in figure legends. Values were considered statistically significant when probability (P) values were equal or below 0.05 (^*), 0.01 (^**), or 0.001 (^***).

To assess the impact of NKG2D on the development of EAE, we applied our standard protocol on mice lacking NKG2D (Klrk1^−/−) and wild type (Klrk1^+/+) counterparts. We observed a modest and transient (day 11 to 14) enhanced peak of disease in Klrk1^−/− compared to wild type mice when large groups of animals were analyzed (n = 22–25) (Figure 1A). We observed similar proportions of infiltrating CD4 (67–72% of total CD3 T lymphocytes) and CD8 (14–17% of total CD3 T lymphocytes) T cells in the CNS of both groups (Figures 1B,C). Within these CNS-infiltrated T lymphocytes, only a very small proportion of CD4 T lymphocytes expressed NKG2D (less than 8%) whereas 28% of CD8 T lymphocytes carried detectable levels of this receptor in Klrk1^+/+ mice (Figure 1D). As expected, a negligible proportion of T lymphocytes expressed NKG2D in the Klrk1^−/− mice (Figures 1B,D). The total number of T cells, CD4 and CD8 T cells that infiltrated the CNS during EAE were similar in both genotypes (Figure 1E). Similar to what others published (19), NKG2D had a minimal impact on the neurobehavioral score of EAE mice. Our results show that during EAE, amongst the CNS-infiltrated T lymphocytes, a greater proportion of CD8 than CD4 T lymphocytes carried NKG2D but as CD4 T cells were more abundant, NKG2D-expressing CD4 and CD8 T lymphocytes were present in similar number per CNS in wild type mice (Figure 1E).

As each NKG2DL can be induced in response to different triggers and exhibits distinct biological functions, we assessed the expression of the specific ligands (H60 (b-c), Rae1 (α-ε) and MULT1) known to be transcribed in C57BL/6 mice (9, 27–29), during the development of EAE. To investigate whether the presence of NKG2D-expressing immune effectors affects the ligands' expression, we quantified the relative mRNA levels of NKG2DL in Klrk1^−/− and wild type (Klrk1^+/+) mice at the different EAE stages: pre-symptomatic, onset and peak stages. We analyzed three CNS areas: spinal cord, brain stem/cerebellum and forebrain and observed that the mRNA levels of H60b did not significantly vary in the wild type mice (Figure 2A). We detected a slight but significant decrease in the spinal cord of Klrk1^−/− mice compared to controls at disease onset and a slight increase in the brainstem-cerebellum (Figure 2A). H60c mRNA was not detected in all tested organs (CNS, spleen, etc.) except the eye (data not shown). Using a pan-Rae-1 qRT-PCR assay to detect all forms of Rae-1, we observed a small but significant increase at the peak of disease in wild type mice in the spinal cord compared to controls as previously reported by others (30). However, Rae-1 mRNA levels were decreased in the spinal cord of Klrk1^−/− mice at the presymptomatic and onset stages as well as in the forebrain at the presymptomatic stage (Figure 2A). Notably, the levels in Klrk1^−/− mice were significantly lower than in Klrk1^+/+ at the same disease stage for the presymptomatic and onset in the spinal cord and the brainstem-cerebellum (Figure 2A). We also assessed the protein expression of Rae-1 in these CNS areas during the development of EAE. The relative amounts of Rae-1 did not vary between control mice (CFA without MOG) and EAE animals in all three CNS areas tested (spinal cord, brainstem/cerebellum, forebrain) (Figures 2B,C).

In contrast, MULT1 mRNA levels were significantly enhanced in the spinal cord of both Klrk1^+/+ and Klrk1^−/− groups at the peak of EAE compared to controls, presymptomatic and onset time points (Figure 3A). Upregulated MULT1 mRNA levels during EAE were previously reported by others (30). Moreover, MULT1 mRNA levels were greater in Klrk1^−/− mice compared to wild type counterparts (Figure 3A) without reaching statistical significance. These observations suggest that in the presence of NKG2D-expressing immune effectors, MULT1-expressing cells may down-regulate such expression or may be destroyed. In the brainstem-cerebellum and forebrain, we also detected significantly augmented MULT1 mRNA levels at the peak of EAE compared with other disease stages in both Klrk1^+/+ or Klrk1^−/− mice (Figure 3A). We observed a small increase of MULT1 mRNA levels in the spleen of Klrk1^−/− and Klrk1^+/+ mice at disease peak and no variation in the liver (Figure 3B). These results demonstrate that during active EAE, MULT1 expression is significantly upregulated (over 900% increase in spinal cord) in the CNS.

