Multiple sclerosis (MS) is a demyelinating disease of the central nervous system (CNS) characterized by plaques of infiltrating CD4+ and CD8+ T cells. Studies of MS and experimental autoimmune encephalomyelitis (EAE), an animal model of MS, focus on the contribution of CD4+ myelin-specific T cells. The role of CD8+ myelin-specific T cells in mediating EAE or MS has not been described previously. Here, we demonstrate that myelin-specific CD8+ T cells induce severe CNS autoimmunity in mice. The pathology and clinical symptoms in CD8+ T cell–mediated CNS autoimmunity demonstrate similarities to MS not seen in myelin-specific CD4+ T cell–mediated EAE. These data suggest that myelin-specific CD8+ T cells could function as effector cells in the pathogenesis of MS.
Multiple sclerosis (MS) is a demyelinating disease of the central nervous system (CNS) of unknown etiology that is characterized by a wide spectrum of neurological symptoms 1. Clinical problems in MS patients frequently result from upper motor neuron disease and include hyperreflexia, ataxia, spasticity, and visual defects 2. MS is believed to be an autoimmune disease because inflammatory infiltrates of the CNS contain T and B lymphocytes 3,4,5 and because genetic susceptibility has been linked to MHC class II genes 6. In addition, experimental autoimmune encephalomyelitis (EAE), an animal model of MS, is induced by generating T cell–mediated immunity to myelin antigens.
Oligoclonal CD4+ and CD8+ T cell populations have been detected within MS plaques, however, the function and antigenic specificity of these T cells is not known 7,8. Attention has focused on CD4+ T cells because susceptibility to MS is associated with MHC class II genes 9,10,11 and CD4+ T cells are critical to induction of EAE. In addition, a subset of TCR V gene sequences detected within MS plaques have been reported to express TCR sequence motifs similar to peripheral CD4+ T cells specific for myelin basic protein (MBP; reference 7). CD4+ T cells may not be the only cells that mediate MS pathology, however, as depletion of CD4+ T cells in MS patients made only minor improvements in relapse rates or levels of MRI activity 12,13.
We chose to investigate the role of CD8+ T cells in CNS autoimmunity for several reasons. First, CD8+ T cell clones specific for myelin antigens have been isolated from the periphery of MS patients and normal donors 14,15. Second, MHC class I genes are in linkage disequilibrium with the MHC class II genes associated with susceptibility to MS 16. Third, the ability of CD8+ myelin-specific T cells to function as effector cells in CNS autoimmunity has not been investigated. Earlier studies of EAE in CD8-deficient mice suggested that CD8+ T cells with undefined antigen specificity may function as suppressor/regulatory T cells in CNS autoimmune disease 17,18,19. CD8+ T cells derived from rodents fed MBP also suppress in vitro immune responses to MBP and protect animals from induction of EAE via a TGF-β–mediated pathway 19. Here we show that MBP-specific CD8+ T cells isolated from wild-type mice do not function as regulatory T cells but instead mediate severe CNS autoimmunity. The disease induced by CD8+ MBP-specific T cells exhibits some similarities to MS not seen in myelin-specific CD4+ T cell–mediated EAE.
Materials And Methods
C3HeB/FeJ, C3HeB/FeJ-MBPshi/+, C3HeB/FeJ-lpr, and C3HSmn.C-scid/J were purchased from The Jackson Laboratory. All mice were housed in a specific pathogen free animal facility at the University of Washington. All procedures have been approved by the animal care and use committee of the University of Washington.
HPLC-purified MBP79-87 peptide was purchased from Research Genetics.
Immunization and Isolation of MBP-specific T Cell Clones.
4–6-wk-old C3H and C3H.shi mice were immunized in the flank with 50 μg MBP79-87 peptide emulsified in Complete Freund's adjuvant. Draining lymph nodes from four immunized C3H and C3H.shi mice were harvested 10 d later and 3 × 107 draining lymph node cells were restimulated in vitro with 107 MBP 79–87-pulsed splenocytes and 10 U/ml IL-2 (Hoffman-La Roche) in 10 mls RPMI supplemented with 10% FCS. After 1 wk in culture, T cell lines were cloned by limiting dilution by culturing 104, 103, and 102 T cells with 106 syngenic spleen cells, 104 L cells infected with Vaccinia virus expressing MBP, and 50 U/ml IL-2 per well in 200 μl final volume. T cell clones from positive wells were then restimulated, expanded in 24-well plates, and tested in a 51Cr release assay 20.
T cell clones were stained with FITC-labeled anti-TCR-αβ, biotin-conjugated anti-CD8, and PE-labeled anti-CD4 mAbs (BD PharMingen). Biotin-conjugated anti-CD8 was visualized with streptavidin-conjugated Tricolor (Caltag). Data were collected on a FACScan™ (Becton Dickinson) and analyzed using CELLQuest™ software (Becton Dickinson).
T Cell Adoptive Transfers.
For disease induction experiments using intravenous transfer of MBP–CTL, recipient mice were sublethally irradiated (400 rads) on day −1. 2 × 107 T cells that had been stimulated for 2 d in vitro were transferred intravenously on day 0. 104 U IL-2 were administered intraperitoneally at the time of T cell transfer. For disease induction experiments using intrathecal transfer, 106 T cells were transferred (4 d after in vitro stimulation) with or without neutralizing reagents intrathecally into T4/5 lumbar section of the spinal cord of recipient mice in 20 μl PBS. An additional dose of neutralizing reagents was intrathecally injected on day 1. Cytokine blocking experiments used 40 ng of anti–IFN-γ (Clone H22, provided by Dr. R. Schreiber) or hamster IgG anti-TNP (BD PharMingen); 100 ng TNFR:Fc (Immunex) or human IgG1 (Sigma-Aldrich) per injection.
Clinical Scoring Scale for CTL-induced CNS Autoimmunity.
Clinical scoring scale for CTL-induced CNS autoimmunity was as follows. Grade 1, ataxia, loss of coordinated movement, and spastic reflexes seen in only one rear leg, with spinning and head-tilt in some mice; Grade 2, a loss of coordinated movement and spastic reflexes of both rear legs; Grade 3, paralysis of one leg (frequently sensation is retained); Grade 4, paralysis in two legs (frequently sensation is retained); and Grade 5, moribund. All Grade 5 mice were killed.
Histologic Analysis of Brain and Spinal Cords.
Brain and spinal cord of 15 recipient mice of MBP–CTL (affected) and eight recipient mice of VAC-CTL (control) were dissected, fixed in 10% Millonig's modified phosphate-buffered formalin for 48 h, sectioned, and embedded in paraffin. To ensure reproducibility, a rodent brain matrix (Electron Microscopy Sciences) was used for all brain sectioning. 4–6 serial sections of five brain cross-sections (20–30 total sections), two cross-sections of the spinal cord (T2, L2) and three longitudinal sections (C1-6, T3-13, L3-cauda equina) per mouse were stained with hematoxylin and eosin (H&E), Luxol Fast Blue (LFB), Crystal Violet, and anti-GFAP (Vector Laboratories). Sections from other major organs from these mice were stained with H&E.
MBP-specific CD8+ T Cells Circulate in the Periphery in Wild-Type Mice.
Previously we demonstrated that MHC class I–restricted, MBP-specific CD8+ T cells can be isolated from C3H.shi (MBP-deficient) mice primed with recombinant virus encoding MBP 20. The CD8+ CTLs are specific for a single epitope, MBP79-87, associated with the Kk MHC molecule. In this study MBP79-87–specific T cells were not readily detected in primary T cell lines derived from C3H wild-type mice infected with virus expressing MBP. This observation indicated that the MBP79-87–specific T cells normally undergo substantial tolerance induction in vivo, similar to many T cells specific for MHC class II–restricted myelin epitopes 21,22,23,24.
To determine whether some MBP-specific CTLs escape tolerance, we immunized wild-type C3H and MBP-deficient C3H.shi mice with MBP79-87 peptide and cloned T cells from the draining lymph nodes by limiting dilution. Six independent MBP-specific T cell clones were established from C3H wild-type mice and 10 T cell clones were established from C3H.shi mice using this protocol. All T cell clones derived from C3H and C3H.shi mice were of the CD8α/β+ TCR α/β+ lineage (Fig. 1 A and data not shown). All wild-type clones lysed H-2Kk expressing, MHC class II–deficient target cells infected with recombinant vaccinia virus encoding MBP as well as target cells coated with MBP79-87 peptide in a similar dose-dependent manner as T cell clones derived from C3H.shi mice (Fig. 1B and Fig. C). These data indicate that CD8+ MHC class I–restricted MBP-specific cytotoxic T cells circulate in the periphery of wild-type mice and exhibit the same sensitivity to antigen as T cells from MBP-deficient mice.
MBP-specific CD8+ T Cells Induce CNS Autoimmunity.