To determine whether MULT1 expression is also increased at the protein level, we performed western blot analysis on the same CNS areas. We observed two bands representing the full length (50–55 kDa) and the putative cleaved extracellular domain (25 kDa) of MULT1 at the peak of disease, which was absent in controls mice (Figure 3C). The quantification of MULT1 relative to actin showed that the spinal cord contained greater levels of both MULT1 forms (55 and 25 kDa) compared to brainstem/cerebellum and forebrain areas in Klrk1^+/+ as well as Klrk1^−/− mice at disease peak (Figure 3D). The full and cleaved forms of MULT1 were detected in the spinal cord at the onset of EAE only in a subset of mice (4 out of 12 mice). Notably, greater amounts of the 25 kDa band relative to the 55 kDa band at disease peak were observed in both Klrk1^+/+ and Klrk1^−/− mice (Figures 3E,F). Our results strongly suggest that MULT1 is the NKG2DL with the greatest enhanced mRNA and protein expression levels at the peak of EAE in the CNS, especially in the spinal cord, regardless of the presence of NKG2D-expressing immune cells.

To establish whether soluble MULT1 diffuses in the CNS, we performed western blots on CSF collected from control or EAE mice at different stages (pre-symptomatic, onset, peak) in Klrk1^+/+ and Klrk1^−/− mice (Figure 4A). We observed that both forms (25 and 55 kDa) of MULT1 were detectable in the CSF and those levels were greater at disease peak compared to other EAE stages (Figure 4B). Notably, the 25kDa form was 7-fold more abundant than the 55 kDa form in the CSF (Figure 4C) compared to 4.5-fold increased expression in the spinal cord protein homogenates at the same disease stage (Figure 3F).

We next sought to determine whether soluble MULT1 could impact activated T lymphocytes, which are abundantly present in the CNS of EAE mice. Splenocytes from naïve Klrk1^+/+ and Klrk1^−/− mice were activated in vitro with anti-CD3+anti-CD28+IL-2+IL-15 to enhance NKG2D expression on CD8 T lymphocytes to similar levels observed on CNS-infiltrating CD8 T cells. After 5 days, recombinant mouse MULT1 or PBS was added to these activated cultures and 2 days later, cells were evaluated for their production of mediators by flow cytometry. Recombinant MULT1 was obtained from the same commercial source than what has been used by the Raulet's group to perform in vitro experiments with activated murine immune cells (10). Our in vitro culture conditions gave rise to a similar proportion (37%) of activated Klrk1^+/+ CD8 T lymphocytes expressing NKG2D (Figures 5A,B) as we observed on CNS-infiltrating CD8 T lymphocytes in EAE mice (Figure 1C, 28%). As expected, NKG2D expression was negligible on Klrk1^−/− cells. Notably, the addition of soluble MULT1 did not alter the percentage of cells expressing NKG2D; but slightly reduced NKG2D expression levels per cell in Klrk1^+/+ CD8 T cells [Figure 5C; MFI for NKG2D on Klrk1^+/+ CD8 T cells exposed to PBS (1279) vs. MULT1 (1068)]. The addition of MULT1 increased the percentage of granzyme B-expressing CD8 T lymphocytes (Figure 5B; % granzyme-expressing Klrk1^+/+ CD8 T cells exposed to PBS (44%) vs. MULT1 (67%) p = 0.09); these cells expressed greater levels of this lytic enzyme compared to cells not treated with MULT1 (Figure 5C; granzyme B-MFI in Klrk1^+/+ CD8 T cells exposed to PBS (3094) vs. MULT1 (4854) p = 0.1). MULT1 did not alter the percentage of IFNγ producing Klrk1^+/+ CD8 T lymphocytes (80–86%) but significantly increased their cellular expression level (Figures 5A,C; IFNγ-MFI on Klrk1^+/+ CD8 T cells exposed to PBS (3095) vs. MULT1 (4901) ^*p = 0.0133). The enhancing impact of MULT1 on CD8 T cell effector functions was dependent on the presence of NKG2D as this soluble ligand did not alter the production of granzyme B or IFNγ by Klrk1^−/− CD8 T lymphocytes as similar percentage of cells and MFI were observed for these cells in the PBS and MULT1 conditions (Figure 5). Our results suggest a novel activating role of soluble MULT1 on CD8 T lymphocytes.