To investigate whether CD8+ T cells could induce CNS autoimmunity, activated MBP-specific CD8+ T cell clones isolated from C3H wild-type mice (MBP-CTLs) were transferred intravenously into recipient mice. Both C3H wild-type and C3H.scid recipient mice exhibited pronounced neurological disease. The clinical symptoms were primarily associated with upper motor neuron impairment including ataxia, spastic reflexes, loss of coordinated movement, spinning, and head-tilt, with some mice developing hind limb paralysis. Mice often lost motor function in their hind legs yet retained sensory feelings. Because the clinical disease course is distinct from classic EAE in which the primary symptom is ascending flaccid paralysis, we developed a new scale for scoring the severity of clinical signs (see Materials and Methods). The disease course was rapid and severe in both wild-type and C3H.scid mice, with significant weight loss and a high incidence of mortality observed (Fig. 2). Wild-type mice receiving vaccinia-specific CTLs or C3H.shi−/− mice receiving MBP–CTLs were clinically normal and experienced no weight loss.
The clinical signs described previously are very distinct from those seen in CD4+ T cell–mediated disease. Histologic analyses shown in (Fig. 3) also revealed striking differences between MBP-specific CTL-mediated autoimmune disease and EAE in both the distribution and type of lesions in the CNS. Lesions of clinically affected mice that received MBP–CTLs were confined to the brain and involved small blood vessels and the surrounding nervous tissue. These lesions were most frequent within the white matter of the cerebellum (Fig. 3 B), but widely scattered, focal involvement of gray (Fig. 3 J) and white matter of the brain stem, midbrain, and cerebral cortex was also observed. Focal inflammation of the meninges with involvement of adjacent nervous tissue occurred in some severely affected animals. Spinal cord lesions were not seen in this series of animals. The distribution of lesions in mice receiving MBP–CTLs at various times post disease onset is summarized in Table. The average number of lesions in each anatomical compartment per slide did not vary significantly between mice at different time points after disease onset with the cerebellum having an average of six independent lesions per slide, the brain stem with two, the midbrain with four, and the cerebral cortex with two lesions per slide.
The vascular lesions in MBP-specific CTL-mediated disease involved capillaries and venules and consisted of perivascular cuffing with rare disruption of the vascular wall. The perivascular cuffs were composed of mixtures of lymphocytes, macrophages, and a few neutrophils (Fig. 3 C). Inflammatory cells were focally marginated to the prominent endothelial surface of cuffed vessels. The changes seen in nervous tissue surrounding these affected vessels were consistent with focal cytotoxic or ischemic injury, although blood clots were not observed. These changes included the presence of degenerative, necrotic, and apoptotic cells (Fig. 3D, Fig. G, and Fig. H) and loss of staining intensity (Fig. 3B and Fig. F). Few inflammatory cells were present outside of the perivascular cuff. Vacuolation of the surrounding nervous tissue was common and consisted of demyelination and cytoplasmic swelling (Fig. 3F, Fig. G, and Fig. J). Astrogliosis was occasionally noted in tissues surrounding affected vessels (Fig. 3 L) although in a few instances perivascular injury manifests as a loss of all nucleated cells. Brain tissues from control mice were histologically normal (Fig. 3A, Fig. E, Fig. I, and Fig. K). Examination of lung, liver, kidney, and myocardium from affected and control mice revealed no significant lesions.
Effector Mechanism of MBP-specific CTL-mediated CNS Autoimmunity.
The major differences between the pathology described here and that seen in EAE are the predominance of lesions in the brain versus the spinal cord, the general lack of inflammation (except that directly associated with vascular walls) and the severe demyelination and perivascular cell death suggesting a cytotoxic or ischemic injury. It is likely that some of these differences are due to different effector mechanisms used by CD8+ versus CD4+ T cells. To begin to address effector mechanisms in MBP-specific CTL-mediated disease, we assessed the requirements for IFN-γ, TNF-α, and Fas–FasL interactions. Disease was induced in recipient mice by intrathecally coinjecting MBP–CTLs with and without neutralizing reagents specific for IFN-γ and TNF-α. Intrathecal transfer of MBP–CTLs alone induced a potent CNS autoimmune disease (Fig. 4 A). Clinical signs appear more rapidly after intrathecal versus intravenous transfer, presumably because the T cells do not have to traffic to the CNS from the periphery. Coinjecting a neutralizing Ab to IFN-γ with MBP–CTLs significantly reduced severity of disease compared with control mice (Fig. 4 B), indicating that IFN-γ plays an important role in CD8+ T cells specific for MBP (MBP–CTL)–mediated CNS autoimmune disease. In contrast to the results with IFN-γ, coinjection of TNFR:Fc fusion protein with MBP–CTLs had no effect on disease course compared with mice receiving isotype-matched control Abs (Fig. 4 C). MBP–CTLs also do not depend on Fas–FasL signaling as CNS autoimmunity was induced in C3H.lpr mice that was indistinguishable from wild-type C3H mice in time course and severity (Fig. 4 D).
The role of CD8+ T cells in MS is not known. The question is of considerable interest because significant numbers of CD8+ T cells of unknown function and antigen specificity are present in MS plaques 3,4,8. Our results demonstrate that MBP-specific CD8+ T cells can function as effector cells to induce CNS autoimmune disease in mice. Demonstrating a pathogenic role for CD8+ T cells in CNS autoimmunity contrasts with observations made in EAE that suggest that CD8+ T cells inhibit the induction of CNS autoimmunity or alter the clinical course of disease 17,18,19. Finding divergent roles for CD8+ T cells in CNS autoimmunity suggests that functionally distinct subsets of CD8+ T cells contribute to disease in different ways. TGF-β–expressing CD8+ T cells appear to function as regulatory cells 19 while our results indicate that IFN-γ producing MBP-specific CD8+ T cells induce severe CNS autoimmunity. This hypothesis has a direct corollary with the role of CD4+ T cells in autoimmune disease. CD25+CD4+ T cells of undefined antigen specificity perform a regulatory function in inhibiting autoimmunity 25,26 while CD4+ T cells specific for CNS antigens function as effector cells in inducing EAE.
The disease mediated by adoptive transfer of CD8+ MBP-specific T cells described here is characterized by very different clinical signs and CNS pathology than classic EAE mediated by transfer of CD4+ T cells. C3H mice have been considered fairly resistant to EAE induced by active immunization, however, adoptive transfer of CD4+ Th1 MBP-specific T cell lines into C3H mice results in classic EAE 27. These differences between CD4 T cell– and CTL-mediated EAE may be due to recognition of different cell types within the CNS by CD4+ and CD8+ T cells and/or the different effector mechanisms employed by CD4+ versus CD8+ T cells. The MBP–CTL-mediated disease described here is independent of CD4+ T cells as well as nonspecific CD8+ T cells, as shown by the induction of disease in C3H.scid recipients. This result indicates that MBP-specific CTLs can directly induce CNS autoimmunity by interacting with MHC class I molecules presenting the MBP epitope within the CNS and not by activating endogenous myelin-specific CD4+ T cells. Interestingly, there are some reports of “atypical” EAE in some EAE-resistant mouse strains immunized with certain PLP peptides that exhibits some similarities to the CTL-mediated disease described here 28,29. T cells mediating this atypical EAE were not characterized and it would be interesting to determine if they were CD4+ or CD8+ T cells. In human autoimmune disease, both CD4+ and CD8+ T cells may contribute to disease, such that the pathology reflects the activities of both types of effector cells.
CD8+ T cell–mediated CNS autoimmunity is largely inhibited by neutralizing IFN-γ activity. IFN-γ may function to increase MHC class I expression which is extremely limited in the CNS but is upregulated on resident cells by exposure to IFN-γ 30,31. IFN-γ may also promote demyelination through activation of macrophage and microglia or by a deleterious effect on oligodendrocytes as described for transgenic mice that express IFN-γ under the control of the MBP promoter 32. In contrast to the dependence on IFN-γ of CTL-mediated disease, IFN-γ is a protective cytokine in CD4+ T cell–mediated EAE 33. Interestingly, clinical trials of IFN-γ in MS resulted in significant exacerbation of symptoms in patients receiving IFN-γ as compared with controls 34 suggesting a stronger parallel for IFN-γ activity to CD8+ T cell-mediated CNS autoimmunity than CD4+ T cell–mediated EAE.
In addition to IFN-γ, TNF-α and Fas appear to play disparate roles in CD4+ versus CD8+ T cell–mediated CNS autoimmunity. Neutralizing TNF-α activity had no effect on CD8+ T cell–mediated CNS autoimmune disease course. This result differs from the amelioration of disease that occurs when TNF-α activity is neutralized in EAE 35, and may be more similar to MS in which an essential role for TNF-α has not been established 36,37. CD8+ T cell–mediated CNS autoimmunity is also independent of Fas–FasL interactions, while these interactions have been shown to influence EAE in some mouse models 38,39.