Active EAE is strongly mediated by CD4 T lymphocytes but only a small proportion of these lymphocytes can express NKG2D upon activation. In contrast, a great proportion of activated murine CD8 T cells acquire NKG2D. To investigate the contribution of this receptor, we optimized a passive EAE model in which the transferred T cells include a similar proportion of CD8 and CD4 T cells. Lymphocytes from MOG-immunized donor mice were activated in vitro in the presence of IL-2, IL-12, and IL-15 prior to their transfer into naïve recipients. The addition of this cytokine cocktail led to the expansion of similar proportions of CD4 and CD8 T cells (data not shown). Activated cells were transferred into naive Klrk1^+/+ and Klrk1^−/− mice and clinical score recorded for 2 months (Figure 6A). Disease onset occurred similarly around day 10 post-transfer in both mouse groups (Figure 6A). However, Klrk1^+/+ recipients achieved a significantly more elevated clinical score than the Klrk1^−/− recipients (Figure 6A). The dampened disease in Klrk1^−/− recipients was maintained for the whole observation window. As these mice were injected in parallel with the same activated lymphocytes, the differences we observed could not be due to the transferred cells. Our results suggest that in passive EAE, endogenous immune cells contribute to disease pathobiology and that such impact depends on the endogenous expression of NKG2D.

We evaluated whether the injected cells contain MOG-specific T cells. Lymph node leukocytes from MOG-immunized donor mice were labeled with CFSE and activated in vitro following our usual protocol in the presence of MOG. Control cells were cultured in the same conditions but in the absence of the myelin peptide. We observed that our activation protocol efficiently induced proliferation as assessed by CFSE dilution (Figure 6B). Moreover, the proportion of CD4 and CD8 T lymphocytes that divided during the in vitro expansion and could produce key immune effectors (GM-CSF, IFNγ, IL-17 for CD4; GM-CSF, IFNγ, and Granzyme B for CD8) was greater in the presence of MOG (Figure 6B). We suggest that in vitro MOG addition had a more modest impact on CD4 than CD8 T lymphocytes due to the greater efficiency of antigen presenting cells activated upon immunization at activating CD4 than CD8 T lymphocytes. These results support the idea that at least a subset of transferred T lymphocytes was MOG-specific. In addition, the passive transfer of MOG-activated lymphocytes caused demyelination in recipient mice. Indeed, fluoromyelin staining revealed zones of demyelination (white arrows Figure 6C) in both Klrk1^+/+ and Klrk1^−/− mice. Finally, we also detected soluble MULT1 in the CSF of passive EAE mice supporting the notion that this NKG2DL is also elevated in this form of disease (Figure 6D). Notably, the relative amount of the 25kDa form of MULT1 was significantly more elevated in CSF from Klrk1^+/+ compared to Klrk1^−/− (Figure 6D).

To investigate the contribution of endogenous vs. transferred T cells, we used C57BL/6 mice expressing the pan leukocyte marker CD45.1 as donor mice for the transfer into the CD45.2-expressing Klrk1^+/+ and Klrk1^−/− recipients. Mice were sacrificed 25 days post-transfer and CNS cells were analyzed by flow cytometry. We could easily discriminate donor (CD45.1⁺) from recipient immune cells (CD45.2⁺) (Figure 7A). CD45^int CD11b⁺ microglia all expressed the endogenous CD45.2 allele (data no shown). We compared CD45.2^hi+ immune cells to the CD45.1⁺ cells in Klrk1^+/+ and Klrk1^−/− recipients. Most transferred cells (CD45.1⁺) detected in the CNS were either CD4 or CD8 T cells (Figure 7B). Despite the injection of similar numbers of CD45.1⁺ CD4 and CD8 T cells, donor CD8 T cells represented a very small proportion of CNS-infiltrated T cells, especially in Klrk1^−/− mice, while donor CD4 T cells were well represented. In contrast, amongst the endogenous immune cells present within the CNS, a greater proportion were CD8 T cells (41 % in Klrk1^+/+ and 44% in Klrk1^−/−) compared to CD4 T cells (30.7% and 28.8%) (Figure 7B). The spleens of all animals contained very low proportion of CD45.1⁺ (donor) T cells (around 1%) suggesting that the transferred cells either died or migrated to other sites such as the CNS at the time point tested (over 20 days after transfer). Overall, less transferred T cells were detected in the CNS of Klrk1^−/− recipients than wild type counterparts correlating with the reduced disease severity in Klrk1^−/− recipients (Figure 7B).