The observation that CD4+ and CD8+ T cells employ distinct effector mechanisms within the CNS may have important implications for understanding the pathogenesis of MS. MS exhibits heterogeneity in clinical course and CNS lesions and it has been difficult to formulate a unifying hypothesis that accounts for this heterogeneity. Recent histological studies have categorized the different types of MS lesions into separate patterns or groups. These efforts led to the suggestion that the heterogeneity in pathology and clinical outcome in MS patients reflects fundamentally different disease mechanisms 40. The MBP–CTL-mediated model that we describe here provides experimental evidence for this idea. The pathology observed in the MBP–CTL model is most similar to that of Pattern IV described by Lucchinetti et al. 41 for three patients with primary progressive MS. In these patients, inflammatory lesions centered on veins and venules and oligodendrocyte death is suggested that is reminiscent of virus or toxin-induced demyelination. Patients exhibiting pattern IV pathology may share a common pathogenic mechanism that includes CD8+ lymphocytes. Not all primary progressive MS patients exhibit this pattern, however, and most patients with primary progressive MS have numerous lesions in the spinal cord rather than the predominance of lesions in the brain seen in MBP-specific CTL-mediated disease. Thus, the CTL-mediated model, like CD4+ T cell–mediated EAE, is valuable in exploring the pathogenic potential and effector mechanisms of different T cell subsets but does not recapitulate all of the features of human disease.
One reason that accurate animal models of MS are difficult to generate is that the low precursor frequency of both CD4+ and CD8+ myelin-specific T cells in healthy animals requires artificial stimuli to induce disease and these stimuli tend to focus on activation of one particular T cell subset. Multiple factors are likely to be involved in triggering MS and variation in these factors may preferentially activate different types of myelin-specific T cells in different patients. Predisposing conditions for MS could include perturbations that impair the efficiency of tolerance induction for either MHC class I– or MHC class II–restricted T cells specific for myelin antigens. The type of perturbation may vary among patients such that the precursor frequency of one class of autoreactive T cells may increase more than another class. Therefore, identifying all of the cell types that can contribute to disease as well as the conditions under which these cells may be activated is essential to understand the pathogenesis of MS.
Our results now implicate CD8+ MBP-specific CTLs as well as CD4+ T cells as potential mediators of CNS autoimmune disease. Importantly, our data indicate that the disease mediated by MBP-specific CTLs is distinct from CD4+ T cell–mediated disease, supporting the idea that preferential activation of one T cell subset may emphasize different aspects of MS. The MBP–CTL-mediated model may be particularly useful in delineating events occurring during the initial stages of MS. Widespread inflammation in the CNS is a common feature of chronic MS, however, it has been suggested that demyelination may be triggered by an initial perivascular lymphocyte infiltration 40,42. Similar to the perivascular infiltrates in the myelin-specific CD8+ T cell model, CD8+ T cells present in MS lesions exhibit perivascular localization 3,4. Endothelial cells as well as resident phagocytic perivascular cells 43 may be able to present MBP epitopes to CTLs via a cross-presentation pathway 44 as both cell types constitutively express MHC class I molecules. Even though endothelial cells and phagocytic antigen presenting cells do not synthesize MBP, cell-associated antigens phagocytosed by antigen presenting cells are processed through the MHC class I pathway 44. Thus, perivascular injury mediated by myelin-specific CD8+ T cells may recapitulate events that occur during particular stages of MS that have not been accessible for study in other animal models.
The authors thank Drs. E.C. Alvord, D.R. Jeffrey, and G.H. Kraft for helpful discussions; Dr. M.J. Bevan, A. Perchellet, and A. Seamons for critical reading of the manuscript; and B. Sather for technical assistance.
This work was supported by a grant from the Research Royalty Fund at the University of Washington. E.S. Huseby was supported by a Predoctoral Fellowship from the National Institutes of Health (CA09537-13) and by the Samuel and Althea Stroum Endowed Diabetes Fellowship. J. Goverman was supported in part by a Harry Weaver Junior Faculty Award (2080-A-2) from the National Multiple Sclerosis Society.
E.S. Huseby's current address is Howard Hughes Medical Institute, Dept. of Immunology, National Jewish Medical Center, 1400 Jackson St., Denver, CO 80206.
Abbreviations used in this paper: CNS, central nervous system; EAE, experimental autoimmune encephalomyelitis; H&E, hematoxylin and eosin; LFB, Luxol Fast Blue; MBP, myelin basic protein; MBP–CTL, CD8+ T cells specific for MBP; MS, multiple sclerosis; Vac–CTL, Vaccinia-specific CTLs.