To characterize the properties of CNS-infiltrated T cells, we assessed the production of key immune effector molecules: IFNγ, GM-CSF and granzyme B in CD45.1⁺ or CD45.1-negative (CD45.2⁺) CD4 and CD8 T cells in both Klrk1^+/+ and Klrk1^−/− recipients. Production of IFNγ by CNS-infiltrated CD4 and CD8 T cells was easily detected in both mouse groups. Notably, a significantly lower proportion of endogenous (CD45.2) CD4 T cells produced this cytokine compared to the counterparts originating from the transfer (^*p < 0.05) in Klrk1^+/+ mice (Figure 7C). In contrast, the proportion of endogenous (CD45.2) CD8 T cells producing IFNγ was significantly higher than within transferred CD8 T cells (^**p < 0.01; Figure 7D). A relatively small but similar proportion of endogenous and transferred CD4 T cells expressed granzyme B (Figure 7C). However, the percentage of CD8 T cells carrying this lytic enzyme was significantly greater in the endogenous cells than in the transferred one (p < 0.05, Figure 7D). We observed a significantly greater proportion of transferred (CD45.1) than endogenous (CD45.2) CD4 T cells producing GM-CSF (Figure 7C). Only few or no CD8 T cells produced GM-CSF or IL-17 above background level (data not shown). Notably, elevated proportions of endogenous CD4 and CD8 T cells in the CNS of Klrk1^−/− recipients produced the immune mediators tested (IFNγ, GM-CSF or granzyme B).

We also measured the expression of NKG2D and could observe that as expected none of the endogenous CD8 T cells carry this receptor in the Klrk1^−/− recipients. Nevertheless, a significantly higher percentage of endogenous (CD45.2) than transferred (CD45.1) CD8 T cells exhibited NKG2D in wild type recipients. It was not possible to compare transferred and endogenous CD8 T cells in Klrk1^−/− recipients due to the absence or very low number of CD45.1⁺ CD8 T cells in these mice (Figure 7D). A very low percentage of CD4 T cells expressed NKG2D in our model precluding any comparison. Our results suggest that both endogenous and transferred CD8 T cells could recognize NKG2D ligands in the CNS during passive EAE.

To illustrate the striking differences within CNS-infiltrated T cells, we generated pie charts showing the proportions of cells with an endogenous (CD45.2) or transferred (CD45.1) origin amongst CD4 (Figure 7C) or CD8 (Figure 7D) T lymphocytes. These representations underline that amongst all CNS-infiltrating CD4 T cells those producing IFNγ or GM-CSF originated mainly from the transferred cells (Figure 7C). The second and third column of pies also illustrate the greater proportion of transferred CD4 T cells (CD45.1) producing these cytokines. In contrast, in the CD8 T cell compartment, most cells expressing effector molecules (IFNγ, granzyme B or NKG2D) originated from the endogenous compartment (Figure 7D). In fact, as mentioned above, few transferred CD8 T cells (CD45.1⁺) reached the CNS although similar amount of these cells were co-transferred with CD4 T cells. Nevertheless, most CD45.1⁺CD8 T cells that reached the CNS were able to produce IFNγ and an important proportion expressed NKG2D (Figure 7D second column of pies). Given that a greater proportion of CD8 T cells express NKG2D than CD4 T cells, our results identify CD8 T cells as potential players within the endogenous NKG2D⁺ immune cells contributing to disease severity in the adoptive model of EAE.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.